glibc/string/strxfrm.c
Ulrich Drepper 4c7d276eb1 Update.
2000-08-31  Bruno Haible  <haible@clisp.cons.org>

	* wctype/wcfuncs.c (iswblank): New function.
	* wctype/wcfuncs_l.c (__iswblank_l): New function.
	* wctype/wcextra.c: Remove file.
	* wctype/wcextra_l.c: Remove file.
	* wctype/Makefile (distribute): Remove wcextra and wcextra_l.

2000-08-31  Bruno Haible  <haible@clisp.cons.org>

	* locale/programs/ld-collate.c (collate_output): Remove redundant
	assert call.
	* string/strcoll.c: Likewise.
	* string/strxfrm.c: Include assert.h.  Add assert calls like in
	string/strcoll.c.

2000-08-31  Bruno Haible  <haible@clisp.cons.org>

	* locale/categories.def (_NL_COLLATE_HASH_SIZE,
	_NL_COLLATE_HASH_LAYERS, _NL_COLLATE_NAMES, _NL_CTYPE_NAMES,
	_NL_CTYPE_HASH_SIZE, _NL_CTYPE_HASH_LAYERS): Remove definitions.
	* locale/langinfo.h (_NL_COLLATE_HASH_SIZE, _NL_COLLATE_HASH_LAYERS,
	_NL_COLLATE_NAMES): Rename to _NL_COLLATE_GAP1/2/3 respectively.
	(_NL_CTYPE_NAMES): Rename to _NL_CTYPE_GAP3.
	(_NL_CTYPE_GAP3): Rename to _NL_CTYPE_GAP4.
	(_NL_CTYPE_HASH_SIZE, _NL_CTYPE_HASH_LAYERS): Rename to
	_NL_CTYPE_GAP5/6 respectively.

	* locale/C-ctype.c (_nl_C_LC_CTYPE): Update.
	* ctype/ctype-info.c (__ctype_names, __ctype_width): Remove variables.
	* locale/lc-ctype.c (_nl_postload_ctype): Assume new locale format.
	Don't initialize __ctype_names and __ctype_width.
	* wctype/cname-lookup.h: Remove file.
	* wctype/Makefile (distribute): Remove cname-lookup.h.
	* wctype/wchar-lookup.h: Include stdint.h.
	* wctype/wctype.c (__wctype): Assume new locale format.
	* wctype/wctype_l.c (__wctype_l): Likewise.
	* wctype/iswctype.c: Don't include cname-lookup.h.
	(__iswctype): Assume new locale format.
	* wctype/iswctype_l.c: Don't include cname-lookup.h.
	(__iswctype_l): Assume new locale format.
	* wctype/wctrans.c: Don't include ctype.h.
	(wctrans): Assume new locale format.
	* wctype/wctrans_l.c (__wctrans_l): Likewise.
	* wctype/towctrans.c: Don't include cname-lookup.h.
	(__towctrans): Assume new locale format.
	* wctype/towctrans_l.c: Don't include cname-lookup.h.
	(__towctrans_l): Assume new locale format.
	* wctype/wcfuncs.c: Don't include ctype.h and cname-lookup.h. Include
	localeinfo.h instead.
	(__NO_WCTYPE): Remove unused macro.
	(__ctype32_b, __ctype32_toupper, __ctype32_tolower): Remove
	declarations.
	(iswalnum, iswalpha, iswcntrl, iswdigit, iswlower, iswgraph, iswprint,
	iswpunct, iswspace, iswupper, iswxdigit): Assume new locale format.
	(towlower, towupper): Likewise.
	* wctype/wcfuncs_l.c: Don't include cname-lookup.h. Include
	localeinfo.h instead.
	(__NO_WCTYPE): Remove unused macro.
	(__iswalnum_l, __iswalpha_l, __iswcntrl_l, __iswdigit_l, __iswlower_l,
	__iswgraph_l, __iswprint_l, __iswpunct_l, __iswspace_l, __iswupper_l,
	__iswxdigit_l): Assume new locale format.
	(__towlower_l, __towupper_l): Likewise.
	* wcsmbs/wcwidth.h: Don't include cname-lookup.h.
	(__ctype32_b): Remove declaration.
	(internal_wcwidth): Assume new locale format.
	* locale/programs/ld-ctype.c (struct locale_ctype_t): Remove fields
	plane_size, plane_cnt, names, width. Rename map to map_b, rename map32
	to map32_b, rename width_3level to width.
	(ctype_output): Always create new locale format. Don't emit
	_NL_CTYPE_NAMES, _NL_CTYPE_HASH_SIZE, _NL_CTYPE_HASH_LAYERS any more.
	(allocate_arrays): Always create new locale format.

