string/strcoll.c

   1 /* Copyright (C) 1995, 1996, 1997, 1998 Free Software Foundation, Inc.
   2    This file is part of the GNU C Library.
   3    Written by Ulrich Drepper <drepper@gnu.ai.mit.edu>, 1995.
   4
   5    The GNU C Library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Library General Public License as
   7    published by the Free Software Foundation; either version 2 of the
   8    License, or (at your option) any later version.
   9
  10    The GNU C Library is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Library General Public License for more details.
  14
  15    You should have received a copy of the GNU Library General Public
  16    License along with the GNU C Library; see the file COPYING.LIB.  If not,
  17    write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  18    Boston, MA 02111-1307, USA.  */
  19
  20 #include <endian.h>
  21 #include <stddef.h>
  22 #include <stdint.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25
  26 #ifndef STRING_TYPE
  27 # define STRING_TYPE char
  28 # define USTRING_TYPE unsigned char
  29 # ifdef USE_IN_EXTENDED_LOCALE_MODEL
  30 #  define STRCOLL __strcoll_l
  31 # else
  32 #  define STRCOLL strcoll
  33 # endif
  34 # define STRCMP strcmp
  35 #endif
  36
  37 /* Include the shared helper functions.  `strxfrm'/`wcsxfrm' also use
  38    these functions.  */
  39 #include "../locale/weight.h"
  40
  41
  42 /* Compare S1 and S2, returning less than, equal to or
  43    greater than zero if the collated form of S1 is lexicographically
  44    less than, equal to or greater than the collated form of S2.  */
  45 #ifndef USE_IN_EXTENDED_LOCALE_MODEL
  46 int
  47 STRCOLL (s1, s2)
  48      const STRING_TYPE *s1;
  49      const STRING_TYPE *s2;
  50 #else
  51 int
  52 STRCOLL (s1, s2, l)
  53      const STRING_TYPE *s1;
  54      const STRING_TYPE *s2;
  55      __locale_t l;
  56 #endif
  57 {
  58 #ifdef USE_IN_EXTENDED_LOCALE_MODEL
  59   struct locale_data *current = l->__locales[LC_COLLATE];
  60 # if BYTE_ORDER == BIG_ENDIAN
  61   const uint32_t *collate_table = (const uint32_t *)
  62     current->values[_NL_ITEM_INDEX (_NL_COLLATE_TABLE_EB)].string;
  63   const uint32_t *collate_extra = (const uint32_t *)
  64     current->values[_NL_ITEM_INDEX (_NL_COLLATE_EXTRA_EB)].string;
  65 # elif BYTE_ORDER == LITTLE_ENDIAN
  66   const uint32_t *collate_table = (const uint32_t *)
  67     current->values[_NL_ITEM_INDEX (_NL_COLLATE_TABLE_EL)].string;
  68   const uint32_t *collate_extra = (const uint32_t *)
  69     current->values[_NL_ITEM_INDEX (_NL_COLLATE_EXTRA_EL)].string;
  70 # else
  71 #  error bizarre byte order
  72 # endif
  73 #endif
  74   weight_t *s1forw = NULL;
  75   weight_t *s1backw = NULL;
  76   weight_t *s2forw = NULL;
  77   weight_t *s2backw = NULL;
  78   size_t pass;
  79
  80   /* If the current locale does not specify locale data we use normal
  81      8-bit string comparison.  */
  82   if (collate_nrules == 0)
  83     return STRCMP (s1, s2);
  84
  85   /* Handle empty strings as a special case.  */
  86   if (*s1 == '\0')
  87     return *s2 == '\0' ? 0 : -1;
  88   else if (*s2 == '\0')
  89     return 1;
  90
  91   /* Get full information about the strings.  This means we get
  92      information for all passes in a special data structure.  */
  93   get_string (s1, s1forw, s1backw);
  94   get_string (s2, s2forw, s2backw);
  95
  96   /* Now we have all the information.  In at most the given number of
  97      passes we can finally decide about the order.  */
  98   for (pass = 0; pass < collate_nrules; ++pass)
  99     {
 100       int forward = (collate_rules[pass] & sort_forward) != 0;
 101       const weight_t *s1run = forward ? s1forw : s1backw;
 102       const weight_t *s2run = forward ? s2forw : s2backw;
 103       int s1idx = forward ? 0 : s1run->data[pass].number - 1;
