glibc/locale/programs/charmap.c
Roland McGrath 75cd5204dd Wed May 1 09:10:04 1996 Roland McGrath <roland@delasyd.gnu.ai.mit.edu>
* time/strftime.c: Use canonical autoconf nugget for time.h+sys/time.h
	include.

Mon Apr 29 02:48:26 1996  Ulrich Drepper  <drepper@cygnus.com>

	* ctype/ctype-info.c: (__ctype_width): New variable.
	(__ctype_names): Initialize correctly without offset.

	* locale/C-collate.c, locale/C-ctype.c,
	locale/C-messages.c, locale/C-monetary.c,
	locale/C-numeric.c, locale/C-time.c: Change copyright.

	* locale/C-ctype.c (_nl_C_LC_CTYPE_class32): Correct
	endianess for initialization value.

	* locale/lc-ctype.c (current): Add parameter for offset.
	(__ctype32_b, __ctype_width): Add initialization for these
	variables.

	* locale/programs/charmap.c: Finish support for WIDTH information.
	(new_width): New function.

	* locale/programs/charset.h (width_rule): new data structure.
	(charset_t): Add elements for width information.

	* locale/programs/ld-ctype.c (locale_ctype_t): Add element
	for width information.
	(allocate_arrays): Add new argument for charset.
	(ctype_finish): Make sure all characters named in charset
	width table are known to name table.
	(ctype_output): Correct handling of class and map name
	information and write out width information.
	(find_idx): Prepare for being called with NULL pointer as
	TABLE argument.  This means only allocate name entry.
	(allocate_arrays): Correct handling of array element -1.
	Because EOF == -1 the value of element 127 must *not* be
	mirrored here.
	Fill width information from charset tables.

	* locale/programs/localedef.c (main): Correct loop over all
	categories after change of order from Thu Mar 28 14:22:51 1996.
	Add new charset argument to call of `write_all_categories'.

	* locale/programs/locales.h (ctype_finish, ctype_output): New
	charset argument.

	* locale/programs/locfile.c (write_all_categories): Call
	`ctype_output' with additional argument charset.

	* posix/getconf.c (vars): Add _POSIX_SYNC_IO, _POSIX_ASYNC_IO,
	and _POSIX_PRIO_IO definitions.

	* posix/posix2_lim.h: Add definition of _POSIX2_CHARCLASS_NAME_MAX
	and CHARCLASS_NAME_MAX.

	* posix/unistd.h: Document _POSIX_SYNC_IO, _POSIX_ASYNC_IO,
	and _POSIX_PRIO_IO.

	* stdlib/grouping.h: Prepare for use in wide string functions.

	* stdlib/stdlib.h: Correct prototypes for __strto*_internal
	functions.

	* stdlib/strtod.c: Extend for use as `wcsto{f,d,ld}'.

	* stdlib/strtol.c: Extend for use as `wcsto{l,ul,q,uq}'.

	* string/strcoll.c: Extend for use as `wcscoll'.

	* string/strxfrm.c: Extend for use as `wcsxfrm'.

	* sysdeps/generic/confname.h: Add definition of _PC_SYNC_IO,
	_PC_ASYNC_IO, _PC_PRIO_IO and _SC_CHARCLASS_NAME_MAX.

	* sysdeps/generic/stpncpy.c: Correct return value.

	* sysdeps/posix/fpathconf.c: Add handling of _PC_SYNC_IO,
	_PC_ASYNC_IO, and _PC_PRIO_IO.

	* sysdeps/posix/sysconf.c: Add handling of _SC_REALTIME_SIGNALS,
	_SC_PRIORITY_SCHEDULING, _SC_TIMERS, _SC_ASYNCHRONOUS_IO,
	_SC_PRIORITIZED_IO, _SC_SYNCHRONIZED_IO, _SC_FSYNC,
	_SC_MAPPED_FILES, _SC_MEMLOCK, _SC_MEMLOCK_RANGE,
	_SC_MEMORY_PROTECTION, _SC_MESSAGE_PASSING, _SC_SEMAPHORES,
	_SC_SHARED_MEMORY_OBJECTS, and _SC_CHARCLASS_NAME_MAX.
	* sysdeps/stub/sysconf.c: Ditto.
	* sysdeps/unix/sysv/sysv4/sysconf.c: Ditto.

	* sysdeps/unix/sysv/linux/Dist: Add sys/sysctl.h.

	* sysdeps/unix/sysv/linux/Makefile [subdir == misc]
	(sysdep_routines): Add s_sysctl and sysctl.

	* sysdeps/unix/sysv/linux/sys/mman.h: Add declaration of mremap.

