-/* Copyright (C) 1996 Free Software Foundation, Inc.
+/* Copyright (C) 1996-2022 Free Software Foundation, Inc.
This file is part of the GNU C Library.
- Written 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.
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 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.
+ Lesser 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. */
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <https://www.gnu.org/licenses/>. */
-#include <alloca.h>
-#include <langinfo.h>
-#include "localeinfo.h"
+#ifndef _WEIGHT_H_
+#define _WEIGHT_H_ 1
-#ifndef STRING_TYPE
-# error STRING_TYPE not defined
-#endif
+#include <libc-diag.h>
-#ifndef USTRING_TYPE
-# error USTRING_TYPE not defined
-#endif
-
-typedef struct weight_t
+/* Find index of weight. */
+static inline int32_t __attribute__ ((always_inline))
+findidx (const int32_t *table,
+ const int32_t *indirect,
+ const unsigned char *extra,
+ const unsigned char **cpp, size_t len)
{
- struct weight_t *prev;
- struct weight_t *next;
- struct data_pair
+ int_fast32_t i = table[*(*cpp)++];
+ const unsigned char *cp;
+ const unsigned char *usrc;
+
+ if (i >= 0)
+ /* This is an index into the weight table. Cool. */
+ return i;
+
+ /* Oh well, more than one sequence starting with this byte.
+ Search for the correct one. */
+ cp = &extra[-i];
+ usrc = *cpp;
+ --len;
+ while (1)
{
- int number;
- const u_int32_t *value;
- } data[0];
-} weight_t;
-
-
-/* The following five macros grant access to the values in the
- collate locale file that do not depend on byte order. */
-#define collate_nrules \
- (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES))
-#define collate_hash_size \
- (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_HASH_SIZE))
-#define collate_hash_layers \
- (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_HASH_LAYERS))
-#define collate_undefined \
- (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_UNDEFINED))
-#define collate_rules \
- (_NL_CURRENT (LC_COLLATE, _NL_COLLATE_RULES))
-
-
-static __inline int get_weight (const STRING_TYPE **str, weight_t *result);
-static __inline int
-get_weight (const STRING_TYPE **str, weight_t *result)
-{
- unsigned int ch = *((USTRING_TYPE *) (*str))++;
- size_t slot;
+ size_t nhere;
- if (sizeof (STRING_TYPE) == 1)
- slot = ch * (collate_nrules + 1);
- else
- {
- const size_t level_size = collate_hash_size * (collate_nrules + 1);
- size_t level;
+ /* The first thing is the index. */
+ i = *((const int32_t *) cp);
+ cp += sizeof (int32_t);
- slot = (ch * (collate_nrules + 1)) % collate_hash_size;
+ /* Next is the length of the byte sequence. These are always
+ short byte sequences so there is no reason to call any
+ function (even if they are inlined). */
+ nhere = *cp++;
- level = 0;
- while (__collate_table[slot] != (u_int32_t) ch)
+ if (i >= 0)
{
- if (__collate_table[slot + 1] == 0
- || ++level >= collate_hash_layers)
- {
- size_t idx = collate_undefined;
- size_t cnt;
+ /* It is a single character. If it matches we found our
+ index. Note that at the end of each list there is an
+ entry of length zero which represents the single byte
+ sequence. The first (and here only) byte was tested
+ already. */
+ size_t cnt;
- for (cnt = 0; cnt < collate_nrules; ++cnt)
- {
- result->data[cnt].number = __collate_extra[idx++];
- result->data[cnt].value = &__collate_extra[idx];
- idx += result->data[cnt].number;
- }
- return 0;
+ /* With GCC 5.3 when compiling with -Os the compiler warns
+ that seq2.back_us, which becomes usrc, might be used
+ uninitialized. This can't be true because we pass a length
+ of -1 for len at the same time which means that this loop
+ never executes. */
+ DIAG_PUSH_NEEDS_COMMENT;
+ DIAG_IGNORE_Os_NEEDS_COMMENT (5, "-Wmaybe-uninitialized");
+ for (cnt = 0; cnt < nhere && cnt < len; ++cnt)
+ if (cp[cnt] != usrc[cnt])
+ break;
+ DIAG_POP_NEEDS_COMMENT;
+
+ if (cnt == nhere)
+ {
+ /* Found it. */
+ *cpp += nhere;
+ return i;
}
- slot += level_size;
- }
- }
-
- if (__collate_table[slot + 1] != (u_int32_t) FORWARD_CHAR)
- {
- /* We have a simple form. One one value for each weight. */
- size_t cnt;
- for (cnt = 0; cnt < collate_nrules; ++cnt)
- {
- result->data[cnt].number = 1;
- result->data[cnt].value = &__collate_table[slot + 1 + cnt];
+ /* Up to the next entry. */
+ cp += nhere;
+ if (!LOCFILE_ALIGNED_P (1 + nhere))
+ cp += LOCFILE_ALIGN - (1 + nhere) % LOCFILE_ALIGN;
}
- return ch == 0;
- }
+ else
+ {
+ /* This is a range of characters. First decide whether the
+ current byte sequence lies in the range. */
+ size_t cnt;
+ size_t offset = 0;
- /* We now look for any collation element which starts with CH.
