guile/libguile/ports.c

3323 lines
90 KiB
C
Raw Normal View History

/* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2006,
* 2007, 2008, 2009, 2010, 2011, 2012, 2013,
Deprecate C exports of GOOPS classes. * libguile/deprecated.h: (scm_class_boolean, scm_class_char, scm_class_pair) (scm_class_procedure, scm_class_string, scm_class_symbol) (scm_class_primitive_generic, scm_class_vector, scm_class_null) (scm_class_real, scm_class_complex, scm_class_integer) (scm_class_fraction, scm_class_unknown, scm_class_top) (scm_class_object, scm_class_class, scm_class_applicable) (scm_class_applicable_struct, scm_class_applicable_struct_with_setter) (scm_class_generic, scm_class_generic_with_setter, scm_class_accessor) (scm_class_extended_generic, scm_class_extended_generic_with_setter) (scm_class_extended_accessor, scm_class_method) (scm_class_accessor_method, scm_class_procedure_class) (scm_class_applicable_struct_class, scm_class_number, scm_class_list) (scm_class_keyword, scm_class_port, scm_class_input_output_port) (scm_class_input_port, scm_class_output_port, scm_class_foreign_slot) (scm_class_self, scm_class_protected, scm_class_hidden) (scm_class_opaque, scm_class_read_only, scm_class_protected_hidden) (scm_class_protected_opaque, scm_class_protected_read_only) (scm_class_scm, scm_class_int, scm_class_float) (scm_class_double, scm_port_class, scm_smob_class): Deprecate. * libguile/deprecated.c: * libguile/goops.c: * libguile/goops.h: Adapt to deprecation. * libguile/goops.h * libguile/goops.c (scm_is_generic, scm_is_method): New interfaces. (SCM_GENERICP, SCM_METHODP): Change to use new interfaces. * libguile/ports.c (scm_make_port_type): * libguile/smob.c (scm_make_smob_type, scm_set_smob_apply): Use internal names for the port and smob class arrays.
2015-01-06 14:16:03 -05:00
* 2014, 2015 Free Software Foundation, Inc.
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public License
* as published by the Free Software Foundation; either version 3 of
* the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
* 02110-1301 USA
*/
/* Headers. */
2006-04-17 00:05:42 +00:00
#define _LARGEFILE64_SOURCE /* ask for stat64 etc */
#ifdef HAVE_CONFIG_H
# include <config.h>
#endif
#include <stdio.h>
#include <errno.h>
2006-10-09 23:40:48 +00:00
#include <fcntl.h> /* for chsize on mingw */
#include <assert.h>
#include <iconv.h>
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
#include <uniconv.h>
#include <unistr.h>
#include <striconveh.h>
#include <assert.h>
#include "libguile/_scm.h"
#include "libguile/async.h"
#include "libguile/deprecation.h"
#include "libguile/eval.h"
2006-10-09 23:40:48 +00:00
#include "libguile/fports.h" /* direct access for seek and truncate */
2005-01-18 13:59:04 +00:00
#include "libguile/goops.h"
#include "libguile/smob.h"
#include "libguile/chars.h"
#include "libguile/dynwind.h"
#include "libguile/keywords.h"
#include "libguile/hashtab.h"
#include "libguile/root.h"
#include "libguile/strings.h"
#include "libguile/mallocs.h"
#include "libguile/validate.h"
#include "libguile/ports.h"
#include "libguile/ports-internal.h"
(scm_i_mode_bits_n): New, for counted strings. (scm_mode_bits): Use it. (scm_c_port_for_each): Blocking GC does not seem to work, allocate a vector normally and fill that instead of consing a list with a blocked GC. * strings.h, strings.c: (scm_i_string_chars, scm_i_string_length, scm_i_string_writable_chars, scm_i_string_stop_writing): New, to replace SCM_I_STRING_CHARS and SCM_I_STRING_LENGTH. Updated all uses. (scm_i_make_string, scm_c_make_string): New, to replace scm_allocate_string. Updated all uses. (SCM_STRINGP, SCM_STRING_CHARS, SCM_STRING_UCHARS, SCM_STRING_LENGTH): Deprecated. (scm_allocate_string, scm_take_str, scm_take0str, scm_mem2string, scm_str2string, scm_makfrom0str, scm_makfrom0str_opt): Discouraged. Replaced all uses with scm_from_locale_string or similar, as appropriate. (scm_c_string_length, scm_c_string_ref, scm_c_string_set_x, scm_c_substring, scm_c_substring_shared, scm_c_substring_copy, scm_substring_shared, scm_substring_copy): New. * symbols.c, symbols.h (SCM_SYMBOLP, SCM_SYMBOL_FUNC, SCM_SET_SYMBOL_FUNC, SCM_SYMBOL_PROPS, SCM_SET_SYMBOL_PROPS, SCM_SYMBOL_HASH, SCM_SYMBOL_INTERNED_P, scm_mem2symbol, scm_str2symbol, scm_mem2uninterned_symbol): Discouraged. (SCM_SYMBOL_LENGTH, SCM_SYMBOL_CHARS, scm_c_symbol2str): Deprecated. (SCM_MAKE_SYMBOL_TAG, SCM_SET_SYMBOL_LENGTH, SCM_SET_SYMBOL_CHARS, SCM_PROP_SLOTS, SCM_SET_PROP_SLOTS): Removed. (scm_is_symbol, scm_from_locale_symbol, scm_from_locale_symboln): New, to replace scm_str2symbol and scm_mem2symbol, respectively. Updated all uses. (scm_gensym): Generate only the number suffix in the buffer, just string-append the prefix.
2004-08-19 17:17:22 +00:00
#include "libguile/vectors.h"
#include "libguile/weak-set.h"
2005-03-02 20:42:01 +00:00
#include "libguile/fluids.h"
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
#include "libguile/eq.h"
#include "libguile/alist.h"
#ifdef HAVE_STRING_H
#include <string.h>
#endif
#ifdef HAVE_IO_H
#include <io.h>
#endif
#include <unistd.h>
#ifdef HAVE_SYS_IOCTL_H
#include <sys/ioctl.h>
#endif
2006-10-09 23:40:48 +00:00
/* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
already, but have this code here in case that wasn't so in past versions,
or perhaps to help other minimal DOS environments.
gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
might be possibilities if we've got other systems without ftruncate. */
#if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE
2001-06-26 17:53:09 +00:00
#define ftruncate(fd, size) chsize (fd, size)
2006-10-09 23:40:48 +00:00
#undef HAVE_FTRUNCATE
#define HAVE_FTRUNCATE 1
2001-06-26 17:53:09 +00:00
#endif
/* Port encodings are case-insensitive ASCII strings. */
static char
ascii_toupper (char c)
{
return (c < 'a' || c > 'z') ? c : ('A' + (c - 'a'));
}
/* It is only necessary to use this function on encodings that come from
the user and have not been canonicalized yet. Encodings that are set
on ports or in the default encoding fluid are in upper-case, and can
be compared with strcmp. */
static int
encoding_matches (const char *enc, const char *upper)
{
if (!enc)
enc = "ISO-8859-1";
while (*enc)
if (ascii_toupper (*enc++) != *upper++)
return 0;
return !*upper;
}
static char*
canonicalize_encoding (const char *enc)
{
char *ret;
int i;
if (!enc)
return "ISO-8859-1";
ret = scm_gc_strdup (enc, "port");
for (i = 0; ret[i]; i++)
{
if (ret[i] > 127)
/* Restrict to ASCII. */
scm_misc_error (NULL, "invalid character encoding ~s",
scm_list_1 (scm_from_latin1_string (enc)));
else
ret[i] = ascii_toupper (ret[i]);
}
return ret;
}
/* The port kind table --- a dynamically resized array of port types. */
/* scm_ptobs scm_numptob
* implement a dynamically resized array of ptob records.
* Indexes into this table are used when generating type
* tags for smobjects (if you know a tag you can get an index and conversely).
*/
static scm_t_ptob_descriptor **scm_ptobs = NULL;
static long scm_numptob = 0; /* Number of port types. */
static long scm_ptobs_size = 0; /* Number of slots in the port type
table. */
static scm_i_pthread_mutex_t scm_ptobs_lock = SCM_I_PTHREAD_MUTEX_INITIALIZER;
long
scm_c_num_port_types (void)
{
long ret;
scm_i_pthread_mutex_lock (&scm_ptobs_lock);
ret = scm_numptob;
scm_i_pthread_mutex_unlock (&scm_ptobs_lock);
return ret;
}
scm_t_ptob_descriptor*
scm_c_port_type_ref (long ptobnum)
{
scm_t_ptob_descriptor *ret = NULL;
scm_i_pthread_mutex_lock (&scm_ptobs_lock);
if (0 <= ptobnum && ptobnum < scm_numptob)
ret = scm_ptobs[ptobnum];
scm_i_pthread_mutex_unlock (&scm_ptobs_lock);
if (!ret)
scm_out_of_range ("scm_c_port_type_ref", scm_from_long (ptobnum));
return ret;
}
long
scm_c_port_type_add_x (scm_t_ptob_descriptor *desc)
{
long ret = -1;
scm_i_pthread_mutex_lock (&scm_ptobs_lock);
if (scm_numptob + 1 < SCM_I_MAX_PORT_TYPE_COUNT)
{
if (scm_numptob == scm_ptobs_size)
{
unsigned long old_size = scm_ptobs_size;
scm_t_ptob_descriptor **old_ptobs = scm_ptobs;
/* Currently there are only 9 predefined port types, so one
resize will cover it. */
scm_ptobs_size = old_size + 10;
if (scm_ptobs_size >= SCM_I_MAX_PORT_TYPE_COUNT)
scm_ptobs_size = SCM_I_MAX_PORT_TYPE_COUNT;
scm_ptobs = scm_gc_malloc (sizeof (*scm_ptobs) * scm_ptobs_size,
"scm_ptobs");
memcpy (scm_ptobs, old_ptobs, sizeof (*scm_ptobs) * scm_numptob);
}
ret = scm_numptob++;
scm_ptobs[ret] = desc;
}
scm_i_pthread_mutex_unlock (&scm_ptobs_lock);
if (ret < 0)
scm_out_of_range ("scm_c_port_type_add_x", scm_from_long (scm_numptob));
return ret;
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* Default buffer size. Used if the port type won't supply a value. */
static const size_t default_buffer_size = 1024;
2001-06-14 19:50:43 +00:00
scm_t_bits
scm_make_port_type (char *name,
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
size_t (*read) (SCM port, SCM dst, size_t start,
size_t count),
size_t (*write) (SCM port, SCM src, size_t start,
size_t count))
{
scm_t_ptob_descriptor *desc;
long ptobnum;
1999-08-11 Gary Houston <ghouston@easynet.co.uk> * fports.c (fport_input_waiting): if select is used, return 1 instead of whatever FD_ISSET expands to. maybe it will be useful to interpret the value from the input_waiting ptob procedure as a lower bound on the number of bytes available. * Mikael asked for a few names to be changed... * ports.c (scm_make_port_type): take the write procedure as the second argument instead of the flush procedure. * ports.h (scm_ptob_descriptor): rename the ptob procedures: fflush -> flush, read_flush -> end_input, fclose -> close, fill_buffer -> fill_input, ftruncate -> truncate, input_waiting_p -> input_waiting. * ports.c (end_input_void_port): was read_flush_void_port. (scm_set_port_end_input): was scm_set_port_flush_input. (scm_set_port_flush): was scm_set_port_write. (scm_set_port_input_waiting): was scm_set_port_input_waiting_p (scm_end_input): was scm_read_flush. (scm_fill_input): was scm_fill_buffer. (scm_flush): was scm_fflush. * fports.c (fport_input_waiting): renamed from fport_input_waiting_p. (fport_end_input): was local_read_flush. (fport_flush): was local_fflush. (fport_close): was local_fclose. (fport_truncate): was local_ftruncate. (fport_seek): was local_seek. (fport_free): was local_free. (fport_fill_input): was fport_fill_buffer. * strports.c (st_end_input): was st_read_flush. (st_truncate): was st_ftruncate. * vports.c: (sf_flush): was sfflush. (sf_close): was sfclose. (sf_fill_input): was sf_fill_buffer. * ports.c, fports.c, strports, vports.c, ioext.c, unif.c, filesys.c: change callers.
1999-08-11 18:35:42 +00:00
desc = scm_gc_malloc_pointerless (sizeof (*desc), "port-type");
memset (desc, 0, sizeof (*desc));
1999-08-11 Gary Houston <ghouston@easynet.co.uk> * fports.c (fport_input_waiting): if select is used, return 1 instead of whatever FD_ISSET expands to. maybe it will be useful to interpret the value from the input_waiting ptob procedure as a lower bound on the number of bytes available. * Mikael asked for a few names to be changed... * ports.c (scm_make_port_type): take the write procedure as the second argument instead of the flush procedure. * ports.h (scm_ptob_descriptor): rename the ptob procedures: fflush -> flush, read_flush -> end_input, fclose -> close, fill_buffer -> fill_input, ftruncate -> truncate, input_waiting_p -> input_waiting. * ports.c (end_input_void_port): was read_flush_void_port. (scm_set_port_end_input): was scm_set_port_flush_input. (scm_set_port_flush): was scm_set_port_write. (scm_set_port_input_waiting): was scm_set_port_input_waiting_p (scm_end_input): was scm_read_flush. (scm_fill_input): was scm_fill_buffer. (scm_flush): was scm_fflush. * fports.c (fport_input_waiting): renamed from fport_input_waiting_p. (fport_end_input): was local_read_flush. (fport_flush): was local_fflush. (fport_close): was local_fclose. (fport_truncate): was local_ftruncate. (fport_seek): was local_seek. (fport_free): was local_free. (fport_fill_input): was fport_fill_buffer. * strports.c (st_end_input): was st_read_flush. (st_truncate): was st_ftruncate. * vports.c: (sf_flush): was sfflush. (sf_close): was sfclose. (sf_fill_input): was sf_fill_buffer. * ports.c, fports.c, strports, vports.c, ioext.c, unif.c, filesys.c: change callers.
1999-08-11 18:35:42 +00:00
desc->name = name;
desc->print = scm_port_print;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
desc->read = read;
desc->write = write;
1999-08-11 Gary Houston <ghouston@easynet.co.uk> * fports.c (fport_input_waiting): if select is used, return 1 instead of whatever FD_ISSET expands to. maybe it will be useful to interpret the value from the input_waiting ptob procedure as a lower bound on the number of bytes available. * Mikael asked for a few names to be changed... * ports.c (scm_make_port_type): take the write procedure as the second argument instead of the flush procedure. * ports.h (scm_ptob_descriptor): rename the ptob procedures: fflush -> flush, read_flush -> end_input, fclose -> close, fill_buffer -> fill_input, ftruncate -> truncate, input_waiting_p -> input_waiting. * ports.c (end_input_void_port): was read_flush_void_port. (scm_set_port_end_input): was scm_set_port_flush_input. (scm_set_port_flush): was scm_set_port_write. (scm_set_port_input_waiting): was scm_set_port_input_waiting_p (scm_end_input): was scm_read_flush. (scm_fill_input): was scm_fill_buffer. (scm_flush): was scm_fflush. * fports.c (fport_input_waiting): renamed from fport_input_waiting_p. (fport_end_input): was local_read_flush. (fport_flush): was local_fflush. (fport_close): was local_fclose. (fport_truncate): was local_ftruncate. (fport_seek): was local_seek. (fport_free): was local_free. (fport_fill_input): was fport_fill_buffer. * strports.c (st_end_input): was st_read_flush. (st_truncate): was st_ftruncate. * vports.c: (sf_flush): was sfflush. (sf_close): was sfclose. (sf_fill_input): was sf_fill_buffer. * ports.c, fports.c, strports, vports.c, ioext.c, unif.c, filesys.c: change callers.