	* sysdeps/unix/sysv/linux/sys/socket.h: New file.  Wrapper
	around kernel header.

	* sysdeps/unix/sysv/linux/sys/sysctl.h: New file.  Define
	interface to `sysctl' function.

	* sysdeps/unix/sysv/linux/syscalls.list: Add mremap and _sysctl.

	* sysdeps/unix/sysv/linux/sysconf.c: Add handling of
	_SC_CHARCLASS_NAME_MAX.

	* sysdeps/unix/sysv/linux/sysctl.c: new file.  Implement caller
	of _sysctl system call.

	* sysvipc/Makefile (routines): Add ftok.

	* sysvipc/ftok.c: use variable `proj_id' not `id'.  Patch by
	David Mosberger-Tang.

	* wcsmbs/Makefile (routines): Add wcpcpy, wcpncpy, wcstol,
        wcstoul, wcstoq, wcstouq, wcstod, wcstold, wcstof, wcscoll,
        wcsxfrm, wcwidth, and wcswidth.

	* wcsmbs/wchar.h: Add declarations for wcpcpy, wcpncpy, wcstol,
        wcstoul, wcstoq, wcstouq, wcstod, wcstold, wcstof, wcscoll,
        wcsxfrm, wcwidth, and wcswidth.
	Declare internal interfaces for wcsto* functions.
	[OPTIMIZE]: Define inline functions for wcsto* functions to
	call internal interface functions.

	* wcsmbs/wcpcpy.c, wcsmbs/wcpncpy.c: New files.  Implement non-
	standard function equivalent to stpcpy/stpncpy.

	* wcsmbs/wcscoll.c: Implement `wcscoll' function by using
        `strcoll' implementation.

	* wcsmbs/wcscpy.c, wcsmbs/wcsncpy.c: Use wint_t instead of
        wchar_t.

	* wcsmbs/wcstod.c: Implement `wcstod' function by using `strtod'
        implementation.
	* wcsmbs/wcstof.c: Same for `wcstof'.
	* wcsmbs/wcstold.c: Same for `strtold'.

	* wcsmbs/wcstol.c: Implement `wcstol' function by using `strtol'
        implementation.
	* wcsmbs/wcstoq.c: Same for `wcstoq'.
	* wcsmbs/wcstoul.c: Same for `wcstoul'.
	* wcsmbs/wcstouq.c: Same for `wcstouq'.

	* wcsmbs/wcswidth.c: Implement `wcswidth' function from X/Open
        CAE.
	* wcsmbs/wcwidth.c: Ditto for `wcwidth'.
	* wcsmbs/wcwidth.h: Common function for definitions of above two
	functions.

	* wcsmbs/wcsxfrm.c: Implement `wcsxfrm function by using
        `strxfrm implementation.

	* wctype/wctype.c: Remove case for `wctype_t' being 16 bit type.

	* wctype/wctype.h (wint_t): Protect against multiple definition.
	(wctype_t): Always define as `unsigned long int'.

	* wctype.h: New file.  Wrapper around wctype/wctype.h.

	* hurd/hurdstartup.c (_hurd_split_args): Function removed.
	(_hurd_startup): Use argz functions.

