posix/tst-regex.c

   1 /* Copyright (C) 2001, 2003 Free Software Foundation, Inc.
   2    This file is part of the GNU C Library.
   3
   4    The GNU C Library is free software; you can redistribute it and/or
   5    modify it under the terms of the GNU Lesser General Public
   6    License as published by the Free Software Foundation; either
   7    version 2.1 of the License, or (at your option) any later version.
   8
   9    The GNU C Library is distributed in the hope that it will be useful,
  10    but WITHOUT ANY WARRANTY; without even the implied warranty of
  11    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12    Lesser General Public License for more details.
  13
  14    You should have received a copy of the GNU Lesser General Public
  15    License along with the GNU C Library; if not, write to the Free
  16    Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
  17    02111-1307 USA.  */
  18
  19 #include <spawn.h>
  20 #include "spawn_int.h"
  21
  22 #include <assert.h>
  23 #include <errno.h>
  24 #include <error.h>
  25 #include <fcntl.h>
  26 #include <getopt.h>
  27 #include <iconv.h>
  28 #include <locale.h>
  29 #include <mcheck.h>
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <time.h>
  34 #include <unistd.h>
  35 #include <sys/stat.h>
  36 #include <sys/types.h>
  37 #include <regex.h>
  38
  39
  40 #ifdef _POSIX_CPUTIME
  41 static clockid_t cl;
  42 static int use_clock;
  43 #endif
  44 static iconv_t cd;
  45 static char *mem;
  46 static char *umem;
  47 static size_t memlen;
  48 static size_t umemlen;
  49 static int timing;
  50
  51 static int test_expr (const char *expr, int expected, int expectedicase);
  52 static int run_test (const char *expr, const char *mem, size_t memlen,
  53                      int icase, int expected);
  54 static int run_test_backwards (const char *expr, const char *mem,
  55                                size_t memlen, int icase, int expected);
  56
  57
  58 int
  59 main (int argc, char *argv[])
  60 {
  61   const char *file;
  62   int fd;
  63   struct stat st;
  64   int result;
  65   char *inmem;
  66   char *outmem;
  67   size_t inlen;
  68   size_t outlen;
  69   static const struct option options[] =
  70     {
  71       {"timing",no_argument,    &timing,        1 },
  72       {NULL,    0,              NULL,           0 }
  73     };
  74
  75   mtrace ();
  76
  77   while (getopt_long (argc, argv, "", options, NULL) >= 0);
  78
  79   /* Make the content of the file available in memory.  */
  80   file = "../ChangeLog.8";
  81   fd = open (file, O_RDONLY);
  82   if (fd == -1)
  83     error (EXIT_FAILURE, errno, "cannot open %s", basename (file));
  84
  85   if (fstat (fd, &st) != 0)
  86     error (EXIT_FAILURE, errno, "cannot stat %s", basename (file));
  87   memlen = st.st_size;
  88
  89   mem = (char *) malloc (memlen + 1);
  90   if (mem == NULL)
  91     error (EXIT_FAILURE, errno, "while allocating buffer");
  92
  93   if ((size_t) read (fd, mem, memlen) != memlen)
  94     error (EXIT_FAILURE, 0, "cannot read entire file");
  95   mem[memlen] = '\0';
  96
  97   close (fd);
  98
  99   /* We have to convert a few things from Latin-1 to UTF-8.  */
 100   cd = iconv_open ("UTF-8", "ISO-8859-1");
 101   if (cd == (iconv_t) -1)
 102     error (EXIT_FAILURE, errno, "cannot get conversion descriptor");
 103
 104   /* For the second test we have to convert the file content to UTF-8.
