mpdm_r.c

   1 /*
   2
   3     MPDM - Minimum Profit Data Manager
   4     Copyright (C) 2003/2007 Angel Ortega <angel@triptico.com>
   5
   6     mpdm_r.c - Regular expressions
   7
   8     This program is free software; you can redistribute it and/or
   9     modify it under the terms of the GNU General Public License
  10     as published by the Free Software Foundation; either version 2
  11     of the License, or (at your option) any later version.
  12
  13     This program is distributed in the hope that it will be useful,
  14     but WITHOUT ANY WARRANTY; without even the implied warranty of
  15     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16     GNU General Public License for more details.
  17
  18     You should have received a copy of the GNU General Public License
  19     along with this program; if not, write to the Free Software
  20     Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  21
  22     http://www.triptico.com
  23
  24 */
  25
  26 #include "config.h"
  27
  28 #include <stdio.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31 #include <wchar.h>
  32
  33 #include "mpdm.h"
  34
  35 #ifdef CONFOPT_PCRE
  36 #include <pcreposix.h>
  37 #endif
  38
  39 #ifdef CONFOPT_SYSTEM_REGEX
  40 #include <regex.h>
  41 #endif
  42
  43 #ifdef CONFOPT_INCLUDED_REGEX
  44 #include "gnu_regex.h"
  45 #endif
  46
  47 /*******************
  48         Data
  49 ********************/
  50
  51 /* matching of the last regex */
  52
  53 int mpdm_regex_offset = -1;
  54 int mpdm_regex_size = 0;
  55
  56 /* number of substitutions in last sregex */
  57
  58 int mpdm_sregex_count = 0;
  59
  60 /*******************
  61         Code
  62 ********************/
  63
  64 static wchar_t *regex_flags(const mpdm_t r)
  65 {
  66         return wcsrchr((wchar_t *) r->data, *(wchar_t *) r->data);
  67 }
  68
  69
  70 static mpdm_t mpdm_regcomp(mpdm_t r)
  71 {
  72         mpdm_t c = NULL;
  73         mpdm_t regex_cache = NULL;
  74
  75         /* if cache does not exist, create it */
  76         if ((regex_cache = mpdm_hget_s(mpdm_root(), L"__REGEX_CACHE__")) == NULL) {
  77                 regex_cache = MPDM_H(0);
  78                 mpdm_hset_s(mpdm_root(), L"__REGEX_CACHE__", regex_cache);
  79         }
  80
  81         /* search the regex in the cache */
  82         if ((c = mpdm_hget(regex_cache, r)) == NULL) {
  83                 mpdm_t rmb;
  84                 regex_t re;
  85                 char *regex;
  86                 char *flags;
  87                 int f = REG_EXTENDED;
  88
  89                 /* not found; regex must be compiled */
  90
  91                 /* convert to mbs */
  92                 rmb = MPDM_2MBS(r->data);
  93
  94                 regex = (char *) rmb->data;
  95                 if ((flags = strrchr(regex, *regex)) == NULL)
  96                         return NULL;
  97
  98                 if (strchr(flags, 'i') != NULL)
  99                         f |= REG_ICASE;
 100                 if (strchr(flags, 'm') != NULL)
 101                         f |= REG_NEWLINE;
 102
 103                 regex++;
 104                 *flags = '\0';
 105
 106                 if (!regcomp(&re, regex, f)) {
 107                         void *ptr;
 108
 109                         if ((ptr = malloc(sizeof(regex_t))) != NULL) {
 110                                 /* copies */
 111                                 memcpy(ptr, &re, sizeof(regex_t));
 112
 113                                 /* create value */
 114                                 c = mpdm_new(MPDM_FREE, ptr, sizeof(regex_t));
 115
 116                                 /* stores */
 117                                 mpdm_hset(regex_cache, r, c);
 118                         }
 119                 }
 120         }
 121
 122         return c;
