lib/mbscasestr.c

   1 /* Case-insensitive searching in a string.  -*- coding: utf-8 -*-
   2    Copyright (C) 2005-2017 Free Software Foundation, Inc.
   3    Written by Bruno Haible <bruno@clisp.org>, 2005.
   4
   5    This program is free software: you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation; either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program 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
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  17
  18 #include <config.h>
  19
  20 /* Specification.  */
  21 #include <string.h>
  22
  23 #include <ctype.h>
  24 #include <stdbool.h>
  25 #include <stddef.h>  /* for NULL, in case a nonstandard string.h lacks it */
  26
  27 #include "malloca.h"
  28 #include "mbuiter.h"
  29
  30 #define TOLOWER(Ch) (isupper (Ch) ? tolower (Ch) : (Ch))
  31
  32 /* Knuth-Morris-Pratt algorithm.  */
  33 #define UNIT unsigned char
  34 #define CANON_ELEMENT(c) TOLOWER (c)
  35 #include "str-kmp.h"
  36
  37 /* Knuth-Morris-Pratt algorithm.
  38    See http://en.wikipedia.org/wiki/Knuth-Morris-Pratt_algorithm
  39    Return a boolean indicating success:
  40    Return true and set *RESULTP if the search was completed.
  41    Return false if it was aborted because not enough memory was available.  */
  42 static bool
  43 knuth_morris_pratt_multibyte (const char *haystack, const char *needle,
  44                               const char **resultp)
  45 {
  46   size_t m = mbslen (needle);
  47   mbchar_t *needle_mbchars;
  48   size_t *table;
  49
  50   /* Allocate room for needle_mbchars and the table.  */
  51   char *memory = (char *) nmalloca (m, sizeof (mbchar_t) + sizeof (size_t));
  52   if (memory == NULL)
  53     return false;
  54   needle_mbchars = (mbchar_t *) memory;
  55   table = (size_t *) (memory + m * sizeof (mbchar_t));
  56
  57   /* Fill needle_mbchars.  */
  58   {
  59     mbui_iterator_t iter;
  60     size_t j;
  61
  62     j = 0;
  63     for (mbui_init (iter, needle); mbui_avail (iter); mbui_advance (iter), j++)
  64       {
  65         mb_copy (&needle_mbchars[j], &mbui_cur (iter));
  66         if (needle_mbchars[j].wc_valid)
  67           needle_mbchars[j].wc = towlower (needle_mbchars[j].wc);
  68       }
  69   }
  70
  71   /* Fill the table.
  72      For 0 < i < m:
  73        0 < table[i] <= i is defined such that
  74        forall 0 < x < table[i]: needle[x..i-1] != needle[0..i-1-x],
  75        and table[i] is as large as possible with this property.
  76      This implies:
  77      1) For 0 < i < m:
  78           If table[i] < i,
  79           needle[table[i]..i-1] = needle[0..i-1-table[i]].
  80      2) For 0 < i < m:
  81           rhaystack[0..i-1] == needle[0..i-1]
  82           and exists h, i <= h < m: rhaystack[h] != needle[h]
  83           implies
  84           forall 0 <= x < table[i]: rhaystack[x..x+m-1] != needle[0..m-1].
  85      table[0] remains uninitialized.  */
  86   {
  87     size_t i, j;
  88
  89     /* i = 1: Nothing to verify for x = 0.  */
  90     table[1] = 1;
  91     j = 0;
  92
  93     for (i = 2; i < m; i++)
  94       {
  95         /* Here: j = i-1 - table[i-1].
  96            The inequality needle[x..i-1] != needle[0..i-1-x] is known to hold
  97            for x < table[i-1], by induction.
  98            Furthermore, if j>0: needle[i-1-j..i-2] = needle[0..j-1].  */
  99         mbchar_t *b = &needle_mbchars[i - 1];
 100
 101         for (;;)
 102           {
 103             /* Invariants: The inequality needle[x..i-1] != needle[0..i-1-x]
 104                is known to hold for x < i-1-j.