 105      Since the text is mostly ASCII it should be enough to allocate
 106      twice as much memory for the UTF-8 text than for the Latin-1
 107      text.  */
 108   umem = (char *) calloc (2, memlen);
 109   if (umem == NULL)
 110     error (EXIT_FAILURE, errno, "while allocating buffer");
 111
 112   inmem = mem;
 113   inlen = memlen;
 114   outmem = umem;
 115   outlen = 2 * memlen - 1;
 116   iconv (cd, &inmem, &inlen, &outmem, &outlen);
 117   umemlen = outmem - umem;
 118   if (inlen != 0)
 119     error (EXIT_FAILURE, errno, "cannot convert buffer");
 120
 121 #ifdef _POSIX_CPUTIME
 122   /* See whether we can use the CPU clock.  */
 123   use_clock = clock_getcpuclockid (0, &cl) == 0;
 124 #endif
 125
 126 #ifdef DEBUG
 127   re_set_syntax (RE_DEBUG);
 128 #endif
 129
 130   /* Run the actual tests.  All tests are run in a single-byte and a
 131      multi-byte locale.  */
 132   result = test_expr ("[äáàâéèêíìîñöóòôüúùû]", 2, 2);
 133   result |= test_expr ("G.ran", 2, 3);
 134   result |= test_expr ("G.\\{1\\}ran", 2, 3);
 135   result |= test_expr ("G.*ran", 3, 44);
 136   result |= test_expr ("[äáàâ]", 0, 0);
 137   result |= test_expr ("Uddeborg", 2, 2);
 138   result |= test_expr (".Uddeborg", 2, 2);
 139
 140   /* Free the resources.  */
 141   free (umem);
 142   iconv_close (cd);
 143   free (mem);
 144
 145   return result;
 146 }
 147
 148
 149 static int
 150 test_expr (const char *expr, int expected, int expectedicase)
 151 {
 152   int result;
 153   char *inmem;
 154   char *outmem;
 155   size_t inlen;
 156   size_t outlen;
 157   char *uexpr;
 158
 159   /* First test: search with an ISO-8859-1 locale.  */
 160   if (setlocale (LC_ALL, "de_DE.ISO-8859-1") == NULL)
 161     error (EXIT_FAILURE, 0, "cannot set locale de_DE.ISO-8859-1");
 162
 163   printf ("\nTest \"%s\" with 8-bit locale\n", expr);
 164   result = run_test (expr, mem, memlen, 0, expected);
 165   printf ("\nTest \"%s\" with 8-bit locale, case insensitive\n", expr);
 166   result |= run_test (expr, mem, memlen, 1, expectedicase);
 167   printf ("\nTest \"%s\" backwards with 8-bit locale\n", expr);
 168   result |= run_test_backwards (expr, mem, memlen, 0, expected);
 169   printf ("\nTest \"%s\" backwards with 8-bit locale, case insensitive\n",
 170           expr);
 171   result |= run_test_backwards (expr, mem, memlen, 1, expectedicase);
 172
 173   /* Second test: search with an UTF-8 locale.  */
 174   if (setlocale (LC_ALL, "de_DE.UTF-8") == NULL)
 175     error (EXIT_FAILURE, 0, "cannot set locale de_DE.UTF-8");
 176
 177   inmem = (char *) expr;
 178   inlen = strlen (expr);
 179   outlen = inlen * MB_CUR_MAX;
 180   outmem = uexpr = alloca (outlen + 1);
 181   memset (outmem, '\0', outlen + 1);
 182   iconv (cd, &inmem, &inlen, &outmem, &outlen);
 183   if (inlen != 0)
 184     error (EXIT_FAILURE, errno, "cannot convert expression");
 185
 186   /* Run the tests.  */
 187   printf ("\nTest \"%s\" with multi-byte locale\n", expr);
 188   result |= run_test (uexpr, umem, umemlen, 0, expected);
 189   printf ("\nTest \"%s\" with multi-byte locale, case insensitive\n", expr);
 190   result |= run_test (uexpr, umem, umemlen, 1, expectedicase);
 191   printf ("\nTest \"%s\" backwards with multi-byte locale\n", expr);
 192   result |= run_test_backwards (uexpr, umem, umemlen, 0, expected);
 193   printf ("\nTest \"%s\" backwards with multi-byte locale, case insensitive\n",
 194           expr);
 195   result |= run_test_backwards (uexpr, umem, umemlen, 1, expectedicase);
 196
 197   return result;
 198 }
 199
 200
 201 static int
 202 run_test (const char *expr, const char *mem, size_t memlen, int icase,
 203           int expected)
 204 {
 205 #ifdef _POSIX_CPUTIME
 206   struct timespec start;
 207   struct timespec finish;
 208 #endif
 209   regex_t re;
 210   int err;
 211   size_t offset;
 212   int cnt;
 213
 214 #ifdef _POSIX_CPUTIME
 215   if (use_clock && !timing)
 216     use_clock = clock_gettime (cl, &start) == 0;
 217 #endif
 218
 219   err = regcomp (&re, expr, REG_NEWLINE | (icase ? REG_ICASE : 0));
 220   if (err != REG_NOERROR)
 221     {
 222       char buf[200];
 223       regerror (err, &re, buf, sizeof buf);