	* hurd/hurdexec.c: Use argz functions.
1996-05-01 13:55:18 +00:00

653 lines
15 KiB
C

/* Copyright (C) 1996 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Ulrich Drepper, <drepper@gnu.ai.mit.edu>.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Library General Public License as
published by the Free Software Foundation; either version 2 of the
License, or (at your option) any later version.
The GNU C 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
Library General Public License for more details.
You should have received a copy of the GNU Library General Public
License along with the GNU C Library; see the file COPYING.LIB. If
not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA. */
#ifdef HAVE_CONFIG_H
# include <config.h>
#endif
#include <ctype.h>
#include <errno.h>
#include <libintl.h>
#include <obstack.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "error.h"
#include "linereader.h"
#include "charset.h"
/* Uncomment following line for production version. */
/* define NDEBUG 1 */
#include <assert.h>
/* Define the lookup function. */
#include "charmap-kw.h"
void *xmalloc (size_t __n);
/* Prototypes for local functions. */
static struct charset_t *parse_charmap (const char *filename);
static void new_width (struct linereader *cmfile, struct charset_t *result,
const char *from, const char *to,
unsigned long int width);
struct charset_t *
charmap_read (const char *filename)
{
const char *pathnfile;
struct charset_t *result = NULL;
if (filename != NULL)
{
if (euidaccess (filename, R_OK) >= 0)
pathnfile = filename;
else
{
char *cp = xmalloc (strlen (filename) + sizeof CHARMAP_PATH + 1);
stpcpy (stpcpy (stpcpy (cp, CHARMAP_PATH), "/"), filename);
pathnfile = (const char *) cp;
}
result = parse_charmap (pathnfile);
if (result == NULL)
error (0, errno, _("character map file `%s' not found"), filename);
}
if (result == NULL)
{
pathnfile = CHARMAP_PATH "/" DEFAULT_CHARMAP;
result = parse_charmap (pathnfile);
if (result == NULL)
error (4, errno, _("default character map file `%s' not found"),
DEFAULT_CHARMAP);
}
return result;
}
static struct charset_t *
parse_charmap (const char *filename)
{
struct linereader *cmfile;
struct charset_t *result;
int state;
enum token_t expected_tok = tok_error;
const char *expected_str = NULL;
char *from_name = NULL;
char *to_name = NULL;
/* Determine path. */
cmfile = lr_open (filename, charmap_hash);
if (cmfile == NULL)
{
if (strchr (filename, '/') == NULL)
{
/* Look in the systems charmap directory. */
char *buf = xmalloc (strlen (filename) + 1 + sizeof (CHARMAP_PATH));
stpcpy (stpcpy (stpcpy (buf, CHARMAP_PATH), "/"), filename);
cmfile = lr_open (buf, charmap_hash);
if (cmfile == NULL)
free (buf);
}
if (cmfile == NULL)
return NULL;
}
/* Allocate room for result. */
result = (struct charset_t *) xmalloc (sizeof (struct charset_t));
memset (result, '\0', sizeof (struct charset_t));
/* The default DEFAULT_WIDTH is 1. */
result->width_default = 1;
#define obstack_chunk_alloc xmalloc
#define obstack_chunk_free free
obstack_init (&result->mem_pool);
if (init_hash (&result->char_table, 256))
{
free (result);
return NULL;
}
/* We use a state machine to describe the charmap description file
format. */
state = 1;
while (1)
{
/* What's on? */
struct token *now = lr_token (cmfile, NULL);
enum token_t nowtok = now->tok;
struct token *arg;
if (nowtok == tok_eof)
break;
switch (state)
{
case 1:
/* The beginning. We expect the special declarations, EOL or
`CHARMAP'. */
if (nowtok == tok_eol)
/* Ignore empty lines. */
continue;
if (nowtok == tok_charmap)
{
from_name = NULL;
to_name = NULL;
/* We have to set up the real work. Fill in some
default values. */
if (result->mb_cur_max == 0)
result->mb_cur_max = 1;
if (result->mb_cur_min == 0)
result->mb_cur_min = result->mb_cur_max;
if (result->mb_cur_min > result->mb_cur_max)
{
error (0, 0, _("\
%s: <mb_cur_max> must be greater than <mb_cur_min>\n"),
cmfile->fname);
result->mb_cur_min = result->mb_cur_max;
}
lr_ignore_rest (cmfile, 1);
state = 2;
continue;
}
if (nowtok != tok_code_set_name && nowtok != tok_mb_cur_max
&& nowtok != tok_mb_cur_min && nowtok != tok_escape_char
&& nowtok != tok_comment_char && nowtok != tok_g0esc
&& nowtok != tok_g1esc && nowtok != tok_g2esc
&& nowtok != tok_g3esc)