1999-08-11 18:35:42 +00:00
ptobnum = scm_c_port_type_add_x (desc);
1999-08-11 Gary Houston <ghouston@easynet.co.uk> * fports.c (fport_input_waiting): if select is used, return 1 instead of whatever FD_ISSET expands to. maybe it will be useful to interpret the value from the input_waiting ptob procedure as a lower bound on the number of bytes available. * Mikael asked for a few names to be changed... * ports.c (scm_make_port_type): take the write procedure as the second argument instead of the flush procedure. * ports.h (scm_ptob_descriptor): rename the ptob procedures: fflush -> flush, read_flush -> end_input, fclose -> close, fill_buffer -> fill_input, ftruncate -> truncate, input_waiting_p -> input_waiting. * ports.c (end_input_void_port): was read_flush_void_port. (scm_set_port_end_input): was scm_set_port_flush_input. (scm_set_port_flush): was scm_set_port_write. (scm_set_port_input_waiting): was scm_set_port_input_waiting_p (scm_end_input): was scm_read_flush. (scm_fill_input): was scm_fill_buffer. (scm_flush): was scm_fflush. * fports.c (fport_input_waiting): renamed from fport_input_waiting_p. (fport_end_input): was local_read_flush. (fport_flush): was local_fflush. (fport_close): was local_fclose. (fport_truncate): was local_ftruncate. (fport_seek): was local_seek. (fport_free): was local_free. (fport_fill_input): was fport_fill_buffer. * strports.c (st_end_input): was st_read_flush. (st_truncate): was st_ftruncate. * vports.c: (sf_flush): was sfflush. (sf_close): was sfclose. (sf_fill_input): was sf_fill_buffer. * ports.c, fports.c, strports, vports.c, ioext.c, unif.c, filesys.c: change callers.
1999-08-11 18:35:42 +00:00
/* Make a class object if GOOPS is present. */
Deprecate C exports of GOOPS classes. * libguile/deprecated.h: (scm_class_boolean, scm_class_char, scm_class_pair) (scm_class_procedure, scm_class_string, scm_class_symbol) (scm_class_primitive_generic, scm_class_vector, scm_class_null) (scm_class_real, scm_class_complex, scm_class_integer) (scm_class_fraction, scm_class_unknown, scm_class_top) (scm_class_object, scm_class_class, scm_class_applicable) (scm_class_applicable_struct, scm_class_applicable_struct_with_setter) (scm_class_generic, scm_class_generic_with_setter, scm_class_accessor) (scm_class_extended_generic, scm_class_extended_generic_with_setter) (scm_class_extended_accessor, scm_class_method) (scm_class_accessor_method, scm_class_procedure_class) (scm_class_applicable_struct_class, scm_class_number, scm_class_list) (scm_class_keyword, scm_class_port, scm_class_input_output_port) (scm_class_input_port, scm_class_output_port, scm_class_foreign_slot) (scm_class_self, scm_class_protected, scm_class_hidden) (scm_class_opaque, scm_class_read_only, scm_class_protected_hidden) (scm_class_protected_opaque, scm_class_protected_read_only) (scm_class_scm, scm_class_int, scm_class_float) (scm_class_double, scm_port_class, scm_smob_class): Deprecate. * libguile/deprecated.c: * libguile/goops.c: * libguile/goops.h: Adapt to deprecation. * libguile/goops.h * libguile/goops.c (scm_is_generic, scm_is_method): New interfaces. (SCM_GENERICP, SCM_METHODP): Change to use new interfaces. * libguile/ports.c (scm_make_port_type): * libguile/smob.c (scm_make_smob_type, scm_set_smob_apply): Use internal names for the port and smob class arrays.
2015-01-06 14:16:03 -05:00
if (SCM_UNPACK (scm_i_port_class[0]) != 0)
scm_make_port_classes (ptobnum, name);
return scm_tc7_port + ptobnum * 256;
}
void
2006-06-17 23:15:59 +00:00
scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port,
scm_print_state *pstate))
{
scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->print = print;
}
void
scm_set_port_close (scm_t_bits tc, void (*close) (SCM))
{
scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->close = close;
}
void
scm_set_port_needs_close_on_gc (scm_t_bits tc, int needs_close_p)
{
scm_t_ptob_descriptor *ptob = scm_c_port_type_ref (SCM_TC2PTOBNUM (tc));
if (needs_close_p)
ptob->flags |= SCM_PORT_TYPE_NEEDS_CLOSE_ON_GC;
else
ptob->flags &= ~SCM_PORT_TYPE_NEEDS_CLOSE_ON_GC;
}
void
scm_set_port_seek (scm_t_bits tc, scm_t_off (*seek) (SCM, scm_t_off, int))
{
scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->seek = seek;
}
void
scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off))
{
scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->truncate = truncate;
}
void
2006-06-17 23:15:59 +00:00
scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM))
{
scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->input_waiting = input_waiting;
}
void
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_set_port_get_natural_buffer_sizes
(scm_t_bits tc, void (*get_natural_buffer_sizes) (SCM, size_t *, size_t *))
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_ptob_descriptor *ptob = scm_c_port_type_ref (SCM_TC2PTOBNUM (tc));
ptob->get_natural_buffer_sizes = get_natural_buffer_sizes;
}
static void
scm_i_set_pending_eof (SCM port)
{
SCM_PTAB_ENTRY (port)->read_buf->has_eof_p = SCM_BOOL_T;
}
static void
scm_i_clear_pending_eof (SCM port)
{
SCM_PTAB_ENTRY (port)->read_buf->has_eof_p = SCM_BOOL_F;
}
SCM_DEFINE (scm_i_port_property, "%port-property", 2, 0, 0,
(SCM port, SCM key),
"Return the property of @var{port} associated with @var{key}.")
#define FUNC_NAME s_scm_i_port_property
{
scm_i_pthread_mutex_t *lock;
SCM result;
SCM_VALIDATE_OPPORT (1, port);
scm_c_lock_port (port, &lock);
result = scm_assq_ref (SCM_PORT_GET_INTERNAL (port)->alist, key);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return result;
}
#undef FUNC_NAME
SCM_DEFINE (scm_i_set_port_property_x, "%set-port-property!", 3, 0, 0,
(SCM port, SCM key, SCM value),
"Set the property of @var{port} associated with @var{key} to @var{value}.")
#define FUNC_NAME s_scm_i_set_port_property_x
{
scm_i_pthread_mutex_t *lock;
scm_t_port_internal *pti;
SCM_VALIDATE_OPPORT (1, port);
scm_c_lock_port (port, &lock);
pti = SCM_PORT_GET_INTERNAL (port);
pti->alist = scm_assq_set_x (pti->alist, key, value);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
/* Standard ports --- current input, output, error, and more(!). */
static SCM cur_inport_fluid = SCM_BOOL_F;
static SCM cur_outport_fluid = SCM_BOOL_F;
static SCM cur_errport_fluid = SCM_BOOL_F;
static SCM cur_warnport_fluid = SCM_BOOL_F;
static SCM cur_loadport_fluid = SCM_BOOL_F;
2005-03-02 20:42:01 +00:00
SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0,
(void),
"Return the current input port. This is the default port used\n"
"by many input procedures. Initially, @code{current-input-port}\n"
"returns the @dfn{standard input} in Unix and C terminology.")
#define FUNC_NAME s_scm_current_input_port
{
if (scm_is_true (cur_inport_fluid))
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
return scm_fluid_ref (cur_inport_fluid);
else
return SCM_BOOL_F;
}
#undef FUNC_NAME
SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0,
(void),
"Return the current output port. This is the default port used\n"
2001-11-11 15:01:52 +00:00
"by many output procedures. Initially,\n"
"@code{current-output-port} returns the @dfn{standard output} in\n"
"Unix and C terminology.")
#define FUNC_NAME s_scm_current_output_port
{
if (scm_is_true (cur_outport_fluid))
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
return scm_fluid_ref (cur_outport_fluid);
else
return SCM_BOOL_F;
}
#undef FUNC_NAME
SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0,
(void),
"Return the port to which errors and warnings should be sent (the\n"
"@dfn{standard error} in Unix and C terminology).")
#define FUNC_NAME s_scm_current_error_port
{
if (scm_is_true (cur_errport_fluid))
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
return scm_fluid_ref (cur_errport_fluid);
else
return SCM_BOOL_F;
}
#undef FUNC_NAME
SCM_DEFINE (scm_current_warning_port, "current-warning-port", 0, 0, 0,
(void),
"Return the port to which diagnostic warnings should be sent.")
#define FUNC_NAME s_scm_current_warning_port
{
if (scm_is_true (cur_warnport_fluid))
return scm_fluid_ref (cur_warnport_fluid);
else
return SCM_BOOL_F;
}
#undef FUNC_NAME
SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0,
(),
"Return the current-load-port.\n"
"The load port is used internally by @code{primitive-load}.")
#define FUNC_NAME s_scm_current_load_port
{
2005-03-02 20:42:01 +00:00
return scm_fluid_ref (cur_loadport_fluid);
}
#undef FUNC_NAME
SCM
scm_set_current_input_port (SCM port)
#define FUNC_NAME "set-current-input-port"
{
2005-03-02 20:42:01 +00:00
SCM oinp = scm_fluid_ref (cur_inport_fluid);
SCM_VALIDATE_OPINPORT (1, port);
2005-03-02 20:42:01 +00:00
scm_fluid_set_x (cur_inport_fluid, port);
return oinp;
}
#undef FUNC_NAME
SCM
scm_set_current_output_port (SCM port)
#define FUNC_NAME "scm-set-current-output-port"
{
2005-03-02 20:42:01 +00:00
SCM ooutp = scm_fluid_ref (cur_outport_fluid);
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
2005-03-02 20:42:01 +00:00
scm_fluid_set_x (cur_outport_fluid, port);
return ooutp;
}
#undef FUNC_NAME
SCM
scm_set_current_error_port (SCM port)
#define FUNC_NAME "set-current-error-port"
{
2005-03-02 20:42:01 +00:00
SCM oerrp = scm_fluid_ref (cur_errport_fluid);
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
2005-03-02 20:42:01 +00:00
scm_fluid_set_x (cur_errport_fluid, port);
return oerrp;
}
#undef FUNC_NAME
SCM
scm_set_current_warning_port (SCM port)
#define FUNC_NAME "set-current-warning-port"
{
SCM owarnp = scm_fluid_ref (cur_warnport_fluid);
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
scm_fluid_set_x (cur_warnport_fluid, port);
return owarnp;
}
#undef FUNC_NAME
void
scm_dynwind_current_input_port (SCM port)
2005-03-02 20:42:01 +00:00
#define FUNC_NAME NULL
{
2005-03-02 20:42:01 +00:00
SCM_VALIDATE_OPINPORT (1, port);
scm_dynwind_fluid (cur_inport_fluid, port);
}
2005-03-02 20:42:01 +00:00
#undef FUNC_NAME
void
scm_dynwind_current_output_port (SCM port)
2005-03-02 20:42:01 +00:00
#define FUNC_NAME NULL
{
2005-03-02 20:42:01 +00:00
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
scm_dynwind_fluid (cur_outport_fluid, port);
}
2005-03-02 20:42:01 +00:00
#undef FUNC_NAME
void
scm_dynwind_current_error_port (SCM port)
2005-03-02 20:42:01 +00:00
#define FUNC_NAME NULL
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
scm_dynwind_fluid (cur_errport_fluid, port);
2005-03-02 20:42:01 +00:00
}
#undef FUNC_NAME
void
scm_i_dynwind_current_load_port (SCM port)
{
scm_dynwind_fluid (cur_loadport_fluid, port);
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* Port buffers. */
scm_t_port_buffer *
scm_c_make_port_buffer (size_t size)
{
scm_t_port_buffer *ret = scm_gc_typed_calloc (scm_t_port_buffer);
ret->size = size;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
ret->bytevector = scm_c_make_bytevector (size);
ret->buf = (scm_t_uint8 *) SCM_BYTEVECTOR_CONTENTS (ret->bytevector);
ret->has_eof_p = SCM_BOOL_F;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
return ret;
}
/* Retrieving a port's mode. */
/* Return the flags that characterize a port based on the mode
* string used to open a file for that port.
*
* See PORT FLAGS in scm.h
*/
static long
scm_i_mode_bits_n (SCM modes)
{
return (SCM_OPN
| (scm_i_string_contains_char (modes, 'r')
|| scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0)
| (scm_i_string_contains_char (modes, 'w')
|| scm_i_string_contains_char (modes, 'a')
|| scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0)
| (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0)
| (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0));
}
long
scm_mode_bits (char *modes)
{
/* Valid characters are rw+a0l. So, use latin1. */
return scm_i_mode_bits (scm_from_latin1_string (modes));
}
long
scm_i_mode_bits (SCM modes)
{
long bits;
if (!scm_is_string (modes))
scm_wrong_type_arg_msg (NULL, 0, modes, "string");
bits = scm_i_mode_bits_n (modes);
scm_remember_upto_here_1 (modes);
return bits;
}
/* Return the mode flags from an open port.
* Some modes such as "append" are only used when opening
* a file and are not returned here. */
SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0,
(SCM port),
"Return the port modes associated with the open port @var{port}.\n"
"These will not necessarily be identical to the modes used when\n"
"the port was opened, since modes such as \"append\" which are\n"
"used only during port creation are not retained.")
#define FUNC_NAME s_scm_port_mode
{
char modes[4];
modes[0] = '\0';
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPPORT (1, port);
if (SCM_CELL_WORD_0 (port) & SCM_RDNG) {
if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
strcpy (modes, "r+");
else
strcpy (modes, "r");
}
else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
strcpy (modes, "w");
if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
strcat (modes, "0");
return scm_from_latin1_string (modes);
}
#undef FUNC_NAME
/* The port table --- a weak set of all ports.
We need a global registry of ports to flush them all at exit, and to
get all the ports matching a file descriptor. */
SCM scm_i_port_weak_set;
/* Port finalization. */
static SCM
do_close (void *data)
{
return scm_close_port (SCM_PACK_POINTER (data));
}
/* Finalize the object (a port) pointed to by PTR. */
static void
finalize_port (void *ptr, void *data)
{
SCM port = SCM_PACK_POINTER (ptr);
if (!SCM_PORTP (port))
abort ();
if (SCM_OPENP (port))
{
scm_internal_catch (SCM_BOOL_T, do_close, ptr,
scm_handle_by_message_noexit, NULL);
scm_gc_ports_collected++;
}
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
static void
initialize_port_buffers (SCM port)
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port);
size_t read_buf_size, write_buf_size;
if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
read_buf_size = write_buf_size = 1;
else
{
read_buf_size = write_buf_size = default_buffer_size;
if (ptob->get_natural_buffer_sizes)
ptob->get_natural_buffer_sizes (port, &read_buf_size, &write_buf_size);
if (read_buf_size == 0)
read_buf_size = 1;
if (write_buf_size == 0)
write_buf_size = 1;
}
if (!SCM_INPUT_PORT_P (port))
read_buf_size = 1;
if (!SCM_OUTPUT_PORT_P (port))
write_buf_size = 1;
pt->read_buffering = read_buf_size;
pt->read_buf = scm_c_make_port_buffer (read_buf_size);
pt->write_buf = scm_c_make_port_buffer (write_buf_size);
}
SCM
scm_c_make_port_with_encoding (scm_t_bits tag, unsigned long mode_bits,
const char *encoding,
scm_t_string_failed_conversion_handler handler,
scm_t_bits stream)
{
SCM ret;
scm_t_port *entry;
scm_t_port_internal *pti;
scm_t_ptob_descriptor *ptob;
entry = scm_gc_typed_calloc (scm_t_port);
pti = scm_gc_typed_calloc (scm_t_port_internal);
ptob = scm_c_port_type_ref (SCM_TC2PTOBNUM (tag));
ret = scm_words (tag | mode_bits, 3);
SCM_SET_CELL_WORD_1 (ret, (scm_t_bits) entry);
SCM_SET_CELL_WORD_2 (ret, (scm_t_bits) ptob);
2002-08-04 16:10:19 +00:00
entry->lock = scm_gc_malloc_pointerless (sizeof (*entry->lock), "port lock");
scm_i_pthread_mutex_init (entry->lock, scm_i_pthread_mutexattr_recursive);
entry->internal = pti;
entry->file_name = SCM_BOOL_F;
/* By default, any port type with a seek function has random-access
ports. */
entry->rw_random = ptob->seek != NULL;
entry->rw_active = SCM_PORT_NEITHER;
entry->port = ret;
entry->stream = stream;
if (encoding_matches (encoding, "UTF-8"))
{
pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8;
entry->encoding = "UTF-8";
}
else if (encoding_matches (encoding, "ISO-8859-1"))
{
pti->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1;
entry->encoding = "ISO-8859-1";
}
else
{
pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV;
entry->encoding = canonicalize_encoding (encoding);
}
entry->ilseq_handler = handler;
pti->iconv_descriptors = NULL;
pti->at_stream_start_for_bom_read = 1;
pti->at_stream_start_for_bom_write = 1;
pti->alist = SCM_EOL;
if (SCM_PORT_DESCRIPTOR (ret)->flags & SCM_PORT_TYPE_NEEDS_CLOSE_ON_GC)
{
scm_i_set_finalizer (SCM2PTR (ret), finalize_port, NULL);
scm_weak_set_add_x (scm_i_port_weak_set, ret);
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
initialize_port_buffers (ret);
return ret;
}
SCM
scm_c_make_port (scm_t_bits tag, unsigned long mode_bits, scm_t_bits stream)
{
return scm_c_make_port_with_encoding (tag, mode_bits,
scm_i_default_port_encoding (),
scm_i_default_port_conversion_handler (),
stream);
}
SCM
scm_new_port_table_entry (scm_t_bits tag)
{
return scm_c_make_port (tag, 0, 0);
}
/* Predicates. */
SCM_DEFINE (scm_port_p, "port?", 1, 0, 0,
(SCM x),
"Return a boolean indicating whether @var{x} is a port.\n"
"Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
"@var{x}))}.")