 224       error (EXIT_FAILURE, 0, "cannot compile expression: %s", buf);
 225     }
 226
 227   cnt = 0;
 228   offset = 0;
 229   assert (mem[memlen] == '\0');
 230   while (offset < memlen)
 231     {
 232       regmatch_t ma[1];
 233       const char *sp;
 234       const char *ep;
 235
 236       err = regexec (&re, mem + offset, 1, ma, 0);
 237       if (err == REG_NOMATCH)
 238         break;
 239
 240       if (err != REG_NOERROR)
 241         {
 242           char buf[200];
 243           regerror (err, &re, buf, sizeof buf);
 244           error (EXIT_FAILURE, 0, "cannot use expression: %s", buf);
 245         }
 246
 247       assert (ma[0].rm_so >= 0);
 248       sp = mem + offset + ma[0].rm_so;
 249       while (sp > mem && sp[-1] != '\n')
 250         --sp;
 251
 252       ep = mem + offset + ma[0].rm_so;
 253       while (*ep != '\0' && *ep != '\n')
 254         ++ep;
 255
 256       printf ("match %d: \"%.*s\"\n", ++cnt, (int) (ep - sp), sp);
 257
 258       offset = ep + 1 - mem;
 259     }
 260
 261   regfree (&re);
 262
 263 #ifdef _POSIX_CPUTIME
 264   if (use_clock && !timing)
 265     {
 266       use_clock = clock_gettime (cl, &finish) == 0;
 267       if (use_clock)
 268         {
 269           if (finish.tv_nsec < start.tv_nsec)
 270             {
 271               finish.tv_nsec -= start.tv_nsec - 1000000000;
 272               finish.tv_sec -= 1 + start.tv_sec;
 273             }
 274           else
 275             {
 276               finish.tv_nsec -= start.tv_nsec;
 277               finish.tv_sec -= start.tv_sec;
 278             }
 279
 280           printf ("elapsed time: %ld.%09ld sec\n",
 281                   finish.tv_sec, finish.tv_nsec);
 282         }
 283     }
 284
 285   if (use_clock && timing)
 286     {
 287       struct timespec mintime = { .tv_sec = 24 * 60 * 60 };
 288
 289       for (int i = 0; i < 10; ++i)
 290         {
 291           offset = 0;
 292           use_clock = clock_gettime (cl, &start) == 0;
 293
 294           if (!use_clock)
 295             continue;
 296
 297           err = regcomp (&re, expr, REG_NEWLINE | (icase ? REG_ICASE : 0));
 298           if (err != REG_NOERROR)
 299             continue;
 300
 301           while (offset < memlen)
 302             {
 303               regmatch_t ma[1];
 304
 305               err = regexec (&re, mem + offset, 1, ma, 0);
 306               if (err != REG_NOERROR)
 307                 break;
 308
 309               offset += ma[0].rm_eo;
 310             }
 311
 312           regfree (&re);
 313
 314           use_clock = clock_gettime (cl, &finish) == 0;
 315           if (use_clock)
 316             {
 317               if (finish.tv_nsec < start.tv_nsec)
 318                 {
 319                   finish.tv_nsec -= start.tv_nsec - 1000000000;
 320                   finish.tv_sec -= 1 + start.tv_sec;
 321                 }
 322               else
 323                 {
 324                   finish.tv_nsec -= start.tv_nsec;
 325                   finish.tv_sec -= start.tv_sec;
 326                 }
 327               if (finish.tv_sec < mintime.tv_sec
 328                   || (finish.tv_sec == mintime.tv_sec
 329                       && finish.tv_nsec < mintime.tv_nsec))
 330                 mintime = finish;
 331             }
 332         }
 333       printf ("elapsed time: %ld.%09ld sec\n",
 334               mintime.tv_sec, mintime.tv_nsec);
 335     }
 336 #endif
 337
 338   /* Return an error if the number of matches found is not match we
 339      expect.  */
 340   return cnt != expected;
 341 }
 342
 343
 344 static int
 345 run_test_backwards (const char *expr, const char *mem, size_t memlen,
 346                     int icase, int expected)
 347 {
 348 #ifdef _POSIX_CPUTIME
 349   struct timespec start;
 350   struct timespec finish;
 351 #endif
 352   struct re_pattern_buffer re;
 353   const char *err;
 354   size_t offset;
 355   int cnt;
 356
 357 #ifdef _POSIX_CPUTIME
 358   if (use_clock && !timing)
 359     use_clock = clock_gettime (cl, &start) == 0;
 360 #endif
 361
 362   re_set_syntax ((RE_SYNTAX_POSIX_BASIC & ~RE_DOT_NEWLINE)
 363                  | RE_HAT_LISTS_NOT_NEWLINE
 364                  | (icase ? RE_ICASE : 0));