 105                Furthermore, if j>0: needle[i-1-j..i-2] = needle[0..j-1].  */
 106             if (mb_equal (*b, needle_mbchars[j]))
 107               {
 108                 /* Set table[i] := i-1-j.  */
 109                 table[i] = i - ++j;
 110                 break;
 111               }
 112             /* The inequality needle[x..i-1] != needle[0..i-1-x] also holds
 113                for x = i-1-j, because
 114                  needle[i-1] != needle[j] = needle[i-1-x].  */
 115             if (j == 0)
 116               {
 117                 /* The inequality holds for all possible x.  */
 118                 table[i] = i;
 119                 break;
 120               }
 121             /* The inequality needle[x..i-1] != needle[0..i-1-x] also holds
 122                for i-1-j < x < i-1-j+table[j], because for these x:
 123                  needle[x..i-2]
 124                  = needle[x-(i-1-j)..j-1]
 125                  != needle[0..j-1-(x-(i-1-j))]  (by definition of table[j])
 126                     = needle[0..i-2-x],
 127                hence needle[x..i-1] != needle[0..i-1-x].
 128                Furthermore
 129                  needle[i-1-j+table[j]..i-2]
 130                  = needle[table[j]..j-1]
 131                  = needle[0..j-1-table[j]]  (by definition of table[j]).  */
 132             j = j - table[j];
 133           }
 134         /* Here: j = i - table[i].  */
 135       }
 136   }
 137
 138   /* Search, using the table to accelerate the processing.  */
 139   {
 140     size_t j;
 141     mbui_iterator_t rhaystack;
 142     mbui_iterator_t phaystack;
 143
 144     *resultp = NULL;
 145     j = 0;
 146     mbui_init (rhaystack, haystack);
 147     mbui_init (phaystack, haystack);
 148     /* Invariant: phaystack = rhaystack + j.  */
 149     while (mbui_avail (phaystack))
 150       {
 151         mbchar_t c;
 152
 153         mb_copy (&c, &mbui_cur (phaystack));
 154         if (c.wc_valid)
 155           c.wc = towlower (c.wc);
 156         if (mb_equal (needle_mbchars[j], c))
 157           {
 158             j++;
 159             mbui_advance (phaystack);
 160             if (j == m)
 161               {
 162                 /* The entire needle has been found.  */
 163                 *resultp = mbui_cur_ptr (rhaystack);
 164                 break;
 165               }
 166           }
 167         else if (j > 0)
 168           {
 169             /* Found a match of needle[0..j-1], mismatch at needle[j].  */
 170             size_t count = table[j];
 171             j -= count;
 172             for (; count > 0; count--)
 173               {
 174                 if (!mbui_avail (rhaystack))
 175                   abort ();
 176                 mbui_advance (rhaystack);
 177               }
 178           }
 179         else
 180           {
 181             /* Found a mismatch at needle[0] already.  */
 182             if (!mbui_avail (rhaystack))
 183               abort ();
 184             mbui_advance (rhaystack);
 185             mbui_advance (phaystack);
 186           }
 187       }
 188   }
 189
 190   freea (memory);
 191   return true;
 192 }
 193
 194 /* Find the first occurrence of the character string NEEDLE in the character
 195    string HAYSTACK, using case-insensitive comparison.
 196    Note: This function may, in multibyte locales, return success even if
 197    strlen (haystack) < strlen (needle) !  */
 198 char *
 199 mbscasestr (const char *haystack, const char *needle)
 200 {
 201   /* Be careful not to look at the entire extent of haystack or needle
 202      until needed.  This is useful because of these two cases:
 203        - haystack may be very long, and a match of needle found early,
 204        - needle may be very long, and not even a short initial segment of
 205          needle may be found in haystack.  */
 206   if (MB_CUR_MAX > 1)
 207     {
 208       mbui_iterator_t iter_needle;
 209
 210       mbui_init (iter_needle, needle);
 211       if (mbui_avail (iter_needle))
 212         {
 213           /* Minimizing the worst-case complexity:
 214              Let n = mbslen(haystack), m = mbslen(needle).
 215              The naïve algorithm is O(n*m) worst-case.
 216              The Knuth-Morris-Pratt algorithm is O(n) worst-case but it needs a
 217              memory allocation.