#define FUNC_NAME s_scm_port_p
{
return scm_from_bool (SCM_PORTP (x));
}
#undef FUNC_NAME
SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0,
(SCM x),
"Return @code{#t} if @var{x} is an input port, otherwise return\n"
"@code{#f}. Any object satisfying this predicate also satisfies\n"
"@code{port?}.")
#define FUNC_NAME s_scm_input_port_p
{
return scm_from_bool (SCM_INPUT_PORT_P (x));
}
#undef FUNC_NAME
SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0,
(SCM x),
"Return @code{#t} if @var{x} is an output port, otherwise return\n"
"@code{#f}. Any object satisfying this predicate also satisfies\n"
"@code{port?}.")
#define FUNC_NAME s_scm_output_port_p
{
x = SCM_COERCE_OUTPORT (x);
return scm_from_bool (SCM_OUTPUT_PORT_P (x));
}
#undef FUNC_NAME
SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0,
(SCM port),
"Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
"open.")
#define FUNC_NAME s_scm_port_closed_p
{
SCM_VALIDATE_PORT (1, port);
return scm_from_bool (!SCM_OPPORTP (port));
}
#undef FUNC_NAME
SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0,
(SCM x),
"Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
"return @code{#f}.")
#define FUNC_NAME s_scm_eof_object_p
{
return scm_from_bool (SCM_EOF_OBJECT_P (x));
}
#undef FUNC_NAME
/* Closing ports. */
static void close_iconv_descriptors (scm_t_iconv_descriptors *id);
/* scm_close_port
* Call the close operation on a port object.
* see also scm_close.
*/
SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0,
(SCM port),
"Close the specified port object. Return @code{#t} if it\n"
"successfully closes a port or @code{#f} if it was already\n"
"closed. An exception may be raised if an error occurs, for\n"
"example when flushing buffered output. See also @ref{Ports and\n"
"File Descriptors, close}, for a procedure which can close file\n"
"descriptors.")
#define FUNC_NAME s_scm_close_port
{
scm_t_port_internal *pti;
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_PORT (1, port);
if (SCM_CLOSEDP (port))
return SCM_BOOL_F;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* May throw an exception. */
if (SCM_OUTPUT_PORT_P (port))
scm_flush (port);
pti = SCM_PORT_GET_INTERNAL (port);
SCM_CLR_PORT_OPEN_FLAG (port);
if (SCM_PORT_DESCRIPTOR (port)->flags & SCM_PORT_TYPE_NEEDS_CLOSE_ON_GC)
scm_weak_set_remove_x (scm_i_port_weak_set, port);
if (SCM_PORT_DESCRIPTOR (port)->close)
/* Note! This may throw an exception. Anything after this point
should be resilient to non-local exits. */
SCM_PORT_DESCRIPTOR (port)->close (port);
if (pti->iconv_descriptors)
{
/* If we don't get here, the iconv_descriptors finalizer will
clean up. */
close_iconv_descriptors (pti->iconv_descriptors);
pti->iconv_descriptors = NULL;
}
return SCM_BOOL_T;
}
#undef FUNC_NAME
SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0,
(SCM port),
"Close the specified input port object. The routine has no effect if\n"
"the file has already been closed. An exception may be raised if an\n"
"error occurs. The value returned is unspecified.\n\n"
"See also @ref{Ports and File Descriptors, close}, for a procedure\n"
"which can close file descriptors.")
#define FUNC_NAME s_scm_close_input_port
{
SCM_VALIDATE_INPUT_PORT (1, port);
scm_close_port (port);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0,
(SCM port),
"Close the specified output port object. The routine has no effect if\n"
"the file has already been closed. An exception may be raised if an\n"
"error occurs. The value returned is unspecified.\n\n"
"See also @ref{Ports and File Descriptors, close}, for a procedure\n"
"which can close file descriptors.")
#define FUNC_NAME s_scm_close_output_port
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OUTPUT_PORT (1, port);
scm_close_port (port);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
/* Encoding characters to byte streams, and decoding byte streams to
characters. */
/* A fluid specifying the default encoding for newly created ports. If it is
a string, that is the encoding. If it is #f, it is in the "native"
(Latin-1) encoding. */
static SCM default_port_encoding_var;
/* Use ENCODING as the default encoding for future ports. */
void
scm_i_set_default_port_encoding (const char *encoding)
{
if (encoding_matches (encoding, "ISO-8859-1"))
scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F);
else
scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var),
scm_from_latin1_string (canonicalize_encoding (encoding)));
}
(scm_i_mode_bits_n): New, for counted strings. (scm_mode_bits): Use it. (scm_c_port_for_each): Blocking GC does not seem to work, allocate a vector normally and fill that instead of consing a list with a blocked GC. * strings.h, strings.c: (scm_i_string_chars, scm_i_string_length, scm_i_string_writable_chars, scm_i_string_stop_writing): New, to replace SCM_I_STRING_CHARS and SCM_I_STRING_LENGTH. Updated all uses. (scm_i_make_string, scm_c_make_string): New, to replace scm_allocate_string. Updated all uses. (SCM_STRINGP, SCM_STRING_CHARS, SCM_STRING_UCHARS, SCM_STRING_LENGTH): Deprecated. (scm_allocate_string, scm_take_str, scm_take0str, scm_mem2string, scm_str2string, scm_makfrom0str, scm_makfrom0str_opt): Discouraged. Replaced all uses with scm_from_locale_string or similar, as appropriate. (scm_c_string_length, scm_c_string_ref, scm_c_string_set_x, scm_c_substring, scm_c_substring_shared, scm_c_substring_copy, scm_substring_shared, scm_substring_copy): New. * symbols.c, symbols.h (SCM_SYMBOLP, SCM_SYMBOL_FUNC, SCM_SET_SYMBOL_FUNC, SCM_SYMBOL_PROPS, SCM_SET_SYMBOL_PROPS, SCM_SYMBOL_HASH, SCM_SYMBOL_INTERNED_P, scm_mem2symbol, scm_str2symbol, scm_mem2uninterned_symbol): Discouraged. (SCM_SYMBOL_LENGTH, SCM_SYMBOL_CHARS, scm_c_symbol2str): Deprecated. (SCM_MAKE_SYMBOL_TAG, SCM_SET_SYMBOL_LENGTH, SCM_SET_SYMBOL_CHARS, SCM_PROP_SLOTS, SCM_SET_PROP_SLOTS): Removed. (scm_is_symbol, scm_from_locale_symbol, scm_from_locale_symboln): New, to replace scm_str2symbol and scm_mem2symbol, respectively. Updated all uses. (scm_gensym): Generate only the number suffix in the buffer, just string-append the prefix.
2004-08-19 17:17:22 +00:00
/* Return the name of the default encoding for newly created ports. */
const char *
scm_i_default_port_encoding (void)
{
SCM encoding;
encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var));
if (!scm_is_string (encoding))
return "ISO-8859-1";
else
return scm_i_string_chars (encoding);
}
/* A fluid specifying the default conversion handler for newly created
ports. Its value should be one of the symbols below. */
static SCM default_conversion_strategy_var;
/* The possible conversion strategies. */
static SCM sym_error;
static SCM sym_substitute;
static SCM sym_escape;
/* Return the default failed encoding conversion policy for new created
ports. */
scm_t_string_failed_conversion_handler
scm_i_default_port_conversion_handler (void)
{
SCM value;
value = scm_fluid_ref (SCM_VARIABLE_REF (default_conversion_strategy_var));
if (scm_is_eq (sym_substitute, value))
return SCM_FAILED_CONVERSION_QUESTION_MARK;
else if (scm_is_eq (sym_escape, value))
return SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE;
else
/* Default to 'error also when the fluid's value is not one of
the valid symbols. */
return SCM_FAILED_CONVERSION_ERROR;
}
/* Use HANDLER as the default conversion strategy for future ports. */
void
scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler
handler)
{
SCM strategy;
switch (handler)
{
case SCM_FAILED_CONVERSION_ERROR:
strategy = sym_error;
break;
case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE:
strategy = sym_escape;
break;
case SCM_FAILED_CONVERSION_QUESTION_MARK:
strategy = sym_substitute;
break;
default:
abort ();
}
scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var),
strategy);
}
static void
scm_i_unget_bytes_unlocked (const unsigned char *buf, size_t len, SCM port);
/* If the next LEN bytes from PORT are equal to those in BYTES, then
return 1, else return 0. Leave the port position unchanged. */
static int
looking_at_bytes (SCM port, const unsigned char *bytes, int len)
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
int i = 0;
while (i < len && scm_peek_byte_or_eof_unlocked (port) == bytes[i])
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
pt->read_buf->cur++;
i++;
}
scm_i_unget_bytes_unlocked (bytes, i, port);
return (i == len);
}
static const unsigned char scm_utf8_bom[3] = {0xEF, 0xBB, 0xBF};
static const unsigned char scm_utf16be_bom[2] = {0xFE, 0xFF};
static const unsigned char scm_utf16le_bom[2] = {0xFF, 0xFE};
static const unsigned char scm_utf32be_bom[4] = {0x00, 0x00, 0xFE, 0xFF};
static const unsigned char scm_utf32le_bom[4] = {0xFF, 0xFE, 0x00, 0x00};
/* Decide what byte order to use for a UTF-16 port. Return "UTF-16BE"
or "UTF-16LE". MODE must be either SCM_PORT_READ or SCM_PORT_WRITE,
and specifies which operation is about to be done. The MODE
determines how we will decide the byte order. We deliberately avoid
reading from the port unless the user is about to do so. If the user
is about to read, then we look for a BOM, and if present, we use it
to determine the byte order. Otherwise we choose big endian, as
recommended by the Unicode Standard. Note that the BOM (if any) is
not consumed here. */
static const char *
decide_utf16_encoding (SCM port, scm_t_port_rw_active mode)
{
if (mode == SCM_PORT_READ
&& SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read
&& looking_at_bytes (port, scm_utf16le_bom, sizeof scm_utf16le_bom))
return "UTF-16LE";
else
return "UTF-16BE";
}
/* Decide what byte order to use for a UTF-32 port. Return "UTF-32BE"
or "UTF-32LE". See the comment above 'decide_utf16_encoding' for
details. */
static const char *
decide_utf32_encoding (SCM port, scm_t_port_rw_active mode)
{
if (mode == SCM_PORT_READ
&& SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read
&& looking_at_bytes (port, scm_utf32le_bom, sizeof scm_utf32le_bom))
return "UTF-32LE";
else
return "UTF-32BE";
}
static void
finalize_iconv_descriptors (void *ptr, void *data)
{
close_iconv_descriptors (ptr);
}
static scm_t_iconv_descriptors *
open_iconv_descriptors (const char *encoding, int reading, int writing)
{
scm_t_iconv_descriptors *id;
iconv_t input_cd, output_cd;
size_t i;
input_cd = (iconv_t) -1;
output_cd = (iconv_t) -1;
for (i = 0; encoding[i]; i++)
if (encoding[i] > 127)
goto invalid_encoding;
if (reading)
{
/* Open an input iconv conversion descriptor, from ENCODING
to UTF-8. We choose UTF-8, not UTF-32, because iconv
implementations can typically convert from anything to
UTF-8, but not to UTF-32 (see
<http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */
/* Assume opening an iconv descriptor causes about 16 KB of
allocation. */
scm_gc_register_allocation (16 * 1024);
input_cd = iconv_open ("UTF-8", encoding);
if (input_cd == (iconv_t) -1)
goto invalid_encoding;
}
if (writing)
{
/* Assume opening an iconv descriptor causes about 16 KB of
allocation. */
scm_gc_register_allocation (16 * 1024);
output_cd = iconv_open (encoding, "UTF-8");
if (output_cd == (iconv_t) -1)
{
if (input_cd != (iconv_t) -1)
iconv_close (input_cd);
goto invalid_encoding;
}
}
id = scm_gc_malloc_pointerless (sizeof (*id), "iconv descriptors");
id->input_cd = input_cd;
id->output_cd = output_cd;
/* Register a finalizer to close the descriptors. */
scm_i_set_finalizer (id, finalize_iconv_descriptors, NULL);
return id;
invalid_encoding:
{
SCM err;
err = scm_from_latin1_string (encoding);
scm_misc_error ("open_iconv_descriptors",
"invalid or unknown character encoding ~s",
scm_list_1 (err));
}
}
static void
close_iconv_descriptors (scm_t_iconv_descriptors *id)
{
if (id->input_cd != (iconv_t) -1)
iconv_close (id->input_cd);
if (id->output_cd != (iconv_t) -1)
iconv_close (id->output_cd);
id->input_cd = (void *) -1;
id->output_cd = (void *) -1;
}
scm_t_iconv_descriptors *
scm_i_port_iconv_descriptors (SCM port, scm_t_port_rw_active mode)
{
scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port);
assert (pti->encoding_mode == SCM_PORT_ENCODING_MODE_ICONV);
if (!pti->iconv_descriptors)
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
const char *precise_encoding;
if (!pt->encoding)
pt->encoding = "ISO-8859-1";
/* If the specified encoding is UTF-16 or UTF-32, then make
that more precise by deciding what byte order to use. */
if (strcmp (pt->encoding, "UTF-16") == 0)
precise_encoding = decide_utf16_encoding (port, mode);
else if (strcmp (pt->encoding, "UTF-32") == 0)
precise_encoding = decide_utf32_encoding (port, mode);
else
precise_encoding = pt->encoding;
pti->iconv_descriptors =
open_iconv_descriptors (precise_encoding,
SCM_INPUT_PORT_P (port),
SCM_OUTPUT_PORT_P (port));
}
return pti->iconv_descriptors;
}
/* The name of the encoding is itself encoded in ASCII. */
void
scm_i_set_port_encoding_x (SCM port, const char *encoding)
{
scm_t_port *pt;
scm_t_port_internal *pti;
scm_t_iconv_descriptors *prev;
/* Set the character encoding for this port. */
pt = SCM_PTAB_ENTRY (port);
pti = SCM_PORT_GET_INTERNAL (port);
prev = pti->iconv_descriptors;
/* In order to handle cases where the encoding changes mid-stream
(e.g. within an HTTP stream, or within a file that is composed of
segments with different encodings), we consider this to be "stream
start" for purposes of BOM handling, regardless of our actual file
position. */
pti->at_stream_start_for_bom_read = 1;
pti->at_stream_start_for_bom_write = 1;
if (encoding_matches (encoding, "UTF-8"))
{
pt->encoding = "UTF-8";
pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8;
}
else if (encoding_matches (encoding, "ISO-8859-1"))
{
pt->encoding = "ISO-8859-1";
pti->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1;
}
else
{
pt->encoding = canonicalize_encoding (encoding);
pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV;
}
pti->iconv_descriptors = NULL;
if (prev)
close_iconv_descriptors (prev);
}
SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0,
(SCM port),
"Returns, as a string, the character encoding that @var{port}\n"
"uses to interpret its input and output.\n")
#define FUNC_NAME s_scm_port_encoding
{
SCM_VALIDATE_PORT (1, port);
return scm_from_latin1_string (SCM_PTAB_ENTRY (port)->encoding);
}
#undef FUNC_NAME
SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0,
(SCM port, SCM enc),
"Sets the character encoding that will be used to interpret all\n"
"port I/O. New ports are created with the encoding\n"
"appropriate for the current locale if @code{setlocale} has \n"
"been called or ISO-8859-1 otherwise\n"
"and this procedure can be used to modify that encoding.\n")
#define FUNC_NAME s_scm_set_port_encoding_x
{
char *enc_str;
SCM_VALIDATE_PORT (1, port);
SCM_VALIDATE_STRING (2, enc);
enc_str = scm_to_latin1_string (enc);
scm_i_set_port_encoding_x (port, enc_str);
free (enc_str);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy",
1, 0, 0, (SCM port),
"Returns the behavior of the port when handling a character that\n"
"is not representable in the port's current encoding.\n"
"It returns the symbol @code{error} if unrepresentable characters\n"
"should cause exceptions, @code{substitute} if the port should\n"
"try to replace unrepresentable characters with question marks or\n"
"approximate characters, or @code{escape} if unrepresentable\n"
"characters should be converted to string escapes.\n"
"\n"
"If @var{port} is @code{#f}, then the current default behavior\n"
"will be returned. New ports will have this default behavior\n"