 365
 366   memset (&re, 0, sizeof (re));
 367   re.fastmap = malloc (256);
 368   if (re.fastmap == NULL)
 369     error (EXIT_FAILURE, errno, "cannot allocate fastmap");
 370
 371   err = re_compile_pattern (expr, strlen (expr), &re);
 372   if (err != NULL)
 373     error (EXIT_FAILURE, 0, "cannot compile expression: %s", err);
 374
 375   if (re_compile_fastmap (&re))
 376     error (EXIT_FAILURE, 0, "couldn't compile fastmap");
 377
 378   cnt = 0;
 379   offset = memlen;
 380   assert (mem[memlen] == '\0');
 381   while (offset <= memlen)
 382     {
 383       int start;
 384       const char *sp;
 385       const char *ep;
 386
 387       start = re_search (&re, mem, memlen, offset, -offset, NULL);
 388       if (start == -1)
 389         break;
 390
 391       if (start == -2)
 392         error (EXIT_FAILURE, 0, "internal error in re_search");
 393
 394       sp = mem + start;
 395       while (sp > mem && sp[-1] != '\n')
 396         --sp;
 397
 398       ep = mem + start;
 399       while (*ep != '\0' && *ep != '\n')
 400         ++ep;
 401
 402       printf ("match %d: \"%.*s\"\n", ++cnt, (int) (ep - sp), sp);
 403
 404       offset = sp - 1 - mem;
 405     }
 406
 407   regfree (&re);
 408
 409 #ifdef _POSIX_CPUTIME
 410   if (use_clock && !timing)
 411     {
 412       use_clock = clock_gettime (cl, &finish) == 0;
 413       if (use_clock)
 414         {
 415           if (finish.tv_nsec < start.tv_nsec)
 416             {
 417               finish.tv_nsec -= start.tv_nsec - 1000000000;
 418               finish.tv_sec -= 1 + start.tv_sec;
 419             }
 420           else
 421             {
 422               finish.tv_nsec -= start.tv_nsec;
 423               finish.tv_sec -= start.tv_sec;
 424             }
 425
 426           printf ("elapsed time: %ld.%09ld sec\n",
 427                   finish.tv_sec, finish.tv_nsec);
 428         }
 429     }
 430
 431   if (use_clock && timing)
 432     {
 433       struct timespec mintime = { .tv_sec = 24 * 60 * 60 };
 434
 435       for (int i = 0; i < 10; ++i)
 436         {
 437           offset = memlen;
 438           use_clock = clock_gettime (cl, &start) == 0;
 439
 440           if (!use_clock)
 441             continue;
 442
 443           memset (&re, 0, sizeof (re));
 444           re.fastmap = malloc (256);
 445           if (re.fastmap == NULL)
 446             continue;
 447
 448           err = re_compile_pattern (expr, strlen (expr), &re);
 449           if (err != NULL)
 450             continue;
 451
 452           if (re_compile_fastmap (&re))
 453             {
 454               regfree (&re);
 455               continue;
 456             }
 457
 458           while (offset <= memlen)
 459             {
 460               int start;
 461               const char *sp;
 462
 463               start = re_search (&re, mem, memlen, offset, -offset, NULL);
 464               if (start < -1)
 465                 break;
 466
 467               sp = mem + start;
 468               while (sp > mem && sp[-1] != '\n')
 469                 --sp;
 470
 471               offset = sp - 1 - mem;
 472             }
 473
 474           regfree (&re);
 475
 476           use_clock = clock_gettime (cl, &finish) == 0;
 477           if (use_clock)
 478             {
 479               if (finish.tv_nsec < start.tv_nsec)
 480                 {
 481                   finish.tv_nsec -= start.tv_nsec - 1000000000;
 482                   finish.tv_sec -= 1 + start.tv_sec;
 483                 }
 484               else
 485                 {
 486                   finish.tv_nsec -= start.tv_nsec;
 487                   finish.tv_sec -= start.tv_sec;
 488                 }
 489               if (finish.tv_sec < mintime.tv_sec
 490                   || (finish.tv_sec == mintime.tv_sec
 491                       && finish.tv_nsec < mintime.tv_nsec))
 492                 mintime = finish;
 493             }
 494         }
 495       printf ("elapsed time: %ld.%09ld sec\n",
 496               mintime.tv_sec, mintime.tv_nsec);
 497     }
 498 #endif
 499
 500   /* Return an error if the number of matches found is not match we
 501      expect.  */
 502   return cnt != expected;
 503 }