- There might none, but the last list member is a catch-all case
- because it is simple the character CH. The value of this entry
- might be the same as UNDEFINED. */
- slot = __collate_table[slot + 2];
+ for (cnt = 0; cnt < nhere && cnt < len; ++cnt)
+ if (cp[cnt] != usrc[cnt])
+ break;
- while (1)
- {
- size_t idx;
+ if (cnt != nhere)
+ {
+ if (cnt == len || cp[cnt] > usrc[cnt])
+ {
+ /* Cannot be in this range. */
+ cp += 2 * nhere;
+ if (!LOCFILE_ALIGNED_P (1 + 2 * nhere))
+ cp += (LOCFILE_ALIGN
+ - (1 + 2 * nhere) % LOCFILE_ALIGN);
+ continue;
+ }
- /* This is a comparison between a u_int32_t array (aka wchar_t) and
- an 8-bit string. */
- for (idx = 0; __collate_extra[slot + 2 + idx] != 0; ++idx)
- if (__collate_extra[slot + 2 + idx] != (u_int32_t) str[idx])
- break;
+ /* Test against the end of the range. */
+ for (cnt = 0; cnt < nhere; ++cnt)
+ if (cp[nhere + cnt] != usrc[cnt])
+ break;
- /* When the loop finished with all character of the collation
- element used, we found the longest prefix. */
- if (__collate_extra[slot + 2 + idx] == 0)
- {
- size_t cnt;
+ if (cnt != nhere && cp[nhere + cnt] < usrc[cnt])
+ {
+ /* Cannot be in this range. */
+ cp += 2 * nhere;
+ if (!LOCFILE_ALIGNED_P (1 + 2 * nhere))
+ cp += (LOCFILE_ALIGN
+ - (1 + 2 * nhere) % LOCFILE_ALIGN);
+ continue;
+ }
- idx += slot + 3;
- for (cnt = 0; cnt < collate_nrules; ++cnt)
- {
- result->data[cnt].number = __collate_extra[idx++];
- result->data[cnt].value = &__collate_extra[idx];
- idx += result->data[cnt].number;
+ /* This range matches the next characters. Now find
+ the offset in the indirect table. */
+ for (cnt = 0; cp[cnt] == usrc[cnt]; ++cnt);
+
+ do
+ {
+ offset <<= 8;
+ /* With GCC 7 when compiling with -Os the compiler
+ warns that seq1.back_us and seq2.back_us, which
+ become usrc, might be used uninitialized. This
+ is impossible for the same reason as described
+ above. */
+ DIAG_PUSH_NEEDS_COMMENT;
+ DIAG_IGNORE_Os_NEEDS_COMMENT (7, "-Wmaybe-uninitialized");
+ offset += usrc[cnt] - cp[cnt];
+ DIAG_POP_NEEDS_COMMENT;
+ }
+ while (++cnt < nhere);
}
- return 0;
- }
- /* To next entry in list. */
- slot += __collate_extra[slot];
+ *cpp += nhere;
+ return indirect[-i + offset];
+ }
}
+
/* NOTREACHED */
- return 0; /* To calm down gcc. */
+ return 0x43219876;
}
-
-/* To process a string efficiently we retrieve all information about
- the string at once. The following macro constructs a double linked
- list of this information. It is a macro because we use `alloca'
- and we use a double linked list because of the backward collation
- order. */
-#define get_string(str, forw, backw) \
- do \
- { \
- weight_t *newp; \
- do \
- { \
- newp = (weight_t *) alloca (sizeof (weight_t) \
- + (collate_nrules \
- * sizeof (struct data_pair))); \
- \
- newp->prev = backw; \
- if (backw == NULL) \
- forw = newp; \
- else \
- backw->next = newp; \
- newp->next = NULL; \
- backw = newp; \
- } \
- while (get_weight (&str, newp) == 0); \
- } \
- while (0)
+#endif /* weight.h */