 104       int s2idx = forward ? 0 : s2run->data[pass].number - 1;
 105
 106       while (1)
 107         {
 108           int s1ignore = 0;
 109           int s2ignore = 0;
 110           uint32_t w1 = 0;
 111           uint32_t w2 = 0;
 112
 113           /* Here we have to check for IGNORE entries.  If these are
 114              found we count them and go on with the next value.  */
 115           while (s1run != NULL
 116                  && ((w1 = s1run->data[pass].value[s1idx])
 117                      == (uint32_t) IGNORE_CHAR))
 118             {
 119               ++s1ignore;
 120               if (forward
 121                   ? ++s1idx >= s1run->data[pass].number
 122                   : --s1idx < 0)
 123                 {
 124                   weight_t *nextp = forward ? s1run->next : s1run->prev;
 125                   if (nextp == NULL)
 126                     {
 127                       w1 = 0;
 128                       /* No more non-INGOREd elements means lowest
 129                          possible value.  */
 130                       s1ignore = -1;
 131                     }
 132                   else
 133                     s1idx = forward ? 0 : nextp->data[pass].number - 1;
 134                   s1run = nextp;
 135                 }
 136             }
 137
 138           while (s2run != NULL
 139                  && ((w2 = s2run->data[pass].value[s2idx])
 140                      == (uint32_t) IGNORE_CHAR))
 141             {
 142               ++s2ignore;
 143               if (forward
 144                   ? ++s2idx >= s2run->data[pass].number
 145                   : --s2idx < 0)
 146                 {
 147                   weight_t *nextp = forward ? s2run->next : s2run->prev;
 148                   if (nextp == NULL)
 149                     {
 150                       w2 = 0;
 151                       /* No more non-INGOREd elements means lowest
 152                          possible value.  */
 153                       s2ignore = -1;
 154                     }
 155                   else
 156                     s2idx = forward ? 0 : nextp->data[pass].number - 1;
 157                   s2run = nextp;
 158                 }
 159             }
 160
 161           /* If one string is completely processed stop.  */
 162           if (s1run == NULL || s2run == NULL)
 163             break;
 164
 165           /* Now we have information of the number of ignored
 166              weights and the value of the next weight.  */
 167           if ((collate_rules[pass] & sort_position) != 0
 168               && s1ignore != s2ignore && (w1 != 0 || w2 != 0))
 169             return s1ignore < s2ignore ? -1 : 1;
 170
 171           if (w1 != w2)
 172             return w1 < w2 ? -1 : 1;
 173
 174           /* We have to increment the index counters.  */
 175           if (forward)
 176             {
 177               if (++s1idx >= s1run->data[pass].number)
 178                 {
 179                   s1run = s1run->next;
 180                   s1idx = 0;
 181                 }
 182               if (++s2idx >= s2run->data[pass].number)
 183                 {
 184                   s2run = s2run->next;
 185                   s2idx = 0;
 186                 }
 187             }
 188           else
 189             {
 190               if (--s1idx < 0)
 191                 {
 192                   s1run = s1run->prev;
 193                   if (s1run != NULL)
 194                     s1idx = s1run->data[pass].number - 1;
 195                 }
 196               if (--s2idx < 0)
 197                 {
 198                   s2run = s2run->prev;
 199                   if (s2run != NULL)
 200                     s2idx = s2run->data[pass].number - 1;
 201                 }
 202             }
 203         }
 204
 205       if (s1run != s2run)
 206         return s1run != NULL ? 1 : -1;
 207     }
 208
 209   return 0;
 210 }