	* locale/C-collate.c (_nl_C_LC_COLLATE): Update.
	* locale/weightwc.h (findidx): Assume new locale format.
	* string/strcoll.c (wcscoll): Remove local variables size, layers,
	names.
	* string/strxfrm.c (wcsxfrm): Likewise.
	* posix/fnmatch_loop.c (internal_fnwmatch): Likewise. Change type of
	local variable collseq to 'const char *'.
	(SUFFIX): Don't use, don't undefine.
	* posix/fnmatch.c: Remove SUFFIX definition.
	* locale/programs/ld-collate.c (struct locale_collate_t): Remove
	fields plane_size, plane_cnt, wcheads, wcseqorder. Rename
	wcheads_3level to wcheads, rename wcseqorder_3level to wcseqorder.
	(collate_finish): Always create new locale format. Remove local
	variables min_total, act_size.
	(collate_output): Always create new locale format. Remove local
	variables table_size, names, tablewc. Rename tablewc_3level to
	tablewc.

2000-08-31  Bruno Haible  <haible@clisp.cons.org>

	* locale/programs/ld-collate.c (obstack_int32_grow,
	obstack_int32_grow_fast): New inline functions.
	(output_weightwc, collate_output): Use them where possible.
2000-09-01 19:36:44 +00:00