{
lr_error (cmfile, _("syntax error in prolog: %s"),
_("illegal definition"));
lr_ignore_rest (cmfile, 0);
continue;
}
/* We know that we need an argument. */
arg = lr_token (cmfile, NULL);
switch (nowtok)
{
case tok_code_set_name:
if (arg->tok != tok_ident)
{
badarg:
lr_error (cmfile, _("syntax error in prolog: %s"),
_("bad argument"));
lr_ignore_rest (cmfile, 0);
continue;
}
result->code_set_name = obstack_copy0 (&result->mem_pool,
arg->val.str.start,
arg->val.str.len);
lr_ignore_rest (cmfile, 1);
continue;
case tok_mb_cur_max:
case tok_mb_cur_min:
if (arg->tok != tok_number)
goto badarg;
if (arg->val.num < 1 || arg->val.num > 4)
{
lr_error (cmfile,
_("value for <%s> must lie between 1 and 4"),
nowtok == tok_mb_cur_min ? "mb_cur_min"
: "mb_cur_max");
lr_ignore_rest (cmfile, 0);
continue;
}
if ((nowtok == tok_mb_cur_max && result->mb_cur_min != 0
&& arg->val.num < result->mb_cur_min)
|| (nowtok == tok_mb_cur_min && result->mb_cur_max != 0
&& arg->val.num > result->mb_cur_max))
{
lr_error (cmfile, _("\
value of <mb_cur_max> must be greater than the value of <mb_cur_min>"));
lr_ignore_rest (cmfile, 0);
continue;
}
if (nowtok == tok_mb_cur_max)
result->mb_cur_max = arg->val.num;
else
result->mb_cur_min = arg->val.num;
lr_ignore_rest (cmfile, 1);
continue;
case tok_escape_char:
case tok_comment_char:
if (arg->tok != tok_ident)
goto badarg;
if (arg->val.str.len != 1)
{
lr_error (cmfile, _("\
argument to <%s> must be a single character"),
nowtok == tok_escape_char ? "escape_char"
: "comment_char");
lr_ignore_rest (cmfile, 0);
continue;
}
if (nowtok == tok_escape_char)
cmfile->escape_char = *arg->val.str.start;
else
cmfile->comment_char = *arg->val.str.start;
lr_ignore_rest (cmfile, 1);
continue;
case tok_g0esc:
case tok_g1esc:
case tok_g2esc:
case tok_g3esc:
lr_ignore_rest (cmfile, 0); /* XXX */
continue;
default:
/* Cannot happen. */
assert (! "Should not happen");
}
break;
case 2:
/* We have seen `CHARMAP' and now are in the body. Each line
must have the format "%s %s %s\n" or "%s...%s %s %s\n". */
if (nowtok == tok_eol)
/* Ignore empty lines. */
continue;
if (nowtok == tok_end)
{
expected_tok = tok_charmap;
expected_str = "CHARMAP";
state = 90;
continue;
}
if (nowtok != tok_bsymbol)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"CHARMAP", _("no symbolic name given"));
lr_ignore_rest (cmfile, 0);
continue;
}
/* If the previous line was not completely correct free the
used memory. */
if (from_name != NULL)
obstack_free (&result->mem_pool, from_name);
from_name = (char *) obstack_copy0 (&result->mem_pool,
now->val.str.start,
now->val.str.len);
to_name = NULL;
state = 3;
continue;
case 3:
/* We have two possibilities: We can see an ellipsis or an
encoding value. */
if (nowtok == tok_ellipsis)
{
state = 4;
continue;
}
/* FALLTHROUGH */
case 5:
if (nowtok != tok_charcode && nowtok != tok_ucs2
&& nowtok != tok_ucs4)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"CHARMAP", _("illegal encoding given"));
lr_ignore_rest (cmfile, 0);
state = 2;
continue;
}
if (nowtok == tok_charcode)
/* Write char value in table. */
charset_new_char (cmfile, result, now->val.charcode.nbytes,
now->val.charcode.val, from_name, to_name);
else
/* Determine ISO 10646 value and write into table. */
charset_new_unicode (cmfile, result, now->val.charcode.nbytes,
now->val.charcode.val, from_name, to_name);
/* Ignore trailing comment silently. */
lr_ignore_rest (cmfile, 0);
from_name = NULL;
to_name = NULL;
state = 2;
continue;
case 4:
if (nowtok != tok_bsymbol)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"CHARMAP",
_("no symbolic name given for end of range"));
lr_ignore_rest (cmfile, 0);
continue;
}
/* If the previous line was not completely correct free the
used memory. */
to_name = (char *) obstack_copy0 (&result->mem_pool,
cmfile->token.val.str.start,
cmfile->token.val.str.len);
state = 3;
continue;
case 90:
if (nowtok != expected_tok)
lr_error (cmfile, _("\
`%1$s' definition does not end with `END %1$s'"), expected_str);
lr_ignore_rest (cmfile, nowtok == expected_tok);
state = 91;
continue;
case 91:
/* Waiting for WIDTH... */
if (nowtok == tok_eol)
/* Ignore empty lines. */
continue;
if (nowtok == tok_width_default)
{
state = 92;
continue;
}
if (nowtok == tok_width)
{
