contrib/less/pattern.c

   1 /*
   2  * Copyright (C) 1984-2015  Mark Nudelman
   3  *
   4  * You may distribute under the terms of either the GNU General Public
   5  * License or the Less License, as specified in the README file.
   6  *
   7  * For more information, see the README file.
   8  */
   9
  10 /*
  11  * Routines to do pattern matching.
  12  */
  13
  14 #include "less.h"
  15 #include "pattern.h"
  16
  17 extern int caseless;
  18
  19 /*
  20  * Compile a search pattern, for future use by match_pattern.
  21  */
  22         static int
  23 compile_pattern2(pattern, search_type, comp_pattern, show_error)
  24         char *pattern;
  25         int search_type;
  26         void **comp_pattern;
  27         int show_error;
  28 {
  29         if (search_type & SRCH_NO_REGEX)
  30                 return (0);
  31   {
  32 #if HAVE_GNU_REGEX
  33         struct re_pattern_buffer *comp = (struct re_pattern_buffer *)
  34                 ecalloc(1, sizeof(struct re_pattern_buffer));
  35         struct re_pattern_buffer **pcomp =
  36                 (struct re_pattern_buffer **) comp_pattern;
  37         re_set_syntax(RE_SYNTAX_POSIX_EXTENDED);
  38         if (re_compile_pattern(pattern, strlen(pattern), comp))
  39         {
  40                 free(comp);
  41                 if (show_error)
  42                         error("Invalid pattern", NULL_PARG);
  43                 return (-1);
  44         }
  45         if (*pcomp != NULL)
  46                 regfree(*pcomp);
  47         *pcomp = comp;
  48 #endif
  49 #if HAVE_POSIX_REGCOMP
  50         regex_t *comp = (regex_t *) ecalloc(1, sizeof(regex_t));
  51         regex_t **pcomp = (regex_t **) comp_pattern;
  52         if (regcomp(comp, pattern, REGCOMP_FLAG))
  53         {
  54                 free(comp);
  55                 if (show_error)
  56                         error("Invalid pattern", NULL_PARG);
  57                 return (-1);
  58         }
  59         if (*pcomp != NULL)
  60                 regfree(*pcomp);
  61         *pcomp = comp;
  62 #endif
  63 #if HAVE_PCRE
  64         pcre *comp;
  65         pcre **pcomp = (pcre **) comp_pattern;
  66         constant char *errstring;
  67         int erroffset;
  68         PARG parg;
  69         comp = pcre_compile(pattern, 0,
  70                         &errstring, &erroffset, NULL);
  71         if (comp == NULL)
  72         {
  73                 parg.p_string = (char *) errstring;
  74                 if (show_error)
  75                         error("%s", &parg);
  76                 return (-1);
  77         }
  78         *pcomp = comp;
  79 #endif
  80 #if HAVE_RE_COMP
  81         PARG parg;
  82         int *pcomp = (int *) comp_pattern;
  83         if ((parg.p_string = re_comp(pattern)) != NULL)
  84         {
  85                 if (show_error)
  86                         error("%s", &parg);
  87                 return (-1);
  88         }
  89         *pcomp = 1;
  90 #endif
  91 #if HAVE_REGCMP
  92         char *comp;
  93         char **pcomp = (char **) comp_pattern;
  94         if ((comp = regcmp(pattern, 0)) == NULL)
  95         {
  96                 if (show_error)
  97                         error("Invalid pattern", NULL_PARG);
  98                 return (-1);
  99         }
 100         if (pcomp != NULL)
 101                 free(*pcomp);
 102         *pcomp = comp;
 103 #endif
 104 #if HAVE_V8_REGCOMP
 105         struct regexp *comp;
 106         struct regexp **pcomp = (struct regexp **) comp_pattern;
 107         reg_show_error = show_error;
 108         comp = regcomp(pattern);
 109         reg_show_error = 1;
 110         if (comp == NULL)
 111         {
 112                 /*
 113                  * regcomp has already printed an error message
 114                  * via regerror().
 115                  */
 116                 return (-1);
 117         }
 118         if (*pcomp != NULL)
 119                 free(*pcomp);
 120         *pcomp = comp;
 121 #endif
 122   }
 123         return (0);
 124 }
 125
 126 /*
 127  * Like compile_pattern2, but convert the pattern to lowercase if necessary.