"when they are created.\n")
#define FUNC_NAME s_scm_port_conversion_strategy
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
scm_t_string_failed_conversion_handler h;
if (scm_is_false (port))
h = scm_i_default_port_conversion_handler ();
else
{
scm_t_port *pt;
SCM_VALIDATE_OPPORT (1, port);
pt = SCM_PTAB_ENTRY (port);
h = pt->ilseq_handler;
}
if (h == SCM_FAILED_CONVERSION_ERROR)
return scm_from_latin1_symbol ("error");
else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK)
return scm_from_latin1_symbol ("substitute");
else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
return scm_from_latin1_symbol ("escape");
else
abort ();
/* Never gets here. */
return SCM_UNDEFINED;
}
#undef FUNC_NAME
SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!",
2, 0, 0,
(SCM port, SCM sym),
"Sets the behavior of the interpreter when outputting a character\n"
"that is not representable in the port's current encoding.\n"
"@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
"@code{'escape}. If it is @code{'error}, an error will be thrown\n"
"when an unconvertible character is encountered. If it is\n"
"@code{'substitute}, then unconvertible characters will \n"
"be replaced with approximate characters, or with question marks\n"
"if no approximately correct character is available.\n"
"If it is @code{'escape},\n"
"it will appear as a hex escape when output.\n"
"\n"
"If @var{port} is an open port, the conversion error behavior\n"
"is set for that port. If it is @code{#f}, it is set as the\n"
"default behavior for any future ports that get created in\n"
"this thread.\n")
#define FUNC_NAME s_scm_set_port_conversion_strategy_x
{
scm_t_string_failed_conversion_handler handler;
if (scm_is_eq (sym, sym_error))
handler = SCM_FAILED_CONVERSION_ERROR;
else if (scm_is_eq (sym, sym_substitute))
handler = SCM_FAILED_CONVERSION_QUESTION_MARK;
else if (scm_is_eq (sym, sym_escape))
handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE;
else
SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym));
if (scm_is_false (port))
scm_i_set_default_port_conversion_handler (handler);
else
{
SCM_VALIDATE_OPPORT (1, port);
SCM_PTAB_ENTRY (port)->ilseq_handler = handler;
}
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
/* The port lock. */
static void
lock_port (void *mutex)
{
scm_i_pthread_mutex_lock ((scm_i_pthread_mutex_t *) mutex);
}
static void
unlock_port (void *mutex)
{
scm_i_pthread_mutex_unlock ((scm_i_pthread_mutex_t *) mutex);
}
void
scm_dynwind_lock_port (SCM port)
#define FUNC_NAME "dynwind-lock-port"
{
scm_i_pthread_mutex_t *lock;
SCM_VALIDATE_OPPORT (SCM_ARG1, port);
scm_c_lock_port (port, &lock);
if (lock)
{
scm_dynwind_unwind_handler (unlock_port, lock, SCM_F_WIND_EXPLICITLY);
scm_dynwind_rewind_handler (lock_port, lock, 0);
}
}
#undef FUNC_NAME
/* Input. */
int
scm_get_byte_or_eof (SCM port)
{
scm_i_pthread_mutex_t *lock;
int ret;
scm_c_lock_port (port, &lock);
ret = scm_get_byte_or_eof_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
int
scm_peek_byte_or_eof (SCM port)
{
scm_i_pthread_mutex_t *lock;
int ret;
scm_c_lock_port (port, &lock);
ret = scm_peek_byte_or_eof_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
static size_t
scm_i_read_bytes_unlocked (SCM port, SCM dst, size_t start, size_t count)
{
size_t filled;
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port);
assert (count <= SCM_BYTEVECTOR_LENGTH (dst));
assert (start + count <= SCM_BYTEVECTOR_LENGTH (dst));
filled = ptob->read (port, dst, start, count);
assert (filled <= count);
return filled;
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* scm_i_read_unlocked is used internally to add bytes to the given port
buffer. If the number of available bytes in the buffer does not
increase after a call to scm_i_read_unlocked, that indicates EOF. */
static void
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_i_read_unlocked (SCM port, scm_t_port_buffer *buf)
{
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
size_t count;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
assert (buf->end < buf->size);
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
count = scm_i_read_bytes_unlocked (port, buf->bytevector, buf->end,
buf->size - buf->end);
buf->end += count;
buf->has_eof_p = scm_from_bool (count == 0);
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
/* Used by an application to read arbitrary number of bytes from an SCM
port. Same semantics as libc read, except that scm_c_read_bytes only
returns less than SIZE bytes if at end-of-file.
Warning: Doesn't update port line and column counts! */
static size_t
scm_c_read_bytes_unlocked (SCM port, SCM dst, size_t start, size_t count)
#define FUNC_NAME "scm_c_read_bytes"
{
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
size_t to_read = count;
scm_t_port *pt;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *read_buf;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
signed char *dst_ptr = SCM_BYTEVECTOR_CONTENTS (dst) + start;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_VALIDATE_OPINPORT (1, port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
read_buf = pt->read_buf;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_WRITE)
scm_flush_unlocked (port);
pt->rw_active = SCM_PORT_READ;
}
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
/* Take bytes first from the port's read buffer. */
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (read_buf->cur < read_buf->end)
{
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
size_t to_copy = count;
to_copy = min (to_copy, read_buf->end - read_buf->cur);
memcpy (dst_ptr, read_buf->buf + read_buf->cur, to_copy);
dst_ptr += to_copy;
to_read -= to_copy;
read_buf->cur += to_copy;
}
while (to_read)
{
/* If the read is smaller than the buffering on the read side of
this port, then go through the buffer. Otherwise fill our
buffer directly. */
if (to_read < pt->read_buffering)
{
size_t to_copy = to_read;
read_buf = scm_fill_input_unlocked (port);
to_copy = min (to_copy, read_buf->end - read_buf->cur);
memcpy (dst_ptr, read_buf->buf + read_buf->cur, to_copy);
if (to_copy == 0)
{
/* Consider that we've read off this EOF. */
read_buf->has_eof_p = SCM_BOOL_F;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
break;
}
dst_ptr += to_copy;
to_read -= to_copy;
read_buf->cur += to_copy;
}
else
{
size_t filled;
filled = scm_i_read_bytes_unlocked (port, dst,
start + count - to_read,
to_read);
if (filled == 0)
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
break;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
to_read -= filled;
dst_ptr += filled;
}
}
return count - to_read;
}
#undef FUNC_NAME
/* Like scm_c_read_bytes, but always proxies reads through the port's
read buffer. Used by an application when it wants to read into a
memory chunk that's not owned by Guile's GC. */
size_t
scm_c_read_unlocked (SCM port, void *buffer, size_t size)
#define FUNC_NAME "scm_c_read"
{
size_t copied = 0;
scm_t_port *pt;
scm_t_port_buffer *read_buf;
scm_t_uint8 *dst = buffer;
SCM_VALIDATE_OPINPORT (1, port);
pt = SCM_PTAB_ENTRY (port);
read_buf = pt->read_buf;
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_WRITE)
scm_flush_unlocked (port);
pt->rw_active = SCM_PORT_READ;
}
while (copied < size)
{
read_buf = scm_fill_input_unlocked (port);
/* Take bytes first from the port's read buffer. */
if (read_buf->cur < read_buf->end)
{
size_t to_copy = size - copied;
to_copy = min (to_copy, read_buf->end - read_buf->cur);
memcpy (dst + copied, read_buf->buf + read_buf->cur, to_copy);
copied += to_copy;
read_buf->cur += to_copy;
}
else
{
/* Consider that we've read off this EOF. */
read_buf->has_eof_p = SCM_BOOL_F;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
break;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
}
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
return copied;
}
2007-12-04 17:57:44 +00:00
#undef FUNC_NAME
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
size_t
scm_c_read_bytes (SCM port, SCM dst, size_t start, size_t count)
{
scm_i_pthread_mutex_t *lock;
size_t ret;
scm_c_lock_port (port, &lock);
ret = scm_c_read_bytes_unlocked (port, dst, start, count);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
size_t
scm_c_read (SCM port, void *buffer, size_t size)
{
scm_i_pthread_mutex_t *lock;
size_t ret;
scm_c_lock_port (port, &lock);
ret = scm_c_read_unlocked (port, buffer, size);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
/* Update the line and column number of PORT after consumption of C. */
static inline void
update_port_lf (scm_t_wchar c, SCM port)
{
switch (c)
{
case '\a':
case EOF:
break;
case '\b':
SCM_DECCOL (port);
break;
case '\n':
SCM_INCLINE (port);
break;
case '\r':
SCM_ZEROCOL (port);
break;
case '\t':
SCM_TABCOL (port);
break;
default:
SCM_INCCOL (port);
break;
}
}
#define SCM_MBCHAR_BUF_SIZE (4)
/* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint.
UTF8_BUF is assumed to contain a valid UTF-8 sequence. */
static scm_t_wchar
utf8_to_codepoint (const scm_t_uint8 *utf8_buf, size_t size)
{
scm_t_wchar codepoint;
if (utf8_buf[0] <= 0x7f)
{
assert (size == 1);
codepoint = utf8_buf[0];
}
else if ((utf8_buf[0] & 0xe0) == 0xc0)
{
assert (size == 2);
codepoint = ((scm_t_wchar) utf8_buf[0] & 0x1f) << 6UL
| (utf8_buf[1] & 0x3f);
}
else if ((utf8_buf[0] & 0xf0) == 0xe0)
{
assert (size == 3);
codepoint = ((scm_t_wchar) utf8_buf[0] & 0x0f) << 12UL
| ((scm_t_wchar) utf8_buf[1] & 0x3f) << 6UL
| (utf8_buf[2] & 0x3f);
}
else
{
assert (size == 4);
codepoint = ((scm_t_wchar) utf8_buf[0] & 0x07) << 18UL
| ((scm_t_wchar) utf8_buf[1] & 0x3f) << 12UL
| ((scm_t_wchar) utf8_buf[2] & 0x3f) << 6UL
| (utf8_buf[3] & 0x3f);
}
return codepoint;
}
/* Read a UTF-8 sequence from PORT. On success, return 0 and set
*CODEPOINT to the codepoint that was read, fill BUF with its UTF-8
representation, and set *LEN to the length in bytes. Return
`EILSEQ' on error. */
static int
get_utf8_codepoint (SCM port, scm_t_wchar *codepoint,
scm_t_uint8 buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
{
#define ASSERT_NOT_EOF(b) \
if (SCM_UNLIKELY ((b) == EOF)) \
goto invalid_seq
#define CONSUME_PEEKED_BYTE() \
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
pt->read_buf->cur++
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
int byte;
scm_t_port *pt;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
*len = 0;
pt = SCM_PTAB_ENTRY (port);
byte = scm_get_byte_or_eof_unlocked (port);
if (byte == EOF)
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
{
*codepoint = EOF;
return 0;
}
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
buf[0] = (scm_t_uint8) byte;
*len = 1;
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
if (buf[0] <= 0x7f)
/* 1-byte form. */
*codepoint = buf[0];
else if (buf[0] >= 0xc2 && buf[0] <= 0xdf)
{
/* 2-byte form. */
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
if (SCM_UNLIKELY ((byte & 0xc0) != 0x80))
goto invalid_seq;
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
CONSUME_PEEKED_BYTE ();
buf[1] = (scm_t_uint8) byte;
*len = 2;
*codepoint = ((scm_t_wchar) buf[0] & 0x1f) << 6UL
| (buf[1] & 0x3f);
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
}
else if ((buf[0] & 0xf0) == 0xe0)
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
{
/* 3-byte form. */
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
if (SCM_UNLIKELY ((byte & 0xc0) != 0x80
|| (buf[0] == 0xe0 && byte < 0xa0)
|| (buf[0] == 0xed && byte > 0x9f)))
goto invalid_seq;
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
CONSUME_PEEKED_BYTE ();
buf[1] = (scm_t_uint8) byte;
*len = 2;
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
if (SCM_UNLIKELY ((byte & 0xc0) != 0x80))
goto invalid_seq;
CONSUME_PEEKED_BYTE ();
buf[2] = (scm_t_uint8) byte;
*len = 3;
*codepoint = ((scm_t_wchar) buf[0] & 0x0f) << 12UL
| ((scm_t_wchar) buf[1] & 0x3f) << 6UL
| (buf[2] & 0x3f);
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
}
else if (buf[0] >= 0xf0 && buf[0] <= 0xf4)
{
/* 4-byte form. */
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
1999-07-04 Gary Houston <ghouston@easynet.co.uk> * strports.c (scm_strprint_obj): bug fix: get pt from the port, not from the parameter obj. (Thanks to Eric Moore.) * ports.h: SCM_CRDY, SCM_CUC, SCM_CRDYP, SCM_SETRDY, SCM_CUNGET, SCM_CGETUN, SCM_CLRDY, SCM_TRY_CLRDY, SCM_N_READY_CHARS: deleted. * strings.c (scm_make_string): throw error if 2nd arg isn't a char. * unif.c (scm_uniform_array_read_x): fix reading from a port. allow non-fports. (scm_uniform_array_write): likewise. 1999-06-29 Gary Houston <ghouston@easynet.co.uk> * ports.c (scm_drain_input): rewritten. * fports.c (local_fclose): check putback_buf. (local_read_flush): likewise. * ports.c (scm_remove_from_port_table): maybe free putback_buf. * ports.h (scm_port): replace cbuf/cbufend/cp with putback_buf/ putback_buf_size. (SCM_INITIAL_PUTBACK_BUF_SIZE): renamed from SCM_INITIAL_CBUF_SIZE. * ports.c (scm_grow_port_cbuf): deleted. (scm_add_to_port_table): initialise putback_buf to 0. remove cbuf stuff. (scm_char_ready_p): check putback_buf (scm_fill_buffer): likewise. (scm_ungetc): rewritten. 1999-06-27 Gary Houston <ghouston@easynet.co.uk> * fports.c (local_fclose): account for push-back buffer. * ports.c (scm_char_ready_p): check the push-back buffer in a new way. * ioext.c (scm_do_read_line): remove the extra code to handle the push-back buffer. * ports.c (scm_getc): don't use SCM_CRDYP etc. * ioext.c (scm_do_read_line): call scm_fill_buffer. * ports.c (scm_ungetc): don't call SCM_CUNGET. reset the read buffer pointers. scm_fill_buffer: new procedure. (scm_getc): call scm_fill_buffer. * ports.h (struct scm_port): saved_read_buf, saved_read_pos, saved_read_end: new fields. 1999-07-04 Gary Houston <ghouston@easynet.co.uk> * configure.in: don't check for ways to violate stdio abstraction.
1999-07-04 11:02:21 +00:00
if (SCM_UNLIKELY (((byte & 0xc0) != 0x80)
|| (buf[0] == 0xf0 && byte < 0x90)
|| (buf[0] == 0xf4 && byte > 0x8f)))
goto invalid_seq;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
CONSUME_PEEKED_BYTE ();
buf[1] = (scm_t_uint8) byte;
*len = 2;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
if (SCM_UNLIKELY ((byte & 0xc0) != 0x80))
goto invalid_seq;
CONSUME_PEEKED_BYTE ();
buf[2] = (scm_t_uint8) byte;
*len = 3;
byte = scm_peek_byte_or_eof_unlocked (port);
ASSERT_NOT_EOF (byte);
if (SCM_UNLIKELY ((byte & 0xc0) != 0x80))
goto invalid_seq;
CONSUME_PEEKED_BYTE ();
buf[3] = (scm_t_uint8) byte;
*len = 4;
*codepoint = ((scm_t_wchar) buf[0] & 0x07) << 18UL
| ((scm_t_wchar) buf[1] & 0x3f) << 12UL
| ((scm_t_wchar) buf[2] & 0x3f) << 6UL
| (buf[3] & 0x3f);
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
}
else
goto invalid_seq;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
return 0;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
invalid_seq:
/* Here we could choose the consume the faulty byte when it's not a
valid starting byte, but it's not a requirement. What Section 3.9
of Unicode 6.0.0 mandates, though, is to not consume a byte that
would otherwise be a valid starting byte. */
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
return EILSEQ;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
#undef CONSUME_PEEKED_BYTE
#undef ASSERT_NOT_EOF
}
/* Read an ISO-8859-1 codepoint (a byte) from PORT. On success, return
0 and set *CODEPOINT to the codepoint that was read, fill BUF with
its UTF-8 representation, and set *LEN to the length in bytes.