lr_ignore_rest (cmfile, 1);
state = 93;
continue;
}
if (nowtok == tok_width_variable)
{
lr_ignore_rest (cmfile, 1);
state = 98;
continue;
}
lr_error (cmfile, _("\
only WIDTH definitions are allowed to follow the CHARMAP definition"));
lr_ignore_rest (cmfile, 0);
continue;
case 92:
if (nowtok != tok_number)
lr_error (cmfile, _("value for %s must be an integer"),
"WIDTH_DEFAULT");
else
result->width_default = now->val.num;
lr_ignore_rest (cmfile, nowtok == tok_number);
state = 91;
continue;
case 93:
/* We now expect `END WIDTH' or lines of the format "%s %d\n" or
"%s...%s %d\n". */
if (nowtok == tok_eol)
/* ignore empty lines. */
continue;
if (nowtok == tok_end)
{
expected_tok = tok_width;
expected_str = "WIDTH";
state = 90;
continue;
}
if (nowtok != tok_bsymbol)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"WIDTH", _("no symbolic name given"));
lr_ignore_rest (cmfile, 0);
continue;
}
if (from_name != NULL)
obstack_free (&result->mem_pool, from_name);
from_name = (char *) obstack_copy0 (&result->mem_pool,
now->val.str.start,
now->val.str.len);
to_name = NULL;
state = 94;
continue;
case 94:
if (nowtok == tok_ellipsis)
{
state = 95;
continue;
}
case 96:
if (nowtok != tok_number)
lr_error (cmfile, _("value for %s must be an integer"),
"WIDTH");
else
{
/* Store width for chars. */
new_width (cmfile, result, from_name, to_name, now->val.num);
from_name = NULL;
to_name = NULL;
}
lr_ignore_rest (cmfile, nowtok == tok_number);
state = 93;
continue;
case 95:
if (nowtok != tok_bsymbol)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"WIDTH", _("no symbolic name given for end of range"));
lr_ignore_rest (cmfile, 0);
state = 93;
continue;
}
to_name = (char *) obstack_copy0 (&result->mem_pool,
now->val.str.start,
now->val.str.len);
state = 96;
continue;
case 98:
/* We now expect `END WIDTH_VARIABLE' or lines of the format
"%s\n" or "%s...%s\n". */
if (nowtok == tok_eol)
/* ignore empty lines. */
continue;
if (nowtok == tok_end)
{
expected_tok = tok_width_variable;
expected_str = "WIDTH_VARIABLE";
state = 90;
continue;
}
if (nowtok != tok_bsymbol)
{
lr_error (cmfile, _("syntax error in %s definition: %s"),
"WIDTH_VARIABLE", _("no symbolic name given"));
lr_ignore_rest (cmfile, 0);
continue;
}
if (from_name != NULL)
obstack_free (&result->mem_pool, from_name);
from_name = (char *) obstack_copy0 (&result->mem_pool,
now->val.str.start,
now->val.str.len);
to_name = NULL;
state = 99;
continue;
case 99:
if (nowtok == tok_ellipsis)
state = 100;
/* Store info. */
from_name = NULL;
/* Warn */
state = 98;
continue;
case 100:
if (nowtok != tok_bsymbol)
lr_error (cmfile, _("syntax error in %s definition: %s"),
"WIDTH_VARIABLE",
_("no symbolic name given for end of range"));
else
{
to_name = (char *) obstack_copy0 (&result->mem_pool,
now->val.str.start,
now->val.str.len);
/* XXX Enter value into table. */
}
lr_ignore_rest (cmfile, nowtok == tok_bsymbol);
state = 98;
continue;
default:
error (5, 0, _("%s: error in state machine"), __FILE__);
/* NOTREACHED */
}
break;
}
if (state != 91)
error (0, 0, _("%s: premature end of file"), cmfile->fname);
lr_close (cmfile);
return result;
}
static void
new_width (struct linereader *cmfile, struct charset_t *result,
const char *from, const char *to, unsigned long int width)
{
unsigned int from_val, to_val;
from_val = charset_find_value (result, from, strlen (from));
if (from_val == ILLEGAL_CHAR_VALUE)
{
lr_error (cmfile, _("unknown character `%s'"), from);
return;
}
if (to == NULL)
to_val = from_val;
else
{
to_val = charset_find_value (result, to, strlen (to));
if (to_val == ILLEGAL_CHAR_VALUE)
{
lr_error (cmfile, _("unknown character `%s'"), to);
return;
}
}
if (result->nwidth_rules >= result->nwidth_rules_max)
{
size_t new_size = result->nwidth_rules + 32;
struct width_rule *new_rules =
(struct width_rule *) obstack_alloc (&result->mem_pool,
(new_size
* sizeof (struct width_rule)));
memcpy (new_rules, result->width_rules,
result->nwidth_rules_max * sizeof (struct width_rule));
result->width_rules = new_rules;
result->nwidth_rules_max = new_size;
}
result->width_rules[result->nwidth_rules].from = from_val;
result->width_rules[result->nwidth_rules].to = to_val;
result->width_rules[result->nwidth_rules].width = (unsigned int) width;
++result->nwidth_rules;
}