 128  */
 129         public int
 130 compile_pattern(pattern, search_type, comp_pattern)
 131         char *pattern;
 132         int search_type;
 133         void **comp_pattern;
 134 {
 135         char *cvt_pattern;
 136         int result;
 137
 138         if (caseless != OPT_ONPLUS)
 139                 cvt_pattern = pattern;
 140         else
 141         {
 142                 cvt_pattern = (char*) ecalloc(1, cvt_length(strlen(pattern), CVT_TO_LC));
 143                 cvt_text(cvt_pattern, pattern, (int *)NULL, (int *)NULL, CVT_TO_LC);
 144         }
 145         result = compile_pattern2(cvt_pattern, search_type, comp_pattern, 1);
 146         if (cvt_pattern != pattern)
 147                 free(cvt_pattern);
 148         return (result);
 149 }
 150
 151 /*
 152  * Forget that we have a compiled pattern.
 153  */
 154         public void
 155 uncompile_pattern(pattern)
 156         void **pattern;
 157 {
 158 #if HAVE_GNU_REGEX
 159         struct re_pattern_buffer **pcomp = (struct re_pattern_buffer **) pattern;
 160         if (*pcomp != NULL)
 161                 regfree(*pcomp);
 162         *pcomp = NULL;
 163 #endif
 164 #if HAVE_POSIX_REGCOMP
 165         regex_t **pcomp = (regex_t **) pattern;
 166         if (*pcomp != NULL)
 167                 regfree(*pcomp);
 168         *pcomp = NULL;
 169 #endif
 170 #if HAVE_PCRE
 171         pcre **pcomp = (pcre **) pattern;
 172         if (*pcomp != NULL)
 173                 pcre_free(*pcomp);
 174         *pcomp = NULL;
 175 #endif
 176 #if HAVE_RE_COMP
 177         int *pcomp = (int *) pattern;
 178         *pcomp = 0;
 179 #endif
 180 #if HAVE_REGCMP
 181         char **pcomp = (char **) pattern;
 182         if (*pcomp != NULL)
 183                 free(*pcomp);
 184         *pcomp = NULL;
 185 #endif
 186 #if HAVE_V8_REGCOMP
 187         struct regexp **pcomp = (struct regexp **) pattern;
 188         if (*pcomp != NULL)
 189                 free(*pcomp);
 190         *pcomp = NULL;
 191 #endif
 192 }
 193
 194 /*
 195  * Can a pattern be successfully compiled?
 196  */
 197         public int
 198 valid_pattern(pattern)
 199         char *pattern;
 200 {
 201         void *comp_pattern;
 202         int result;
 203
 204         CLEAR_PATTERN(comp_pattern);
 205         result = compile_pattern2(pattern, 0, &comp_pattern, 0);
 206         if (result != 0)
 207                 return (0);
 208         uncompile_pattern(&comp_pattern);
 209         return (1);
 210 }
 211
 212 /*
 213  * Is a compiled pattern null?
 214  */
 215         public int
 216 is_null_pattern(pattern)
 217         void *pattern;
 218 {
 219 #if HAVE_GNU_REGEX
 220         return (pattern == NULL);
 221 #endif
 222 #if HAVE_POSIX_REGCOMP
 223         return (pattern == NULL);
 224 #endif
 225 #if HAVE_PCRE
 226         return (pattern == NULL);
 227 #endif
 228 #if HAVE_RE_COMP
 229         return (pattern == 0);
 230 #endif
 231 #if HAVE_REGCMP
 232         return (pattern == NULL);
 233 #endif
 234 #if HAVE_V8_REGCOMP
 235         return (pattern == NULL);
 236 #endif
 237 #if NO_REGEX
 238         return (pattern == NULL);
 239 #endif
 240 }
 241
 242 /*
 243  * Simple pattern matching function.
 244  * It supports no metacharacters like *, etc.
 245  */
 246         static int
 247 match(pattern, pattern_len, buf, buf_len, pfound, pend)
 248         char *pattern;
 249         int pattern_len;
 250         char *buf;
 251         int buf_len;
 252         char **pfound, **pend;
 253 {
 254         register char *pp, *lp;
 255         register char *pattern_end = pattern + pattern_len;
 256         register char *buf_end = buf + buf_len;
 257
 258         for ( ;  buf < buf_end;  buf++)
 259         {
 260                 for (pp = pattern, lp = buf;  ;  pp++, lp++)
 261                 {
 262                         char cp = *pp;
 263                         char cl = *lp;
 264                         if (caseless == OPT_ONPLUS && ASCII_IS_UPPER(cp))
 265                                 cp = ASCII_TO_LOWER(cp);
 266                         if (cp != cl)
 267                                 break;
 268                         if (pp == pattern_end || lp == buf_end)
 269                                 break;
 270                 }
 271                 if (pp == pattern_end)
 272                 {
 273                         if (pfound != NULL)
 274                                 *pfound = buf;
 275                         if (pend != NULL)
 276                                 *pend = lp;
 277                         return (1);
 278                 }
 279         }
 280         return (0);
 281 }
 282
 283 /*
 284  * Perform a pattern match with the previously compiled pattern.