 123 }
 124
 125
 126 static mpdm_t regex1(mpdm_t r, const mpdm_t v, int offset)
 127 /* test for one regex */
 128 {
 129         mpdm_t w = NULL;
 130         mpdm_t cr;
 131
 132         /* no matching yet */
 133         mpdm_regex_offset = -1;
 134
 135         /* compile the regex */
 136         if ((cr = mpdm_regcomp(r)) != NULL) {
 137                 regmatch_t rm;
 138                 char *ptr;
 139                 wchar_t *last;
 140                 int o = 0;
 141                 int f = 0;
 142
 143                 /* takes pointer to 'last' flag */
 144                 if ((last = regex_flags(r)) != NULL)
 145                         last = wcschr(last, 'l');
 146
 147                 /* convert to mbs */
 148                 ptr = mpdm_wcstombs((wchar_t *) v->data + offset, NULL);
 149
 150                 /* match? */
 151                 while (regexec((regex_t *) cr->data, ptr + o, 1,
 152                                &rm, offset > 0 ? REG_NOTBOL : 0) == 0) {
 153                         f++;
 154
 155                         /* if 'last' is not set, it's done */
 156                         if (last == NULL)
 157                                 break;
 158
 159                         rm.rm_so += o;
 160                         rm.rm_eo += o;
 161                         o = rm.rm_eo;
 162                 }
 163
 164                 if (f) {
 165                         /* converts to mbs the string from the beginning
 166                            to the start of the match, just to know
 167                            the size (and immediately frees it) */
 168                         free(mpdm_mbstowcs(ptr, &mpdm_regex_offset, rm.rm_so));
 169
 170                         /* add the offset */
 171                         mpdm_regex_offset += offset;
 172
 173                         /* create now the matching string */
 174                         w = MPDM_NMBS(ptr + rm.rm_so, rm.rm_eo - rm.rm_so);
 175
 176                         /* and store the size */
 177                         mpdm_regex_size = mpdm_size(w);
 178                 }
 179
 180                 free(ptr);
 181         }
 182
 183         return w;
 184 }
 185
 186
 187 /**
 188  * mpdm_regex - Matches a regular expression.
 189  * @r: the regular expression
 190  * @v: the value to be matched
 191  * @offset: offset from the start of v->data
 192  *
 193  * Matches a regular expression against a value. Valid flags are 'i',
 194  * for case-insensitive matching, 'm', to treat the string as a
 195  * multiline string (i.e., one containing newline characters), so
 196  * that ^ and $ match the boundaries of each line instead of the
 197  * whole string, 'l', to return the last matching instead of the
 198  * first one, or 'g', to match globally; in that last case, an array
 199  * containing all matches is returned instead of a string scalar.
 200  *
 201  * If @r is a string, an ordinary regular expression matching is tried
 202  * over the @v string. If the matching is possible, the match result
 203  * is returned, or NULL otherwise.
 204  *
 205  * If @r is an array (of strings), each element is tried sequentially
 206  * as an individual regular expression over the @v string, each one using
 207  * the offset returned by the previous match. All regular expressions
 208  * must match to be successful. If this is the case, an array (with
 209  * the same number of arguments) is returned containing the matched
 210  * strings, or NULL otherwise.
 211  *
 212  * If @r is NULL, the result of the previous regex matching
 213  * is returned as a two element array. The first element will contain
 214  * the character offset of the matching and the second the number of
 215  * characters matched. If the previous regex was unsuccessful, NULL
 216  * is returned.