463 lines
12 KiB
C

/* Copyright (C) 1995-1999, 2000 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Written by Ulrich Drepper <drepper@cygnus.com>, 1995.
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. */
#include <assert.h>
#include <langinfo.h>
#include <stddef.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <sys/param.h>
#ifndef STRING_TYPE
# define STRING_TYPE char
# define USTRING_TYPE unsigned char
# ifdef USE_IN_EXTENDED_LOCALE_MODEL
# define STRXFRM __strxfrm_l
# else
# define STRXFRM strxfrm
# endif
# define STRCMP strcmp
# define STRLEN strlen
# define STPNCPY __stpncpy
# define WEIGHT_H "../locale/weight.h"
# define SUFFIX MB
# define L(arg) arg
#endif
#define CONCAT(a,b) CONCAT1(a,b)
#define CONCAT1(a,b) a##b
#include "../locale/localeinfo.h"
#ifndef WIDE_CHAR_VERSION
/* We need UTF-8 encoding of numbers. */
static inline int
utf8_encode (char *buf, int val)
{
int retval;
if (val < 0x80)
{
*buf++ = (char) val;
retval = 1;
}
else
{
int step;
for (step = 2; step < 6; ++step)
if ((val & (~(uint32_t)0 << (5 * step + 1))) == 0)
break;
retval = step;
*buf = (unsigned char) (~0xff >> step);
--step;
do
{
buf[step] = 0x80 | (val & 0x3f);
val >>= 6;
}
while (--step > 0);
*buf |= val;
}
return retval;
}
#endif
#ifndef USE_IN_EXTENDED_LOCALE_MODEL
size_t
STRXFRM (STRING_TYPE *dest, const STRING_TYPE *src, size_t n)
#else
size_t
STRXFRM (STRING_TYPE *dest, const STRING_TYPE *src, size_t n, __locale_t l)
#endif
{
#ifdef USE_IN_EXTENDED_LOCALE_MODEL
struct locale_data *current = l->__locales[LC_COLLATE];
uint_fast32_t nrules = *((uint32_t *) current->values[_NL_ITEM_INDEX (_NL_COLLATE_NRULES)].string);
#else
uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
#endif
/* We don't assign the following values right away since it might be
unnecessary in case there are no rules. */
const unsigned char *rulesets;
const int32_t *table;
const USTRING_TYPE *weights;
const USTRING_TYPE *extra;
const int32_t *indirect;
uint_fast32_t pass;
size_t needed;
const USTRING_TYPE *usrc;
size_t srclen = STRLEN (src);
int32_t *idxarr;
unsigned char *rulearr;
size_t idxmax;
size_t idxcnt;
int use_malloc;
#include WEIGHT_H
if (nrules == 0)
{
if (n != 0)
STPNCPY (dest, src, MIN (srclen + 1, n));
return srclen;
}
#ifdef USE_IN_EXTENDED_LOCALE_MODEL
rulesets = (const unsigned char *)
current->values[_NL_ITEM_INDEX (_NL_COLLATE_RULESETS)].string;
table = (const int32_t *)
current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_TABLE,SUFFIX))].string;
weights = (const USTRING_TYPE *)
current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_WEIGHT,SUFFIX))].string;
extra = (const USTRING_TYPE *)
current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_EXTRA,SUFFIX))].string;
indirect = (const int32_t *)
current->values[_NL_ITEM_INDEX (CONCAT(_NL_COLLATE_INDIRECT,SUFFIX))].string;
#else
rulesets = (const unsigned char *)
_NL_CURRENT (LC_COLLATE, _NL_COLLATE_RULESETS);
table = (const int32_t *)
_NL_CURRENT (LC_COLLATE, CONCAT(_NL_COLLATE_TABLE,SUFFIX));
weights = (const USTRING_TYPE *)
_NL_CURRENT (LC_COLLATE, CONCAT(_NL_COLLATE_WEIGHT,SUFFIX));
extra = (const USTRING_TYPE *)
_NL_CURRENT (LC_COLLATE, CONCAT(_NL_COLLATE_EXTRA,SUFFIX));
indirect = (const int32_t *)
_NL_CURRENT (LC_COLLATE, CONCAT(_NL_COLLATE_INDIRECT,SUFFIX));
#endif
use_malloc = 0;
assert (((uintptr_t) table) % sizeof (table[0]) == 0);
assert (((uintptr_t) weights) % sizeof (weights[0]) == 0);
assert (((uintptr_t) extra) % sizeof (extra[0]) == 0);
assert (((uintptr_t) indirect) % sizeof (indirect[0]) == 0);
/* Handle an empty string as a special case. */
if (srclen == 0)
{
if (n != 0)
*dest = L('\0');
return 1;
}
/* We need the elements of the string as unsigned values since they
are used as indeces. */
usrc = (const USTRING_TYPE *) src;
/* Perform the first pass over the string and while doing this find
and store the weights for each character. Since we want this to
be as fast as possible we are using `alloca' to store the temporary
values. But since there is no limit on the length of the string
we have to use `malloc' if the string is too long. We should be
very conservative here. */
if (srclen >= 16384)
{
idxarr = (int32_t *) malloc (srclen * (sizeof (int32_t) + 1));
rulearr = (unsigned char *) &idxarr[srclen];
if (idxarr == NULL)
/* No memory. Well, go with the stack then.
XXX Once this implementation is stable we will handle this
differently. Instead of precomputing the indeces we will
do this in time. This means, though, that this happens for
every pass again. */
goto try_stack;
use_malloc = 1;
}
else
{
try_stack:
idxarr = (int32_t *) alloca (srclen * sizeof (int32_t));
rulearr = (unsigned char *) alloca (srclen);
}
idxmax = 0;
do
{