 285  * Set sp and ep to the start and end of the matched string.
 286  */
 287         public int
 288 match_pattern(pattern, tpattern, line, line_len, sp, ep, notbol, search_type)
 289         void *pattern;
 290         char *tpattern;
 291         char *line;
 292         int line_len;
 293         char **sp;
 294         char **ep;
 295         int notbol;
 296         int search_type;
 297 {
 298         int matched;
 299 #if HAVE_GNU_REGEX
 300         struct re_pattern_buffer *spattern = (struct re_pattern_buffer *) pattern;
 301 #endif
 302 #if HAVE_POSIX_REGCOMP
 303         regex_t *spattern = (regex_t *) pattern;
 304 #endif
 305 #if HAVE_PCRE
 306         pcre *spattern = (pcre *) pattern;
 307 #endif
 308 #if HAVE_RE_COMP
 309         int spattern = (int) pattern;
 310 #endif
 311 #if HAVE_REGCMP
 312         char *spattern = (char *) pattern;
 313 #endif
 314 #if HAVE_V8_REGCOMP
 315         struct regexp *spattern = (struct regexp *) pattern;
 316 #endif
 317
 318         *sp = *ep = NULL;
 319 #if NO_REGEX
 320         search_type |= SRCH_NO_REGEX;
 321 #endif
 322         if (search_type & SRCH_NO_REGEX)
 323                 matched = match(tpattern, strlen(tpattern), line, line_len, sp, ep);
 324         else
 325         {
 326 #if HAVE_GNU_REGEX
 327         {
 328                 struct re_registers search_regs;
 329                 spattern->not_bol = notbol;
 330                 spattern->regs_allocated = REGS_UNALLOCATED;
 331                 matched = re_search(spattern, line, line_len, 0, line_len, &search_regs) >= 0;
 332                 if (matched)
 333                 {
 334                         *sp = line + search_regs.start[0];
 335                         *ep = line + search_regs.end[0];
 336                 }
 337         }
 338 #endif
 339 #if HAVE_POSIX_REGCOMP
 340         {
 341                 regmatch_t rm;
 342                 int flags = (notbol) ? REG_NOTBOL : 0;
 343 #ifdef REG_STARTEND
 344                 flags |= REG_STARTEND;
 345                 rm.rm_so = 0;
 346                 rm.rm_eo = line_len;
 347 #endif
 348                 matched = !regexec(spattern, line, 1, &rm, flags);
 349                 if (matched)
 350                 {
 351 #ifndef __WATCOMC__
 352                         *sp = line + rm.rm_so;
 353                         *ep = line + rm.rm_eo;
 354 #else
 355                         *sp = rm.rm_sp;
 356                         *ep = rm.rm_ep;
 357 #endif
 358                 }
 359         }
 360 #endif
 361 #if HAVE_PCRE
 362         {
 363                 int flags = (notbol) ? PCRE_NOTBOL : 0;
 364                 int ovector[3];
 365                 matched = pcre_exec(spattern, NULL, line, line_len,
 366                         0, flags, ovector, 3) >= 0;
 367                 if (matched)
 368                 {
 369                         *sp = line + ovector[0];
 370                         *ep = line + ovector[1];
 371                 }
 372         }
 373 #endif
 374 #if HAVE_RE_COMP
 375         matched = (re_exec(line) == 1);
 376         /*
 377          * re_exec doesn't seem to provide a way to get the matched string.
 378          */
 379         *sp = *ep = NULL;
 380 #endif
 381 #if HAVE_REGCMP
 382         *ep = regex(spattern, line);
 383         matched = (*ep != NULL);
 384         if (matched)
 385                 *sp = __loc1;
 386 #endif
 387 #if HAVE_V8_REGCOMP
 388 #if HAVE_REGEXEC2
 389         matched = regexec2(spattern, line, notbol);
 390 #else
 391         matched = regexec(spattern, line);
 392 #endif
 393         if (matched)
 394         {
 395                 *sp = spattern->startp[0];
 396                 *ep = spattern->endp[0];
 397         }
 398 #endif
 399         }
 400         matched = (!(search_type & SRCH_NO_MATCH) && matched) ||
 401                         ((search_type & SRCH_NO_MATCH) && !matched);
 402         return (matched);
 403 }
 404