Return `EILSEQ' on error. */
static int
get_latin1_codepoint (SCM port, scm_t_wchar *codepoint,
char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
{
*codepoint = scm_get_byte_or_eof_unlocked (port);
if (*codepoint == EOF)
*len = 0;
else
{
*len = 1;
buf[0] = *codepoint;
}
return 0;
}
/* Likewise, read a byte sequence from PORT, passing it through its
input conversion descriptor. */
static int
get_iconv_codepoint (SCM port, scm_t_wchar *codepoint,
char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
{
scm_t_iconv_descriptors *id;
scm_t_uint8 utf8_buf[SCM_MBCHAR_BUF_SIZE];
size_t input_size = 0;
id = scm_i_port_iconv_descriptors (port, SCM_PORT_READ);
for (;;)
{
int byte_read;
char *input, *output;
size_t input_left, output_left, done;
byte_read = scm_get_byte_or_eof_unlocked (port);
if (SCM_UNLIKELY (byte_read == EOF))
{
if (SCM_LIKELY (input_size == 0))
{
*codepoint = (scm_t_wchar) EOF;
*len = input_size;
return 0;
}
else
{
/* EOF found in the middle of a multibyte character. */
scm_i_set_pending_eof (port);
return EILSEQ;
}
}
buf[input_size++] = byte_read;
input = buf;
input_left = input_size;
output = (char *) utf8_buf;
output_left = sizeof (utf8_buf);
done = iconv (id->input_cd, &input, &input_left, &output, &output_left);
if (done == (size_t) -1)
{
int err = errno;
if (SCM_LIKELY (err == EINVAL))
/* The input byte sequence did not form a complete
character. Read another byte and try again. */
continue;
else
return err;
}
else
{
size_t output_size = sizeof (utf8_buf) - output_left;
if (SCM_LIKELY (output_size > 0))
{
/* iconv generated output. Convert the UTF8_BUF sequence
to a Unicode code point. */
*codepoint = utf8_to_codepoint (utf8_buf, output_size);
*len = input_size;
return 0;
}
else
{
/* iconv consumed some bytes without producing any output.
Most likely this means that a Unicode byte-order mark
(BOM) was consumed, which should not be included in the
returned buf. Shift any remaining bytes to the beginning
of buf, and continue the loop. */
memmove (buf, input, input_left);
input_size = input_left;
continue;
}
}
}
}
/* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF
with the byte representation of the codepoint in PORT's encoding, and
set *LEN to the length in bytes of that representation. Return 0 on
success and an errno value on error. */
static SCM_C_INLINE int
get_codepoint (SCM port, scm_t_wchar *codepoint,
char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
{
int err;
scm_t_port *pt = SCM_PTAB_ENTRY (port);
scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port);
if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
err = get_utf8_codepoint (port, codepoint, (scm_t_uint8 *) buf, len);
else if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1)
err = get_latin1_codepoint (port, codepoint, buf, len);
else
err = get_iconv_codepoint (port, codepoint, buf, len);
if (SCM_LIKELY (err == 0))
{
if (SCM_UNLIKELY (pti->at_stream_start_for_bom_read))
{
/* Record that we're no longer at stream start. */
pti->at_stream_start_for_bom_read = 0;
if (pt->rw_random)
pti->at_stream_start_for_bom_write = 0;
/* If we just read a BOM in an encoding that recognizes them,
then silently consume it and read another code point. */
if (SCM_UNLIKELY
(*codepoint == SCM_UNICODE_BOM
&& (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8
|| strcmp (pt->encoding, "UTF-16") == 0
|| strcmp (pt->encoding, "UTF-32") == 0)))
return get_codepoint (port, codepoint, buf, len);
}
update_port_lf (*codepoint, port);
}
else if (pt->ilseq_handler == SCM_ICONVEH_QUESTION_MARK)
{
*codepoint = '?';
err = 0;
update_port_lf (*codepoint, port);
}
return err;
}
/* Read a codepoint from PORT and return it. */
scm_t_wchar
scm_getc_unlocked (SCM port)
#define FUNC_NAME "scm_getc"
{
int err;
size_t len;
scm_t_wchar codepoint;
char buf[SCM_MBCHAR_BUF_SIZE];
err = get_codepoint (port, &codepoint, buf, &len);
if (SCM_UNLIKELY (err != 0))
/* At this point PORT should point past the invalid encoding, as per
R6RS-lib Section 8.2.4. */
scm_decoding_error (FUNC_NAME, err, "input decoding error", port);
return codepoint;
}
#undef FUNC_NAME
scm_t_wchar
scm_getc (SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_t_wchar ret;
scm_c_lock_port (port, &lock);
ret = scm_getc_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0,
(SCM port),
"Return the next character available from @var{port}, updating\n"
"@var{port} to point to the following character. If no more\n"
"characters are available, the end-of-file object is returned.\n"
"\n"
"When @var{port}'s data cannot be decoded according to its\n"
"character encoding, a @code{decoding-error} is raised and\n"
"@var{port} points past the erroneous byte sequence.\n")
#define FUNC_NAME s_scm_read_char
{
scm_t_wchar c;
if (SCM_UNBNDP (port))
port = scm_current_input_port ();
SCM_VALIDATE_OPINPORT (1, port);
c = scm_getc_unlocked (port);
if (EOF == c)
return SCM_EOF_VAL;
return SCM_MAKE_CHAR (c);
}
#undef FUNC_NAME
/* Pushback. */
static void
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_i_unget_bytes_unlocked (const scm_t_uint8 *buf, size_t len, SCM port)
#define FUNC_NAME "scm_unget_bytes"
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *read_buf = pt->read_buf;
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_WRITE)
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_flush_unlocked (port);
pt->rw_active = SCM_PORT_READ;
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (read_buf->cur < len)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* The bytes don't fit directly in the read_buf. */
if (len <= read_buf->cur + (read_buf->size - read_buf->end))
{
/* But they would fit if we shift the not-yet-read bytes from
the read_buf right. Let's do that. */
size_t to_move = read_buf->end - read_buf->cur;
if (to_move > 0)
memmove (read_buf->buf + (read_buf->size - to_move),
read_buf->buf + read_buf->cur,
to_move);
read_buf->end = read_buf->size;
read_buf->cur = read_buf->end - to_move;
}
else
{
/* Bah, have to expand the read_buf for the putback. */
scm_t_port_buffer *new_buf;
size_t buffered = read_buf->end - read_buf->cur;
size_t size = read_buf->size;
while (size < len + buffered)
size *= 2;
new_buf = scm_c_make_port_buffer (size);
new_buf->end = new_buf->size;
new_buf->cur = new_buf->end - buffered;
new_buf->has_eof_p = read_buf->has_eof_p;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
memcpy (new_buf->buf + new_buf->cur, read_buf->buf + read_buf->cur,
buffered);
pt->read_buf = read_buf = new_buf;
}
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
read_buf->cur -= len;
memcpy (read_buf->buf + read_buf->cur, buf, len);
}
#undef FUNC_NAME
void
scm_unget_bytes_unlocked (const unsigned char *buf, size_t len, SCM port)
{
scm_i_unget_bytes_unlocked (buf, len, port);
}
void
scm_unget_byte_unlocked (int c, SCM port)
{
unsigned char byte = c;
scm_i_unget_bytes_unlocked (&byte, 1, port);
}
void
scm_unget_bytes (const unsigned char *buf, size_t len, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_i_unget_bytes_unlocked (buf, len, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
void
scm_unget_byte (int c, SCM port)
{
unsigned char byte = c;
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_i_unget_bytes_unlocked (&byte, 1, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
void
scm_ungetc_unlocked (scm_t_wchar c, SCM port)
#define FUNC_NAME "scm_ungetc"
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port);
char *result;
char result_buf[10];
size_t len;
len = sizeof (result_buf);
if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
{
if (c < 0x80)
{
result_buf[0] = (char) c;
result = result_buf;
len = 1;
}
else
result =
(char *) u32_to_u8 ((uint32_t *) &c, 1, (uint8_t *) result_buf, &len);
}
else if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1 && c <= 0xff)
{
result_buf[0] = (char) c;
result = result_buf;
len = 1;
}
else
result = u32_conv_to_encoding (pt->encoding,
(enum iconv_ilseq_handler) pt->ilseq_handler,
(uint32_t *) &c, 1, NULL,
result_buf, &len);
if (SCM_UNLIKELY (result == NULL || len == 0))
scm_encoding_error (FUNC_NAME, errno,
"conversion to port encoding failed",
SCM_BOOL_F, SCM_MAKE_CHAR (c));
scm_i_unget_bytes_unlocked ((unsigned char *) result, len, port);
if (SCM_UNLIKELY (result != result_buf))
free (result);
if (c == '\n')
SCM_LINUM (port) -= 1;
SCM_DECCOL (port);
}
#undef FUNC_NAME
void
scm_ungetc (scm_t_wchar c, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_ungetc_unlocked (c, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
void
scm_ungets_unlocked (const char *s, int n, SCM port)
{
/* This is simple minded and inefficient, but unreading strings is
* probably not a common operation, and remember that line and
* column numbers have to be handled...
*
* Please feel free to write an optimized version!
*/
while (n--)
scm_ungetc_unlocked (s[n], port);
}
void
scm_ungets (const char *s, int n, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_ungets_unlocked (s, n, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0,
(SCM port),
"Return the next character available from @var{port},\n"
"@emph{without} updating @var{port} to point to the following\n"
"character. If no more characters are available, the\n"
"end-of-file object is returned.\n"
"\n"
"The value returned by\n"
"a call to @code{peek-char} is the same as the value that would\n"
"have been returned by a call to @code{read-char} on the same\n"
"port. The only difference is that the very next call to\n"
"@code{read-char} or @code{peek-char} on that @var{port} will\n"
"return the value returned by the preceding call to\n"
"@code{peek-char}. In particular, a call to @code{peek-char} on\n"
"an interactive port will hang waiting for input whenever a call\n"
"to @code{read-char} would have hung.\n"
"\n"
"As for @code{read-char}, a @code{decoding-error} may be raised\n"
"if such a situation occurs. However, unlike with @code{read-char},\n"
"@var{port} still points at the beginning of the erroneous byte\n"
"sequence when the error is raised.\n")
#define FUNC_NAME s_scm_peek_char
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
{
int err;
SCM result;
scm_t_wchar c;
char bytes[SCM_MBCHAR_BUF_SIZE];
long column, line;
size_t len = 0;
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
if (SCM_UNBNDP (port))
2005-03-02 20:42:01 +00:00
port = scm_current_input_port ();
SCM_VALIDATE_OPINPORT (1, port);
column = SCM_COL (port);
line = SCM_LINUM (port);
err = get_codepoint (port, &c, bytes, &len);
scm_i_unget_bytes_unlocked ((unsigned char *) bytes, len, port);
SCM_COL (port) = column;
SCM_LINUM (port) = line;
if (SCM_UNLIKELY (err != 0))
{
scm_decoding_error (FUNC_NAME, err, "input decoding error", port);
/* Shouldn't happen since `catch' always aborts to prompt. */
result = SCM_BOOL_F;
}
else if (c == EOF)
{
scm_i_set_pending_eof (port);
result = SCM_EOF_VAL;
}
else
result = SCM_MAKE_CHAR (c);
return result;
}
#undef FUNC_NAME
2001-02-09 07:12:05 +00:00
SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0,
(SCM cobj, SCM port),
"Place character @var{cobj} in @var{port} so that it will be\n"
"read by the next read operation. If called multiple times, the\n"
"unread characters will be read again in last-in first-out\n"
"order. If @var{port} is not supplied, the current input port\n"
"is used.")
#define FUNC_NAME s_scm_unread_char
{
int c;
SCM_VALIDATE_CHAR (1, cobj);
if (SCM_UNBNDP (port))
2005-03-02 20:42:01 +00:00
port = scm_current_input_port ();
SCM_VALIDATE_OPINPORT (2, port);
c = SCM_CHAR (cobj);
scm_ungetc_unlocked (c, port);
return cobj;
}
#undef FUNC_NAME
SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0,
(SCM str, SCM port),
"Place the string @var{str} in @var{port} so that its characters will be\n"
"read in subsequent read operations. If called multiple times, the\n"
"unread characters will be read again in last-in first-out order. If\n"
"@var{port} is not supplied, the current-input-port is used.")
#define FUNC_NAME s_scm_unread_string
{
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
int n;
SCM_VALIDATE_STRING (1, str);
if (SCM_UNBNDP (port))
2005-03-02 20:42:01 +00:00
port = scm_current_input_port ();
SCM_VALIDATE_OPINPORT (2, port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
n = scm_i_string_length (str);
while (n--)
scm_ungetc_unlocked (scm_i_string_ref (str, n), port);
return str;
}
#undef FUNC_NAME
2006-06-17 23:15:59 +00:00
/* Manipulating the buffers. */
SCM_SYMBOL (sym_none, "none");
SCM_SYMBOL (sym_line, "line");
SCM_SYMBOL (sym_block, "block");
SCM_DEFINE (scm_setvbuf, "setvbuf", 2, 1, 0,
(SCM port, SCM mode, SCM size),
"Set the buffering mode for @var{port}. @var{mode} can be one\n"
"of the following symbols:\n"
"@table @code\n"
"@item none\n"
"no buffering\n"
"@item line\n"
"line buffering\n"
"@item block\n"
"block buffering, using a newly allocated buffer of @var{size} bytes.\n"
"If @var{size} is omitted, a default size will be used.\n"
"@end table\n\n"
"Only certain types of ports are supported, most importantly\n"
"file ports.")
#define FUNC_NAME s_scm_setvbuf
{
long csize;
scm_t_port *pt;
scm_t_ptob_descriptor *ptob;
scm_t_bits tag_word;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
size_t read_buf_size, write_buf_size;
scm_t_port_buffer *saved_read_buf;
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
pt = SCM_PTAB_ENTRY (port);
ptob = SCM_PORT_DESCRIPTOR (port);
tag_word = SCM_CELL_WORD_0 (port) & ~(SCM_BUF0 | SCM_BUFLINE);
if (scm_is_eq (mode, sym_none))
{
tag_word |= SCM_BUF0;
if (!SCM_UNBNDP (size) && !scm_is_eq (size, SCM_INUM0))
scm_out_of_range (FUNC_NAME, size);
csize = 0;
}
else if (scm_is_eq (mode, sym_line))
{
csize = SCM_UNBNDP (size) ? -1 : scm_to_int (size);
tag_word |= SCM_BUFLINE;
}
else if (scm_is_eq (mode, sym_block))
{
csize = SCM_UNBNDP (size) ? -1 : scm_to_int (size);
}
else
scm_out_of_range (FUNC_NAME, mode);
if (!SCM_UNBNDP (size) && csize < 0)
scm_out_of_range (FUNC_NAME, size);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (csize >= 0)
read_buf_size = write_buf_size = csize;
else
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
read_buf_size = write_buf_size = default_buffer_size;
if (ptob->get_natural_buffer_sizes)
ptob->get_natural_buffer_sizes (port, &read_buf_size, &write_buf_size);
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* Minimum buffer size is one byte. */
if (read_buf_size == 0)
read_buf_size = 1;
if (write_buf_size == 0)
write_buf_size = 1;
if (SCM_OUTPUT_PORT_P (port))
scm_flush_unlocked (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
saved_read_buf = pt->read_buf;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
SCM_SET_CELL_WORD_0 (port, tag_word);
pt->read_buffering = read_buf_size;
pt->read_buf = scm_c_make_port_buffer (read_buf_size);
pt->write_buf = scm_c_make_port_buffer (write_buf_size);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (saved_read_buf && saved_read_buf->cur < saved_read_buf->end)
scm_unget_bytes (saved_read_buf->buf + saved_read_buf->cur,
saved_read_buf->end - saved_read_buf->cur,
port);
if (saved_read_buf)
pt->read_buf->has_eof_p = saved_read_buf->has_eof_p;
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer*
scm_fill_input (SCM port)
{
scm_i_pthread_mutex_t *lock;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *ret;
scm_c_lock_port (port, &lock);
ret = scm_fill_input_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
return ret;
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* Move up to READ_LEN bytes from PORT's read buffer into memory
starting at DEST. Return the number of bytes moved. PORT's
line/column numbers are left unchanged. */
size_t
scm_take_from_input_buffers (SCM port, char *dest, size_t read_len)
{
scm_t_port *pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *read_buf = pt->read_buf;
size_t count = min (read_buf->end - read_buf->cur, read_len);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
memcpy (dest, read_buf->buf + read_buf->cur, count);
read_buf->cur += count;
return count;
}
/* Clear a port's read buffers, returning the contents. */
SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0,
(SCM port),
"This procedure clears a port's input buffers, similar\n"
"to the way that force-output clears the output buffer. The\n"
"contents of the buffers are returned as a single string, e.g.,\n"
"\n"
"@lisp\n"
"(define p (open-input-file ...))\n"
"(drain-input p) => empty string, nothing buffered yet.\n"
"(unread-char (read-char p) p)\n"
"(drain-input p) => initial chars from p, up to the buffer size.\n"
"@end lisp\n\n"
"Draining the buffers may be useful for cleanly finishing\n"
"buffered I/O so that the file descriptor can be used directly\n"
"for further input.")