 218              To achieve linear complexity and yet amortize the cost of the
 219              memory allocation, we activate the Knuth-Morris-Pratt algorithm
 220              only once the naïve algorithm has already run for some time; more
 221              precisely, when
 222                - the outer loop count is >= 10,
 223                - the average number of comparisons per outer loop is >= 5,
 224                - the total number of comparisons is >= m.
 225              But we try it only once.  If the memory allocation attempt failed,
 226              we don't retry it.  */
 227           bool try_kmp = true;
 228           size_t outer_loop_count = 0;
 229           size_t comparison_count = 0;
 230           size_t last_ccount = 0;                  /* last comparison count */
 231           mbui_iterator_t iter_needle_last_ccount; /* = needle + last_ccount */
 232
 233           mbchar_t b;
 234           mbui_iterator_t iter_haystack;
 235
 236           mbui_init (iter_needle_last_ccount, needle);
 237
 238           mb_copy (&b, &mbui_cur (iter_needle));
 239           if (b.wc_valid)
 240             b.wc = towlower (b.wc);
 241
 242           mbui_init (iter_haystack, haystack);
 243           for (;; mbui_advance (iter_haystack))
 244             {
 245               mbchar_t c;
 246
 247               if (!mbui_avail (iter_haystack))
 248                 /* No match.  */
 249                 return NULL;
 250
 251               /* See whether it's advisable to use an asymptotically faster
 252                  algorithm.  */
 253               if (try_kmp
 254                   && outer_loop_count >= 10
 255                   && comparison_count >= 5 * outer_loop_count)
 256                 {
 257                   /* See if needle + comparison_count now reaches the end of
 258                      needle.  */
 259                   size_t count = comparison_count - last_ccount;
 260                   for (;
 261                        count > 0 && mbui_avail (iter_needle_last_ccount);
 262                        count--)
 263                     mbui_advance (iter_needle_last_ccount);
 264                   last_ccount = comparison_count;
 265                   if (!mbui_avail (iter_needle_last_ccount))
 266                     {
 267                       /* Try the Knuth-Morris-Pratt algorithm.  */
 268                       const char *result;
 269                       bool success =
 270                         knuth_morris_pratt_multibyte (haystack, needle,
 271                                                       &result);
 272                       if (success)
 273                         return (char *) result;
 274                       try_kmp = false;
 275                     }
 276                 }
 277
 278               outer_loop_count++;
 279               comparison_count++;
 280               mb_copy (&c, &mbui_cur (iter_haystack));
 281               if (c.wc_valid)
 282                 c.wc = towlower (c.wc);
 283               if (mb_equal (c, b))
 284                 /* The first character matches.  */
 285                 {
 286                   mbui_iterator_t rhaystack;
 287                   mbui_iterator_t rneedle;
 288
 289                   memcpy (&rhaystack, &iter_haystack, sizeof (mbui_iterator_t));
 290                   mbui_advance (rhaystack);
 291
 292                   mbui_init (rneedle, needle);
 293                   if (!mbui_avail (rneedle))
 294                     abort ();
 295                   mbui_advance (rneedle);
 296
 297                   for (;; mbui_advance (rhaystack), mbui_advance (rneedle))
 298                     {
 299                       if (!mbui_avail (rneedle))
 300                         /* Found a match.  */
 301                         return (char *) mbui_cur_ptr (iter_haystack);
 302                       if (!mbui_avail (rhaystack))
 303                         /* No match.  */
 304                         return NULL;
 305                       comparison_count++;
 306                       if (!mb_caseequal (mbui_cur (rhaystack),
 307                                          mbui_cur (rneedle)))
 308                         /* Nothing in this round.  */
 309                         break;
 310                     }
 311                 }
 312             }
 313         }
 314       else
 315         return (char *) haystack;
 316     }
 317   else
 318     {
 319       if (*needle != '\0')
 320         {
 321           /* Minimizing the worst-case complexity:
 322              Let n = strlen(haystack), m = strlen(needle).
 323              The naïve algorithm is O(n*m) worst-case.
 324              The Knuth-Morris-Pratt algorithm is O(n) worst-case but it needs a
 325              memory allocation.
 326              To achieve linear complexity and yet amortize the cost of the
 327              memory allocation, we activate the Knuth-Morris-Pratt algorithm
 328              only once the naïve algorithm has already run for some time; more
 329              precisely, when
 330                - the outer loop count is >= 10,
 331                - the average number of comparisons per outer loop is >= 5,
 332                - the total number of comparisons is >= m.
 333              But we try it only once.  If the memory allocation attempt failed,
 334              we don't retry it.  */
 335           bool try_kmp = true;
 336           size_t outer_loop_count = 0;
 337           size_t comparison_count = 0;
 338           size_t last_ccount = 0;                  /* last comparison count */
 339           const char *needle_last_ccount = needle; /* = needle + last_ccount */
 340
 341           /* Speed up the following searches of needle by caching its first
 342              character.  */
 343           unsigned char b = TOLOWER ((unsigned char) *needle);
 344
 345           needle++;
 346           for (;; haystack++)
 347             {
 348               if (*haystack == '\0')
 349                 /* No match.  */
 350                 return NULL;
 351
 352               /* See whether it's advisable to use an asymptotically faster
 353                  algorithm.  */
 354               if (try_kmp
 355                   && outer_loop_count >= 10
 356                   && comparison_count >= 5 * outer_loop_count)
 357                 {
 358                   /* See if needle + comparison_count now reaches the end of
 359                      needle.  */
 360                   if (needle_last_ccount != NULL)
 361                     {
 362                       needle_last_ccount +=
 363                         strnlen (needle_last_ccount,
 364                                  comparison_count - last_ccount);
 365                       if (*needle_last_ccount == '\0')
 366                         needle_last_ccount = NULL;
 367                       last_ccount = comparison_count;
 368                     }
 369                   if (needle_last_ccount == NULL)
 370                     {
 371                       /* Try the Knuth-Morris-Pratt algorithm.  */
 372                       const unsigned char *result;
 373                       bool success =
 374                         knuth_morris_pratt ((const unsigned char *) haystack,
 375                                             (const unsigned char *) (needle - 1),
 376                                             strlen (needle - 1),
 377                                             &result);
 378                       if (success)
 379                         return (char *) result;
 380                       try_kmp = false;
 381                     }
 382                 }
 383
 384               outer_loop_count++;
 385               comparison_count++;
 386               if (TOLOWER ((unsigned char) *haystack) == b)
 387                 /* The first character matches.  */
 388                 {
 389                   const char *rhaystack = haystack + 1;
 390                   const char *rneedle = needle;
 391
 392                   for (;; rhaystack++, rneedle++)
 393                     {
 394                       if (*rneedle == '\0')
 395                         /* Found a match.  */
 396                         return (char *) haystack;
 397                       if (*rhaystack == '\0')
 398                         /* No match.  */
 399                         return NULL;
 400                       comparison_count++;
 401                       if (TOLOWER ((unsigned char) *rhaystack)
 402                           != TOLOWER ((unsigned char) *rneedle))
 403                         /* Nothing in this round.  */
 404                         break;
 405                     }
 406                 }
 407             }
 408         }
 409       else
 410         return (char *) haystack;
 411     }
 412 }