int32_t tmp = findidx (&usrc);
rulearr[idxmax] = tmp >> 24;
idxarr[idxmax] = tmp & 0xffffff;
++idxmax;
}
while (*usrc != L('\0'));
/* Now the passes over the weights. We now use the indeces we found
before. */
needed = 0;
for (pass = 0; pass < nrules; ++pass)
{
size_t backw_stop = ~0ul;
int rule = rulesets[rulearr[0] * nrules + pass];
/* We assume that if a rule has defined `position' in one section
this is true for all of them. */
int position = rule & sort_position;
if (position == 0)
{
for (idxcnt = 0; idxcnt < idxmax; ++idxcnt)
{
if ((rule & sort_forward) != 0)
{
size_t len;
if (backw_stop != ~0ul)
{
/* Handle the pushed elements now. */
size_t backw;
for (backw = idxcnt - 1; backw >= backw_stop; --backw)
{
len = weights[idxarr[backw]++];
if (needed + len < n)
while (len-- > 0)
dest[needed++] = weights[idxarr[backw]++];
else
{
/* No more characters fit into the buffer. */
needed += len;
idxarr[backw] += len;
}
}
backw_stop = ~0ul;
}
/* Now handle the forward element. */
len = weights[idxarr[idxcnt]++];
if (needed + len < n)
while (len-- > 0)
dest[needed++] = weights[idxarr[idxcnt]++];
else
{
/* No more characters fit into the buffer. */
needed += len;
idxarr[idxcnt] += len;
}
}
else
{
/* Remember where the backwards series started. */
if (backw_stop == ~0ul)
backw_stop = idxcnt;
}
rule = rulesets[rulearr[idxcnt + 1] * nrules + pass];
}
if (backw_stop != ~0ul)
{
/* Handle the pushed elements now. */
size_t backw;
backw = idxcnt;
while (backw > backw_stop)
{
size_t len = weights[idxarr[--backw]++];
if (needed + len < n)
while (len-- > 0)
dest[needed++] = weights[idxarr[backw]++];
else
{
/* No more characters fit into the buffer. */
needed += len;
idxarr[backw] += len;
}
}
}
}
else
{
int val = 1;
#ifndef WIDE_CHAR_VERSION
char buf[7];
size_t buflen;
#endif
size_t i;
for (idxcnt = 0; idxcnt < idxmax; ++idxcnt)
{
if ((rule & sort_forward) != 0)
{
size_t len;
if (backw_stop != ~0ul)
{
/* Handle the pushed elements now. */
size_t backw;
for (backw = idxcnt - 1; backw >= backw_stop; --backw)
{
len = weights[idxarr[backw]++];
if (len != 0)
{
#ifdef WIDE_CHAR_VERSION
if (needed + 1 + len < n)
{
dest[needed] = val;
for (i = 0; i < len; ++i)
dest[needed + 1 + i] =
weights[idxarr[backw] + i];
}
needed += 1 + len;
#else
buflen = utf8_encode (buf, val);
if (needed + buflen + len < n)
{
for (i = 0; i < buflen; ++i)
dest[needed + i] = buf[i];
for (i = 0; i < len; ++i)
dest[needed + buflen + i] =
weights[idxarr[backw] + i];
}
needed += buflen + len;
#endif
idxarr[backw] += len;
val = 1;
}
else
++val;
}
backw_stop = ~0ul;
}
/* Now handle the forward element. */
len = weights[idxarr[idxcnt]++];
if (len != 0)
{
#ifdef WIDE_CHAR_VERSION
if (needed + 1+ len < n)
{
dest[needed] = val;
for (i = 0; i < len; ++i)
dest[needed + 1 + i] =
weights[idxarr[idxcnt] + i];
}
needed += 1 + len;
#else
buflen = utf8_encode (buf, val);
if (needed + buflen + len < n)
{
for (i = 0; i < buflen; ++i)
dest[needed + i] = buf[i];
for (i = 0; i < len; ++i)
dest[needed + buflen + i] =
weights[idxarr[idxcnt] + i];
}
needed += buflen + len;
#endif
idxarr[idxcnt] += len;
val = 1;
}
else
/* Note that we don't have to increment `idxarr[idxcnt]'
since the length is zero. */
++val;
}
else
{
/* Remember where the backwards series started. */
if (backw_stop == ~0ul)
backw_stop = idxcnt;
}
rule = rulesets[rulearr[idxcnt + 1] * nrules + pass];
}
if (backw_stop != ~0)
{
/* Handle the pushed elements now. */
size_t backw;
backw = idxmax - 1;
while (backw > backw_stop)
{
size_t len = weights[idxarr[--backw]++];
if (len != 0)
{
#ifdef WIDE_CHAR_VERSION
if (needed + 1 + len < n)
{
dest[needed] = val;
for (i = 0; i < len; ++i)
dest[needed + 1 + i] =
weights[idxarr[backw] + i];
}
needed += 1 + len;
#else
buflen = utf8_encode (buf, val);
if (needed + buflen + len < n)
{
for (i = 0; i < buflen; ++i)
dest[needed + i] = buf[i];
for (i = 0; i < len; ++i)
dest[needed + buflen + i] =
weights[idxarr[backw] + i];
}
needed += buflen + len;
#endif
idxarr[backw] += len;
val = 1;
}
else
++val;
}
}
}
/* Finally store the byte to separate the passes or terminate
the string. */
if (needed < n)
dest[needed] = pass + 1 < nrules ? L('\1') : L('\0');
++needed;
}
/* This is a little optimization: many collation specifications have
a `position' rule at the end and if no non-ignored character
is found the last \1 byte is immediately followed by a \0 byte
signalling this. We can avoid the \1 byte(s). */
if (needed <= n && needed > 2 && dest[needed - 2] == L('\1'))
{
/* Remove the \1 byte. */
--needed;
dest[needed - 1] = L('\0');
}
/* Free the memory if needed. */
if (use_malloc)
free (idxarr);
/* Return the number of bytes/words we need, but don't count the NUL
byte/word at the end. */
return needed - 1;
}