#define FUNC_NAME s_scm_drain_input
{
SCM result;
scm_t_port *pt;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *read_buf;
long count;
SCM_VALIDATE_OPINPORT (1, port);
pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
read_buf = pt->read_buf;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
count = read_buf->end - read_buf->cur;
if (count)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_uint8 *ptr = read_buf->buf + read_buf->cur;
result = scm_from_port_stringn ((char *) ptr, count, port);
read_buf->cur = read_buf->end;
}
else
result = scm_nullstr;
return result;
}
#undef FUNC_NAME
void
scm_end_input_unlocked (SCM port)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port *pt;
scm_t_port_buffer *buf;
scm_t_off offset;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
pt = SCM_PTAB_ENTRY (port);
buf = SCM_PTAB_ENTRY (port)->read_buf;
offset = buf->cur - buf->end;
assert (pt->rw_random);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
buf->end = buf->cur = 0;
if (offset != 0)
SCM_PORT_DESCRIPTOR (port)->seek (port, offset, SEEK_CUR);
pt->rw_active = SCM_PORT_NEITHER;
}
void
scm_end_input (SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_end_input_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0,
(SCM port),
"Flush the specified output port, or the current output port if @var{port}\n"
"is omitted. The current output buffer contents are passed to the\n"
"underlying port implementation (e.g., in the case of fports, the\n"
"data will be written to the file and the output buffer will be cleared.)\n"
"It has no effect on an unbuffered port.\n\n"
"The return value is unspecified.")
#define FUNC_NAME s_scm_force_output
{
if (SCM_UNBNDP (port))
port = scm_current_output_port ();
else
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPOUTPORT (1, port);
}
scm_flush_unlocked (port);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
static void scm_i_write_unlocked (SCM port, scm_t_port_buffer *src);
void
scm_flush_unlocked (SCM port)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *buf = SCM_PTAB_ENTRY (port)->write_buf;
if (buf->cur < buf->end)
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
scm_i_write_unlocked (port, buf);
SCM_PTAB_ENTRY (port)->rw_active = SCM_PORT_NEITHER;
}
void
scm_flush (SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_flush_unlocked (port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *
scm_fill_input_unlocked (SCM port)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port *pt = SCM_PTAB_ENTRY (port);
scm_t_port_buffer *read_buf = pt->read_buf;
if (read_buf->cur < read_buf->end || scm_is_true (read_buf->has_eof_p))
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
return read_buf;
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_WRITE)
scm_flush_unlocked (pt->port);
pt->rw_active = SCM_PORT_READ;
}
/* It could be that putback caused us to enlarge the buffer; now that
we've read all the bytes we need to shrink it again. */
if (read_buf->size != pt->read_buffering)
read_buf = pt->read_buf = scm_c_make_port_buffer (pt->read_buffering);
else
read_buf->cur = read_buf->end = 0;
scm_i_read_unlocked (port, read_buf);
return read_buf;
}
/* Output. */
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
void
scm_putc (char c, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_putc_unlocked (c, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
void
scm_puts (const char *s, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_puts_unlocked (s, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
static void
scm_i_write_bytes_unlocked (SCM port, SCM src, size_t start, size_t count)
{
size_t written = 0;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port);
assert (count <= SCM_BYTEVECTOR_LENGTH (src));
assert (start + count <= SCM_BYTEVECTOR_LENGTH (src));
do
written += ptob->write (port, src, start + written, count - written);
while (written < count);
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
assert (written == count);
}
static void
scm_i_write_unlocked (SCM port, scm_t_port_buffer *src)
{
size_t start, count;
assert (src->cur < src->end);
assert (src->end <= src->size);
/* Update cursors before attempting to write, assuming that I/O errors
are sticky. That way if the write throws an error, causing the
computation to abort, and possibly causing the port to be collected
by GC when it's open, any subsequent close-port / force-output
won't signal *another* error. */
start = src->cur;
count = src->end - src->cur;
src->cur = src->end = 0;
scm_i_write_bytes_unlocked (port, src->bytevector, start, count);
}
/* Used by an application to write arbitrary number of bytes to an SCM
port. Similar semantics as libc write. However, unlike libc write,
scm_c_write writes the requested number of bytes.
Warning: Doesn't update port line and column counts! */
static size_t
scm_c_write_bytes_unlocked (SCM port, SCM src, size_t start, size_t count)
#define FUNC_NAME "scm_c_write_bytes"
{
scm_t_port *pt;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *write_buf;
SCM_VALIDATE_OPOUTPORT (1, port);
pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
write_buf = pt->write_buf;
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_READ)
scm_end_input_unlocked (port);
pt->rw_active = SCM_PORT_WRITE;
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
if (count < write_buf->size)
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
{
/* Make it so that write_buf->end is only nonzero if there are
buffered bytes already. */
if (write_buf->cur == write_buf->end)
write_buf->cur = write_buf->end = 0;
/* We buffer writes that are smaller in size than the write
buffer. If the buffer is too full to hold the new data, we
flush it beforehand. Otherwise it could be that the buffer is
full after filling it with the new data; if that's the case, we
flush then instead. */
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
if (write_buf->end + count > write_buf->size)
scm_i_write_unlocked (port, write_buf);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
memcpy (write_buf->buf + write_buf->end,
SCM_BYTEVECTOR_CONTENTS (src) + start,
count);
write_buf->end += count;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (write_buf->end == write_buf->size)
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
scm_i_write_unlocked (port, write_buf);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
}
else
{
/* Our write would overflow the buffer. Flush buffered bytes (if
needed), then write our bytes with just one syscall. */
if (write_buf->cur < write_buf->end)
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
scm_i_write_unlocked (port, write_buf);
scm_i_write_bytes_unlocked (port, src, start, count);
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
}
return count;
}
#undef FUNC_NAME
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
/* Like scm_c_write_bytes, but always writes through the write buffer.
Used when an application wants to write bytes stored in an area not
managed by GC. */
void
scm_c_write_unlocked (SCM port, const void *ptr, size_t size)
#define FUNC_NAME "scm_c_write"
{
scm_t_port *pt;
scm_t_port_buffer *write_buf;
size_t written = 0;
const scm_t_uint8 *src = ptr;
SCM_VALIDATE_OPOUTPORT (1, port);
pt = SCM_PTAB_ENTRY (port);
write_buf = pt->write_buf;
if (pt->rw_random)
{
if (pt->rw_active == SCM_PORT_READ)
scm_end_input_unlocked (port);
pt->rw_active = SCM_PORT_WRITE;
}
while (written < size)
{
size_t to_write = write_buf->size - write_buf->end;
if (to_write > size - written)
to_write = size - written;
memcpy (write_buf->buf + write_buf->end, src, to_write);
write_buf->end += to_write;
written += to_write;
src += to_write;
if (write_buf->end == write_buf->size)
scm_i_write_unlocked (port, write_buf);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
}
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
void
scm_c_write (SCM port, const void *ptr, size_t size)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_c_write_unlocked (port, ptr, size);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
void
scm_c_write_bytes (SCM port, SCM src, size_t start, size_t count)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_c_write_bytes_unlocked (port, src, start, count);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
/* scm_lfwrite
*
* This function differs from scm_c_write; it updates port line and
* column, flushing line-buffered ports when appropriate. */
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
void
scm_lfwrite_unlocked (const char *ptr, size_t size, SCM port)
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
int saved_line;
scm_c_write_unlocked (port, ptr, size);
saved_line = SCM_LINUM (port);
for (; size; ptr++, size--)
update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port);
/* Handle line buffering. */
if ((SCM_CELL_WORD_0 (port) & SCM_BUFLINE) && saved_line != SCM_LINUM (port))
scm_flush_unlocked (port);
}
void
scm_lfwrite (const char *ptr, size_t size, SCM port)
{
scm_i_pthread_mutex_t *lock;
scm_c_lock_port (port, &lock);
scm_lfwrite_unlocked (ptr, size, port);
if (lock)
scm_i_pthread_mutex_unlock (lock);
}
/* Write STR to PORT from START inclusive to END exclusive. */
void
scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port)
{
if (end == (size_t) -1)
end = scm_i_string_length (str);
scm_i_display_substring (str, start, end, port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
/* Querying and setting positions, and character availability. */
SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0,
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
(SCM port),
"Return @code{#t} if a character is ready on input @var{port}\n"
"and return @code{#f} otherwise. If @code{char-ready?} returns\n"
"@code{#t} then the next @code{read-char} operation on\n"
"@var{port} is guaranteed not to hang. If @var{port} is a file\n"
"port at end of file then @code{char-ready?} returns @code{#t}.\n"
"\n"
"@code{char-ready?} exists to make it possible for a\n"
"program to accept characters from interactive ports without\n"
"getting stuck waiting for input. Any input editors associated\n"
"with such ports must make sure that characters whose existence\n"
"has been asserted by @code{char-ready?} cannot be rubbed out.\n"
"If @code{char-ready?} were to return @code{#f} at end of file,\n"
"a port at end of file would be indistinguishable from an\n"
"interactive port that has no ready characters.")
#define FUNC_NAME s_scm_char_ready_p
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
scm_t_port *pt;
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_t_port_buffer *read_buf;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
if (SCM_UNBNDP (port))
port = scm_current_input_port ();
/* It's possible to close the current input port, so validate even in
this case. */
SCM_VALIDATE_OPINPORT (1, port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
pt = SCM_PTAB_ENTRY (port);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
read_buf = pt->read_buf;
if (read_buf->cur < read_buf->end || scm_is_true (read_buf->has_eof_p))
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
/* FIXME: Verify that a whole character is available? */
return SCM_BOOL_T;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
else
{
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port);
if (ptob->input_waiting)
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
return scm_from_bool (ptob->input_waiting (port));
else
return SCM_BOOL_T;
}
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#undef FUNC_NAME
SCM_DEFINE (scm_seek, "seek", 3, 0, 0,
(SCM fd_port, SCM offset, SCM whence),
"Sets the current position of @var{fd_port} to the integer\n"
"@var{offset}, which is interpreted according to the value of\n"
"@var{whence}.\n"
"\n"
"One of the following variables should be supplied for\n"
"@var{whence}:\n"
"@defvar SEEK_SET\n"
"Seek from the beginning of the file.\n"
"@end defvar\n"
"@defvar SEEK_CUR\n"
"Seek from the current position.\n"
"@end defvar\n"
"@defvar SEEK_END\n"
"Seek from the end of the file.\n"
"@end defvar\n"
"If @var{fd_port} is a file descriptor, the underlying system\n"
"call is @code{lseek}. @var{port} may be a string port.\n"
"\n"
"The value returned is the new position in the file. This means\n"
"that the current position of a port can be obtained using:\n"
"@lisp\n"
"(seek port 0 SEEK_CUR)\n"
"@end lisp")
#define FUNC_NAME s_scm_seek
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
int how;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
fd_port = SCM_COERCE_OUTPORT (fd_port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
how = scm_to_int (whence);
if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END)
SCM_OUT_OF_RANGE (3, whence);
if (SCM_OPPORTP (fd_port))
{
scm_t_port *pt = SCM_PTAB_ENTRY (fd_port);
scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (fd_port);
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (fd_port);
off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
off_t_or_off64_t rv;
if (!ptob->seek || !pt->rw_random)
SCM_MISC_ERROR ("port is not seekable",
scm_cons (fd_port, SCM_EOL));
/* FIXME: Avoid flushing buffers for SEEK_CUR with an offset of
0. */
if (pt->rw_active == SCM_PORT_READ)
scm_end_input_unlocked (pt->port);
else if (pt->rw_active == SCM_PORT_WRITE)
scm_flush_unlocked (pt->port);
pt->rw_active = SCM_PORT_NEITHER;
rv = ptob->seek (fd_port, off, how);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
/* Set stream-start flags according to new position. */
pti->at_stream_start_for_bom_read = (rv == 0);
pti->at_stream_start_for_bom_write = (rv == 0);
scm_i_clear_pending_eof (fd_port);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
return scm_from_off_t_or_off64_t (rv);
}
else /* file descriptor?. */
{
off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
off_t_or_off64_t rv;
rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how);
if (rv == -1)
SCM_SYSERROR;
return scm_from_off_t_or_off64_t (rv);
}
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#undef FUNC_NAME
#ifndef O_BINARY
#define O_BINARY 0
#endif
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
/* Mingw has ftruncate(), perhaps implemented above using chsize, but
doesn't have the filename version truncate(), hence this code. */
#if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
static int
truncate (const char *file, off_t length)
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
int ret, fdes;
fdes = open (file, O_BINARY | O_WRONLY);
if (fdes == -1)
return -1;
ret = ftruncate (fdes, length);
if (ret == -1)
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
int save_errno = errno;
close (fdes);
errno = save_errno;
return -1;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
return close (fdes);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0,
(SCM object, SCM length),
"Truncate file @var{object} to @var{length} bytes. @var{object}\n"
"can be a filename string, a port object, or an integer file\n"
"descriptor.\n"
"The return value is unspecified.\n"
"\n"
"For a port or file descriptor @var{length} can be omitted, in\n"
"which case the file is truncated at the current position (per\n"
"@code{ftell} above).\n"
"\n"
"On most systems a file can be extended by giving a length\n"
"greater than the current size, but this is not mandatory in the\n"
"POSIX standard.")
#define FUNC_NAME s_scm_truncate_file
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
int rv;
/* "object" can be a port, fdes or filename.
Negative "length" makes no sense, but it's left to truncate() or
ftruncate() to give back an error for that (normally EINVAL).
*/
if (SCM_UNBNDP (length))
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
/* must supply length if object is a filename. */
if (scm_is_string (object))
SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL);
length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR));
}
object = SCM_COERCE_OUTPORT (object);
if (scm_is_integer (object))
{
off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object),
c_length));
}
else if (SCM_OPOUTPORTP (object))
{
off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (object);
if (!ptob->truncate)
SCM_MISC_ERROR ("port is not truncatable", SCM_EOL);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
scm_i_clear_pending_eof (object);
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
if (SCM_INPUT_PORT_P (object))
scm_end_input_unlocked (object);
scm_flush_unlocked (object);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
ptob->truncate (object, c_length);
rv = 0;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
else
{
off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
char *str = scm_to_locale_string (object);
int eno;
SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length));
eno = errno;
free (str);
errno = eno;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
if (rv == -1)
SCM_SYSERROR;
return SCM_UNSPECIFIED;
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0,
(SCM port),
"Return the current line number for @var{port}.\n"
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
"\n"
"The first line of a file is 0. But you might want to add 1\n"
"when printing line numbers, since starting from 1 is\n"
"traditional in error messages, and likely to be more natural to\n"
"non-programmers.")
#define FUNC_NAME s_scm_port_line
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
return scm_from_long (SCM_LINUM (port));
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
}
#undef FUNC_NAME
SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0,
(SCM port, SCM line),
"Set the current line number for @var{port} to @var{line}. The\n"
"first line of a file is 0.")
#define FUNC_NAME s_scm_set_port_line_x
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0,
(SCM port),
"Return the current column number of @var{port}.\n"
"If the number is\n"
"unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
"- i.e. the first character of the first line is line 0, column 0.\n"
"(However, when you display a file position, for example in an error\n"
"message, we recommend you add 1 to get 1-origin integers. This is\n"
"because lines and column numbers traditionally start with 1, and that is\n"
"what non-programmers will find most natural.)")
#define FUNC_NAME s_scm_port_column
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
return scm_from_int (SCM_COL (port));
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0,
(SCM port, SCM column),
"Set the current column of @var{port}. Before reading the first\n"
"character on a line the column should be 0.")
#define FUNC_NAME s_scm_set_port_column_x
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0,
(SCM port),
"Return the filename associated with @var{port}, or @code{#f}\n"
"if no filename is associated with the port.")
#define FUNC_NAME s_scm_port_filename
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
return SCM_FILENAME (port);
}
#undef FUNC_NAME
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0,
(SCM port, SCM filename),
"Change the filename associated with @var{port}, using the current input\n"
"port if none is specified. Note that this does not change the port's\n"
"source of data, but only the value that is returned by\n"
"@code{port-filename} and reported in diagnostic output.")
#define FUNC_NAME s_scm_set_port_filename_x
{
port = SCM_COERCE_OUTPORT (port);
SCM_VALIDATE_OPENPORT (1, port);
/* We allow the user to set the filename to whatever he likes. */
SCM_SET_FILENAME (port, filename);
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
/* Implementation helpers for port printing functions. */
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
void
scm_print_port_mode (SCM exp, SCM port)
{
scm_puts_unlocked (SCM_CLOSEDP (exp)
? "closed: "
: (SCM_RDNG & SCM_CELL_WORD_0 (exp)
? (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
? "input-output: "
: "input: ")
: (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
? "output: "
: "bogus: ")),
port);
}
int
scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED)
{
char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp));
if (!type)
type = "port";
scm_puts_unlocked ("#<", port);
scm_print_port_mode (exp, port);
scm_puts_unlocked (type, port);
scm_putc_unlocked (' ', port);
scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port);
scm_putc_unlocked ('>', port);
return 1;
}
/* Iterating over all ports. */
struct for_each_data
{
void (*proc) (void *data, SCM p);
void *data;
};
static SCM
for_each_trampoline (void *data, SCM port, SCM result)
{
struct for_each_data *d = data;
d->proc (d->data, port);
return result;
}
void
scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data)
{
struct for_each_data d;
d.proc = proc;
d.data = data;
scm_c_weak_set_fold (for_each_trampoline, &d, SCM_EOL,
scm_i_port_weak_set);
}
static void
scm_for_each_trampoline (void *data, SCM port)
{
scm_call_1 (SCM_PACK_POINTER (data), port);
}
SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0,
(SCM proc),
"Apply @var{proc} to each port in the Guile port table\n"
"in turn. The return value is unspecified. More specifically,\n"
"@var{proc} is applied exactly once to every port that exists\n"
"in the system at the time @code{port-for-each} is invoked.\n"
"Changes to the port table while @code{port-for-each} is running\n"
"have no effect as far as @code{port-for-each} is concerned.")
#define FUNC_NAME s_scm_port_for_each
{
SCM_VALIDATE_PROC (1, proc);
scm_c_port_for_each (scm_for_each_trampoline, SCM_UNPACK_POINTER (proc));
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
static void
flush_output_port (void *closure, SCM port)
{
if (SCM_OPOUTPORTP (port))
scm_flush_unlocked (port);
}
SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0,
(),
"Equivalent to calling @code{force-output} on\n"
"all open output ports. The return value is unspecified.")
#define FUNC_NAME s_scm_flush_all_ports
{
scm_c_port_for_each (&flush_output_port, NULL);
return SCM_UNSPECIFIED;
}
#undef FUNC_NAME
* ioext.c (scm_do_read_line): Rewritten to use memchr to find the newline. A bit faster, and definitely hairier. (scm_read_line): Count newlines here instead. * strings.c (scm_take_str): New function. (scm_take0str): Reimplement in terms of scm_take_str. * strings.h (scm_take_str): New declaration. * ioext.c (scm_read_line): Use scm_take_str, to avoid copying the string. Add some simple-minded support for line buffered ports. * ports.h (SCM_BUFLINE): New flag for ports. * init.c (scm_init_standard_ports): Request line-buffering on the standard output port. * * ports.c (scm_mode_bits): Recognize 'l' as a request for line buffering. (scm_putc, scm_puts, scm_lfwrite): If the port is line-buffered, and there's a newline to be written, flush the port. * ports.c: (scm_lseek): clear buffers even if just reading current position. * fports.c (local_fclose): call local_fflush unconditionally. (various): don't use the scm_must... memory procs. * ports.h (scm_port): make read_pos a pointer to const. strports.c: take care of rw_active and rw_randow. fports.c: scm_fport_drain_input: removed. do it all in ports.c. strports.c (scm_mkstrport): check that pos is reasonable. ioext.c (scm_ftell, scm_fseek): use lseek. (SCM_CLEAR_BUFFERS): macro deleted. ioext.c (redirect_port: use ptob fflush, read_flush. ports.h (scm_ptobfuns): add ftruncate. ports.c (scm_newptob): set ftruncate. adjust ptob tables. * ports.c (scm_ftruncate): new procedure. fports.c (local_ftrunate), strports.c (str_ftruncate): new procs. strports.c (st_seek, st_grow_port): new procs. fports.h (scm_port): change size types from int to off_t. ports.c (scm_init_ports): initialise the seek symbols here instead of in ioext.c. strports.c (scm_call_with_output_string): start with an empty string, so seek and ftruncate can be used. * ports.h (scm_ptobfuns): add a read_flush procedure which is the equivalent to fflush for the read buffer. * ports.c (scm_newptob): set read_flush. ports.c (void_port_ptob): set read_flush. fports.c (local_read_flush): new proc. add to ptob. strport.c (st_read_flush): likewise. vport.c (sf_read_flush): likewise. fports.h (struct scm_fport): remove random member. there's nothing left but fdes. leaving it as a struct to allow for future changes. fports.c: replace usage of scm_fport::random with scm_port::rw_random. ports.c: (scm_putc, scm_puts, scm_lfwrite): call the read_flush ptob proc if the read buffer is filled. * ports.h (scm_port): add a rw_random member and replace reading and writing members with rw_active member. SCM_PORT_READ/SCM_PORT_WRITE: new values. * ports.h (struct scm_port_table): add writing and reading members to replace write_needs_seek: it isn't good enough for non-fports. ports.c, ioext.c, fports.c: corresponding changes. (struct scm_port_table): give it a typedef and rename to scm_port. ports.c, fports.c, strports.c, vports.c, ioext.c, ports.h: corresponding changes. * ports.c (scm_newptob): bugfix: set seek member. * * (scm_lseek): new procedure, using code from ioext.c:scm_fseek and generalised to all port types. * scmsigs.c (scm_init_scmsigs): set the SA_RESTART flag for all signals (it was only being done for handlers installed from Scheme). Otherwise (for example) SIGSTOP followed by SIGCONT on an interpreter waiting for input caused an EINTR error from read. * ports.h (struct scm_port_table): make all the char members unsigned, so they convert to int without becoming negative if large. * fports.c (scm_fdes_wait_for_input): forgot to check compilation with threads enabled. rename this procedure to fport_wait_for_input and take a port instead of a fdes. use scm_fport_input_waiting_p instead of scm_fdes_waiting_p. * readline.c (scm_readline): Applied a patch from Greg Harvey to get readline support working again: use fdopen to get FILE objects. * gc.c (scm_init_storage): install an atexit proc to flush the ports. (cleanup): the new proc. it sets a global variable which can be checked by the ptob flush procs to avoid trying to throw exceptions during exit. not very pleasant but it seems more reliable. * fports.c (local_fflush): check terminating variable and if set don't throw exception. * CHECKME: that the atexit proc is installed if unexec used. * throw.c (scm_handle_by_message): don't flush all ports here. it still causes bus errors. * fports.h (SCM_FPORT_CLEAR_BUFFERS): rename to SCM_CLEAR_BUFFERS and move to ioext.c. * fports.c (scm_fdes_waiting_p): merged into fport_input_waiting_p. * ports.c (scm_char_ready_p): check the port buffer and call the ptob entry if needed. * ports.h (scm_ptobfuns): input_waiting_p added. change all the ptob initialisers. use it in char-ready * ioext.c (scm_do_read_line): moved from ports.c. make it static. * vports.c (sfflush): modified to write a char (since softports currently use shortbuf.) * fports.c (scm_standard_stream_to_port): moved to init.c and made static. * init.c (scm_init_standard_ports): make stdout and stderr unbuffered if connected to a terminal. with stdio they were line-buffered by default. * ports.h (scm_ptobfuns): change fflush return to void. change flush proc definitions. * strports.c (scm_call_with_output_string): get size from buffer instead of port stream. (scm_strprint_obj): likewise. (st_flush): new proc. * ports.h (struct scm_port_table): added write_end member, as an optimisation. set it where write_buf_size is set. * ports.h (struct scm_port_table): change stream from void * back to SCM. SCM presumably must be large enough to hold a pointer (and probably vice versa but who knows.) (SCM_SSTREAM): deleted. change users back to SCM_STREAM. (scm_puts): rewritten * fports.c (local_ffwrite, local_fputs): removed. * strports.c (stputc, stputs, stwrite): dyked out (FIXME) * vports.c (sfputc, sfputs, sfwrite) likewise. * ports.c (write_void_port, puts_void_port): removed. (putc_void_port, getc_void_port, fgets_void_port): likewise. * ports.c (scm_lfwrite): rewritten using fport.c version. * fports.c (local_fputc): deleted. * ports.c (scm_add_to_port_table): initialise write_needs_seek. * ports.h (scm_ptobfuns): add seek function pointer. * fports.c: set it to local_seek, new procedure. * fports.h (SCM_MAYBE_DRAIN_INPUT): moved to ports.c. use ptob for seek. take ptob instead of fport arg. * ports.h (struct scm_port_table): new member write_needs_seek, replaces reading member in fport struct. * vports.c (sfgetc): store the getted char into the buffer. rename to sf_fill_buffer and install it for fill-buffer in ptob. the Scheme interface is still a procedure that gets a char. (scm_make_soft_port): set up the port buffer (shortbuf). * fports.c (local_fgetc, local_fgets): deleted. * strports.c (stgetc): likewise. * ports.c: scm_generic_fgets: likewise. * ports.h (scm_ptobfuns): add fill_buffer. * ports.c (scm_newptob): assign it. * strports.c (scm_mkstrport): set up the buffer. put just the string into the stream, not cons (pos stream). (stfill_buffer): new proc. * ports.h: fport buffer moved into port table: to be used for all port types. * throw.c (scm_handle_by_message): flush ports at exit. * socket.c (scm_sock_fd_to_port): use scm_fdes_to_port. (scm_getsockopt, scm_setsockopt, scm_shutdown, scm_connect, scm_bind, scm_listen, scm_accept, scm_getsockname, scm_getpeername, scm_recv, scm_send, scm_recvfrom, scm_sendto, use SCM_FPORT_FDES. use SCM_OPFPORTP not SCM_FPORTP. * posix.c (scm_getgroups): use SCM_ALLOW/DEFER_INTS. (scm_ttyname): use SCM_FPORT_FDES. (scm_tcgetpgrp, scm_tcsetpgrp): likewise. * ioext.c (scm_isatty_p): use SCM_FPORT_FDES. (scm_fdes_to_ports): modified. (scm_fdopen): use scm_fdes_to_port. * ports.c (scm_init_ports): don't try to flush ports using atexit(). it's too late, errors will cause SEGV. * fports.c (scm_fport_buffer_add): new procedure. * fports.h (SCM_FDES_RANDOM_P): new macro. use it in scm_fdes_to_port and scm_redirect_port. * ioext.c (scm_redirect_port): use setvbuf to set buffers in the new port. reset fp->random. * fports.c (scm_fdes_to_port), ports.c (scm_void_port), filesys.c (scm_opendir): restore defer interrupts while the port is constructed. * (scm_setvbuf): if mode is _IOFBF and size is not supplied, derive buffer size from fdes or use a default. (scm_fdes_to_port): use setvbuf instead of creating the buffers directly. vports.c (various places): use SCM_SSTREAM. strports.c: likewise. * gdbint.c: likewise. * ports.h (SCM_SSTREAM): new macro. * fports.c (scm_input_waiting_p): use scm_return_first, since port may be removed from the stack by the tail call to scm_fdes_waiting_p. * fports.h (SCM_CLEAR_BUFFERS): new macro. * ports.c (scm_force_output): call scm_fflush. * print.c (scm_newline): don't check errno for EPIPE (it wouldn't * reach this point.) don't flush port (if scm_cur_outp). * fports.h (SCM_FPORT_FDES): new macro. * vports.c (sfflush): don't need to set errno. * ports.c: install scm_flush_all_ports to be run on exit. ports.c fports.c ioext.c posix.c socket.c net_db.c filesys.c: removed all uses of SCM_DEFER/ALLOW ints for now. they were mainly just protecting errno. some may need to be put back. * scmsigs.c (take_signal): save and restore errno while this proc runs. *fports.c (print_pipe_port, local_pclose, scm_pipob): deleted. * open-pipe, close-pipe are emulated in (ice-9 popen) ports.c (scm_ports_prehistory): don't init scm_pipob. ports.h (scm_tc16_pipe): deleted. posix.c (scm_open_pipe, scm_close_pipe): deleted. * ioext.c (scm_primitive_move_to_fdes): use fport. * fport.c (scm_fport_fill_buffer): flush write buffer if needed. change arg type from scm_fport to SCM port. fport.h (SCM_SETFDES): removed. (SCM_MAYBE_DRAIN_INPUT): new macro. * ioext.c (scm_dup_to_fdes): use SCM_FSTREAM. (scm_ftell): always use lseek and account for the buffer. (scm_fileno): use fport buffer. (scm_fseek): clear fport buffers. always use lseek. * posix.c (scm_pipe): use fport buffer. * unif.c: include fports.h instead of genio.h. * fports.c (scm_fdes_wait_for_input, scm_fport_fill_buffer): new procedures. (local_fgetc): use them. (local_ffwrite): use buffer. (local_fgets): use buffer. (scm_setbuf0): deleted. (scm_setvbuf): set the buffer. (scm_setfileno): deleted. (scm_evict_ports): set fdes directly. * (scm_freopen): deleted. doesn't seem useful in Guile. (scm_stdio_to_port): deleted. fports.h (struct scm_fport): add shortbuf member to avoid separate code for unbuffered ports. (SCM_FPORTP, SCM_OPFPORTP, SCM_OPINFPORTP, SCM_OPOUTFPORTP): moved from ports.h. * genio.c, genio.h: move contents into ports.c, ports.h. The division wasn't useful. * fports.c, fports.h (scm_fport_drain_input): new procedure. * ports.c (scm_drain_input): call scm_fport_drain_input. * scm_fdes_waiting_p: new procedure. * fports.c (scm_fdes_to_port): allocate read and/or write buffers. (scm_input_waiting_p): check the buffer. (local_fgetc, local_fflush, local_fputc): likewise. * fports.h (scm_fport): read/write_buf,_pos,_buf_end,,_buf_size: new members. * init.c (scm_init_standard_ports): pass fdes instead of FILE *. * * ports.c (scm_drain_input): new procedure. ports.h: prototype. * fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY, pre_read, pre_write): removed. (local_fputc, local_fputs, local_ffwrite): use write, not stdio. (scm_standard_stream_to_port): change first arg from FILE * to int fdes. (local_fflush): flush fdes, not FILE *. * fports.h (SCM_NOFTELL): removed. * genio.c, ports.c: don't include filesys.h. * genio.c (scm_getc): don't use scm_internal_select if FPORT. do it in fports.c:local_fgetc. * genio.c: don't use SCM_SYSCALL when calling ptob procedures. do it where it's needed in the port smobs. * filesys.c (scm_input_waiting_p): moved to fports.c, stdio buffer support removed. take SCM arg, not FILE *. * filesys.h: prototype moved too. * fports.c (scm_fdes_to_port): new procedure. (local_fgetc): use read not fgetc. (local_fclose): use close, not fclose. (local_fgets): use read, not fgets * fports.h: prototype for scm_fdes_to_port. * fports.h (scm_fport): new struct. * fports.c (scm_open_file): use open, not fopen. #include fcntl.h * ports.h (struct scm_port_table): change stream from SCM to void *. * ports.c (scm_add_to_port_table): check for memory allocation error. (scm_prinport): remove MSDOS hair. (scm_void_port): set stream to 0 instead of SCM_BOOL_F. (scm_close_port): don't throw errors: do it in fports.c.
1999-06-09 12:19:58 +00:00
/* Void ports. */
2001-06-14 19:50:43 +00:00
scm_t_bits scm_tc16_void_port = 0;
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
static size_t
void_port_read (SCM port, SCM dst, size_t start, size_t count)
{
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
return 0;
}
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
static size_t
void_port_write (SCM port, SCM src, size_t start, size_t count)
{
Port read/write functions take bytevectors This will allow better Scheme integration for ports. * libguile/ports.h (scm_t_port_buffer): Change "holder" member to be a bytevector defined to have "buf" as its starting point. (scm_t_ptob_descriptor): Change read and write functions to take bytevectors as arguments and to return the number of octets read or written. (scm_make_port_type): Adapt accordingly. (scm_c_read_bytes, scm_c_write_bytes): New functions that take bytevectors. * libguile/ports.c (scm_make_port_type): Adapt to read/write function prototype change. (scm_c_make_port_buffer): Arrange to populate the "bytevector" field. (scm_i_read_bytes_unlocked): New function. (scm_i_read_unlocked): Use scm_i_read_bytes_unlocked. (scm_c_read_bytes_unlocked): New function. (scm_c_read_unlocked): Update comment, and always go through the buffer. (scm_c_read_bytes): New function. (scm_flush_unlocked): Use scm_i_write_unlocked instead of the port's write function. (scm_i_write_bytes_unlocked): New function. (scm_i_write_unlocked): Use scm_i_write_bytes_unlocked. (scm_c_write_bytes_unlocked): New function. (scm_c_write_unlocked): Always write through the buffer. (scm_c_write_bytes): New function. (scm_truncate_file): Remove unused variable. (void_port_read, void_port_write): Adapt to read/write prototype change. * libguile/fports.c (fport_read, fport_write): * libguile/r6rs-ports.c (bytevector_input_port_read) (custom_binary_input_port_read, bytevector_output_port_write) (custom_binary_output_port_write, transcoded_port_write) (transcoded_port_read): Adapt to read/write prototype change. (scm_get_bytevector_n, scm_get_bytevector_n_x) (scm_get_bytevector_all): Use scm_c_read_bytes. (scm_put_bytevector): Use scm_c_write_bytes. * libguile/strports.c (string_port_read, string_port_write): * libguile/vports.c (soft_port_write, soft_port_read): Adapt to read/write prototype change. * test-suite/standalone/test-scm-c-read.c (custom_port_read): Fix for read API change.
2016-04-11 18:40:03 +02:00
return count;
}
static SCM
scm_i_void_port (long mode_bits)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
return scm_c_make_port (scm_tc16_void_port, mode_bits, 0);
}
SCM
scm_void_port (char *mode_str)
{
return scm_i_void_port (scm_mode_bits (mode_str));
}
SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0,
(SCM mode),
"Create and return a new void port. A void port acts like\n"
2002-03-15 10:37:40 +00:00
"@file{/dev/null}. The @var{mode} argument\n"
"specifies the input/output modes for this port: see the\n"
"documentation for @code{open-file} in @ref{File Ports}.")
#define FUNC_NAME s_scm_sys_make_void_port
{
return scm_i_void_port (scm_i_mode_bits (mode));
}
#undef FUNC_NAME
1998-10-05 21:05:04 +00:00
/* Initialization. */
static void
scm_init_ice_9_ports (void)
{
#include "libguile/ports.x"
/* lseek() symbols. */
scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET));
scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR));
scm_c_define ("SEEK_END", scm_from_int (SEEK_END));
/* These bindings are used when boot-9 turns `current-input-port' et
al into parameters. They are then removed from the guile module. */
scm_c_define ("%current-input-port-fluid", cur_inport_fluid);
scm_c_define ("%current-output-port-fluid", cur_outport_fluid);
scm_c_define ("%current-error-port-fluid", cur_errport_fluid);
scm_c_define ("%current-warning-port-fluid", cur_warnport_fluid);
}
void
scm_init_ports (void)
{
Generic port facility provides buffering uniformly * libguile/ports.h (struct scm_t_port_buffer): New data type. (struct scm_t_port): Refactor to use port buffers instead of implementation-managed read and write pointers. Add "read_buffering" member. (SCM_INITIAL_PUTBACK_BUF_SIZE, SCM_READ_BUFFER_EMPTY_P): Remove. (scm_t_ptob_descriptor): Rename "fill_input" function to "read", and take a port buffer, returning void. Likewise "write" takes a port buffer and returns void. Remove "end_input"; instead if there is buffered input and rw_random is true, then there must be a seek function, so just seek back if needed. Remove "flush"; instead all calls to the "write" function implicitly include a "flush", since the buffering happens in the generic port code now. Remove "setvbuf", but add "get_natural_buffer_sizes"; instead the generic port code can buffer any port. (scm_make_port_type): Adapt to read and write prototype changes. (scm_set_port_flush, scm_set_port_end_input, scm_set_port_setvbuf): Remove. (scm_slow_get_byte_or_eof_unlocked) (scm_slow_get_peek_or_eof_unlocked): Remove; the slow path is to call scm_fill_input. (scm_set_port_get_natural_buffer_sizes): New function. (scm_c_make_port_buffer): New internal function. (scm_port_non_buffer): Remove. This was a function for implementations that is no longer needed. Instead open with BUF0 or use (setvbuf port 'none). (scm_fill_input, scm_fill_input_unlocked): Return the filled port buffer. (scm_get_byte_or_eof_unlocked, scm_peek_byte_or_eof_unlocked): Adapt to changes in buffering and EOF management. * libguile/ports.c: Adapt to port interface changes. (initialize_port_buffers): New function, using the port mode flags to set up appropriate initial buffering for all ports. (scm_c_make_port_with_encoding): Create port buffers here instead of delegating to implementations. (scm_close_port): Flush the port if needed instead of delegating to the implementation. * libguile/filesys.c (set_element): Adapt to buffering changes. * libguile/fports.c (fport_get_natural_buffer_sizes): New function, replacing scm_fport_buffer_add. (fport_write, fport_read): Update to let the generic ports code do the buffering. (fport_flush, fport_end_input): Remove. (fport_close): Don't flush in a dynwind; that's the core ports' job. (scm_make_fptob): Adapt. * libguile/ioext.c (scm_redirect_port): Adapt to buffering changes. * libguile/poll.c (scm_primitive_poll): Adapt to buffering changes. * libguile/ports-internal.h (struct scm_port_internal): Remove pending_eof flag; this is now set on the read buffer. * libguile/r6rs-ports.c (struct bytevector_input_port): New type. The new buffering arrangement means that there's now an intermediate buffer between the bytevector and the user of the port; this could lead to a perf degradation, but on the other hand there are some other speedups enabled by the buffering refactor, so probably the memcpy cost is dwarfed by the cost of the other parts of the ports machinery. (make_bytevector_input_port, bytevector_input_port_read): (bytevector_input_port_seek, initialize_bytevector_input_ports): Adapt to new buffering arrangement. (struct custom_binary_port): Remove read buffer, as Guile handles that now. (custom_binary_input_port_setvbuf): Remove; now handled by Guile. (make_custom_binary_input_port, custom_binary_input_port_read) (initialize_custom_binary_input_ports): Adapt. (scm_get_bytevector_some): Adapt to new EOF management. (scm_t_bytevector_output_port_buffer): Hold on to the underlying port, so we can flush it if it's open. (make_bytevector_output_port, bytevector_output_port_write): (bytevector_output_port_seek): Adapt. (bytevector_output_port_procedure): Flush the port as appropriate, so that we get all the bytes. (make_custom_binary_output_port, custom_binary_output_port_write): Adapt. (make_transcoded_port): Don't muck with buffering. (transcoded_port_write): Simply forward the write to the underlying port. (transcoded_port_read): Likewise. (transcoded_port_close): No need to flush. (initialize_transcoded_ports): Adapt. * libguile/read.c (scm_i_scan_for_encoding): Adapt to buffering changes. * libguile/rw.c (scm_write_string_partial): Adapt to buffering changes. * libguile/strports.c: Adapt to the fact that we don't manage the buffer. Probably room for speed improvements here... * libguile/vports.c (soft_port_get_natural_buffer_sizes): New function. Adapt the rest of the file for the new buffering regime. * test-suite/tests/r6rs-ports.test ("8.2.10 Output ports"): Custom binary output ports need to be flushed before you can rely on the write! procedure having been called. Add necessary flush-port invocations. ("8.2.6 Input and output ports"): Transcoded ports now have an internal buffer by default. This test checks that the characters are transcoded one at a time, so to do that, call setvbuf on the transcoded port to remove the buffer. * test-suite/tests/web-client.test (run-with-http-transcript): Fix for different flushing regime on soft ports. (The vestigial flush procedure is now called after each write, which is not what the test was expecting.) * test-suite/standalone/test-scm-c-read.c: Update for changes to the C interface for defining port types. * doc/ref/api-io.texi (Ports): Update to discuss buffering in a generic way, and to remove a hand-wavey paragraph describing string ports as "interesting and powerful". (Reading, Writing): Remove placeholder comments. Document `scm_lfwrite'. (Buffering): New section. (File Ports): Link to buffering. (I/O Extensions): Join subnodes into parent and describe new API, including buffering API. * doc/ref/posix.texi (Ports and File Descriptors): Link to buffering. Remove unread-char etc, as they are documented elsewhere. (Pipes, Network Sockets and Communication): Link to buffering.
2016-04-06 09:21:44 +02:00
scm_tc16_void_port = scm_make_port_type ("void", void_port_read,
void_port_write);
2005-03-02 20:42:01 +00:00
scm_i_port_weak_set = scm_c_make_weak_set (31);
cur_inport_fluid = scm_make_fluid ();
cur_outport_fluid = scm_make_fluid ();
cur_errport_fluid = scm_make_fluid ();
cur_warnport_fluid = scm_make_fluid ();
cur_loadport_fluid = scm_make_fluid ();
2005-03-02 20:42:01 +00:00
sym_substitute = scm_from_latin1_symbol ("substitute");
sym_escape = scm_from_latin1_symbol ("escape");
sym_error = scm_from_latin1_symbol ("error");
Add full Unicode capability to ports and the default reader Ports are given two additional properties: a character encoding and a conversion failure strategy. These properties have getters and setters. The new properties are used to convert any locale text to/from the internal representation of strings. If unspecified, ports use a default value. The default value of these properties is held in a fluid. The default character encoding can be modified by calling setlocale. ISO-8859-1 is treated specially. Since it is a native encoding of strings, it can be processed more quickly. Source code is assumed to be ISO-8859-1 unless otherwise specified. The encoding of a source code file can be given as 'coding: XXXXX' in a magic comment at the top of a file. The C functions that deal with encoding often use a null pointer as shorthand for the native Latin-1 encoding, for efficiency's sake. * test-suite/tests/encoding-iso88591.test: new tests * test-suite/tests/encoding-iso88597.test: new tests * test-suite/tests/encoding-utf8.test: new tests * test-suite/tests/encoding-escapes.test: new tests * test-suite/tests/numbers.test: declare 'binary' encoding * test-suite/tests/ports.test: declare 'binary' encoding * test-suite/tests/r6rs-ports.test: declare 'binary' encoding * module/system/base/compile.scm (compile-file): use source-code file's self-declared encoding when compiling files * libguile/strports.c: store string ports in locale encoding (scm_strport_to_locale_u8vector, scm_call_with_output_locale_u8vector) (scm_open_input_locale_u8vector, scm_get_output_locale_u8vector): new functions * libguile/strings.h: new declaration for scm_i_string_contains_char * libguile/strings.c (scm_i_string_contains_char): new function (scm_from_stringn, scm_to_stringn): use NULL for Latin-1 (scm_from_locale_stringn, scm_to_locale_stringn): respect character encoding of input and output ports * libguile/read.h: declaration for scm_scan_for_encoding * libguile/read.c: (read_token): now takes scheme string instead of C string/length (read_complete_token): new function (scm_read_sexp, scm_read_number, scm_read_mixed_case_symbol) (scm_read_number_and_radix, scm_read_quote, scm_read_semicolon_comment) (scm_read_srfi4_vector, scm_read_bytevector, scm_read_guile_bit_vector) (scm_read_scsh_block_comment, scm_read_commented_expression) (scm_read_extended_symbol, scm_read_sharp_extension, scm_read_shart) (scm_read_expression): use scm_t_wchar for char type, use read_complete_token (scm_scan_for_encoding): new function to find a file's character encoding (scm_file_encoding): new function to find a port's character encoding * libguile/rdelim.c: don't unpack strings * libguile/print.h: declaration for modified function scm_i_charprint * libguile/print.c: use locale when printing characters and strings (scm_i_charprint): input parameter is now scm_t_wchar (scm_simple_format): don't unpack strings * libguile/posix.h: new declaration for scm_setbinary. * libguile/posix.c (scm_setlocale): set default and stdio port encodings based on the locale's character encoding (scm_setbinary): new function * libguile/ports.h (scm_t_port): add encoding and failed conversion handler to port type. Declarations for new or modified functions scm_getc, scm_unget_byte, scm_ungetc, scm_i_get_port_encoding, scm_i_set_port_encoding_x, scm_port_encoding, scm_set_port_encoding_x, scm_i_get_conversion_strategy, scm_i_set_conversion_strategy_x, scm_port_conversion_strategy, scm_set_port_conversion_strategy_x. * libguile/ports.c: assign the current ports to zero on startup so we can see if they've been set. (scm_current_input_port, scm_current_output_port, scm_current_error_port): return #f if the port is not yet initialized (scm_new_port_table_entry): set up a new port's encoding and illegal sequence handler based on the thread's current defaults (scm_i_remove_port): free port encoding name when port is removed (scm_i_mode_bits_n): now takes a scheme string instead of a c string and length. All callers changed. (SCM_MBCHAR_BUF_SIZE): new const (scm_getc): new function, since the scm_getc in inline.h is now scm_get_byte_or_eof. This pulls one codepoint from a port. (scm_lfwrite_substr, scm_lfwrite_str): now uses port's encoding (scm_unget_byte): new function, incorportaing the low-level functionality of scm_ungetc (scm_ungetc): uses scm_unget_byte * libguile/numbers.h (scm_t_wchar): compilation order problem with scm_t_wchar being use in functions in multiple headers. Forward declare scm_t_wchar. * libguile/load.c (scm_primitive_load): scan for file encoding at top of file and use it to set the load port's encoding * libguile/inline.h (scm_get_byte_or_eof): new function incorporating most of the functionality of scm_getc. * libguile/fports.c (fport_fill_input): now returns scm_t_wchar * libguile/chars.h (scm_t_wchar): avoid compilation order problem with declaration of scm_t_wchar
2009-08-25 07:54:37 -07:00
/* Use Latin-1 as the default port encoding. */
default_port_encoding_var =
scm_c_define ("%default-port-encoding",
scm_make_fluid_with_default (SCM_BOOL_F));
default_conversion_strategy_var =
scm_c_define ("%default-port-conversion-strategy",
scm_make_fluid_with_default (sym_substitute));
scm_c_register_extension ("libguile-" SCM_EFFECTIVE_VERSION,
"scm_init_ice_9_ports",
(scm_t_extension_init_func) scm_init_ice_9_ports,
NULL);
/* The following bindings are used early in boot-9.scm. */
/* Used by `include'. */
scm_c_define_gsubr (s_scm_set_port_encoding_x, 2, 0, 0,
(scm_t_subr) scm_set_port_encoding_x);
scm_c_define_gsubr (s_scm_eof_object_p, 1, 0, 0,
(scm_t_subr) scm_eof_object_p);
/* Used by a number of error/warning-printing routines. */
scm_c_define_gsubr (s_scm_force_output, 0, 1, 0,
(scm_t_subr) scm_force_output);
/* Used by `file-exists?' and related functions if `stat' is
unavailable. */
scm_c_define_gsubr (s_scm_close_port, 1, 0, 0,
(scm_t_subr) scm_close_port);
/* Used by error routines. */
scm_c_define_gsubr (s_scm_current_error_port, 0, 0, 0,
(scm_t_subr) scm_current_error_port);
scm_c_define_gsubr (s_scm_current_warning_port, 0, 0, 0,
(scm_t_subr) scm_current_warning_port);
}
/*
Local Variables:
c-file-style: "gnu"
End:
*/