 217  * [Regular Expressions]
 218  */
 219 mpdm_t mpdm_regex(mpdm_t r, const mpdm_t v, int offset)
 220 {
 221         mpdm_t w = NULL;
 222
 223         /* special case: if r is NULL, return previous match */
 224         if (r == NULL) {
 225                 /* if previous regex was successful... */
 226                 if (mpdm_regex_offset != -1) {
 227                         w = MPDM_A(2);
 228
 229                         mpdm_aset(w, MPDM_I(mpdm_regex_offset), 0);
 230                         mpdm_aset(w, MPDM_I(mpdm_regex_size), 1);
 231                 }
 232
 233                 return w;
 234         }
 235
 236         /* if the string to be tested is NULL, return NULL */
 237         if (v == NULL)
 238                 return NULL;
 239
 240         if (r->flags & MPDM_MULTIPLE) {
 241                 int n;
 242                 mpdm_t t;
 243
 244                 /* multiple value; try sequentially all regexes,
 245                    moving the offset forward */
 246
 247                 w = MPDM_A(mpdm_size(r));
 248
 249                 for (n = 0; n < mpdm_size(r); n++) {
 250                         t = mpdm_regex(mpdm_aget(r, n), v, offset);
 251
 252                         /* if not found, invalid all search and exit */
 253                         if (t == NULL) {
 254                                 w = NULL;
 255                                 break;
 256                         }
 257
 258                         /* found; store and move forward */
 259                         mpdm_aset(w, t, n);
 260                         offset = mpdm_regex_offset + mpdm_regex_size;
 261                 }
 262         }
 263         else {
 264                 wchar_t *global;
 265
 266                 /* takes pointer to 'global' flag */
 267                 if ((global = regex_flags(r)) != NULL)
 268                         global = wcschr(global, 'g');
 269
 270                 if (global != NULL) {
 271                         mpdm_t t;
 272
 273                         /* match sequentially until done */
 274                         w = MPDM_A(0);
 275
 276                         while ((t = regex1(r, v, offset)) != NULL) {
 277                                 mpdm_push(w, t);
 278
 279                                 offset = mpdm_regex_offset + mpdm_regex_size;
 280                         }
 281
 282                         /* no matches? convert to NULL */
 283                         if (mpdm_size(w) == 0)
 284                                 w = NULL;
 285                 }
 286                 else
 287                         w = regex1(r, v, offset);
 288         }
 289
 290         return w;
 291 }
 292
 293
 294 static mpdm_t expand_ampersands(const mpdm_t s, const mpdm_t t)
 295 /* substitutes all unescaped ampersands in s with t */
 296 {
 297         const wchar_t *sptr = mpdm_string(s);
 298         wchar_t *wptr;
 299         mpdm_t r = NULL;
 300
 301         if (s == NULL)
 302                 return s;
 303
 304         while ((wptr = wcschr(sptr, L'\\')) != NULL ||
 305                 (wptr = wcschr(sptr, L'&')) != NULL) {
 306                 int n = wptr - sptr;
 307
 308                 /* add the leading part */
 309                 r = mpdm_strcat(r, MPDM_NS(sptr, n));
 310
 311                 if (*wptr == L'\\') {
 312                         if (*(wptr + 1) == L'&' || *(wptr + 1) == L'\\')
 313                                 wptr++;
 314
 315                         r = mpdm_strcat(r, MPDM_NS(wptr, 1));
 316                 }
 317                 else
 318                 if (*wptr == '&')
 319                         r = mpdm_strcat(r, t);
 320
 321                 sptr = wptr + 1;
 322         }
 323
 324         if (r == NULL)
 325                 r = s;
 326         else {
 327                 /* add the rest of the string */
 328                 r = mpdm_strcat(r, MPDM_S(sptr));
 329         }
 330
 331         return r;
 332 }
 333
 334
 335 /**
 336  * mpdm_sregex - Matches and substitutes a regular expression.
 337  * @r: the regular expression
 338  * @v: the value to be matched
 339  * @s: the substitution string, hash or code
 340  * @offset: offset from the start of v->data
 341  *
 342  * Matches a regular expression against a value, and substitutes the
 343  * found substring with @s. Valid flags are 'i', for case-insensitive
 344  * matching, and 'g', for global replacements (all ocurrences in @v
 345  * will be replaced, instead of just the first found one).
 346  *
 347  * If @s is executable, it's executed with the matched part as
 348  * the only argument and its return value is used as the
 349  * substitution string.
 350  *
 351  * If @s is a hash, the matched string is used as a key to it and
 352  * its value used as the substitution. If this value itself is
 353  * executable, it's executed with the matched string as its only
 354  * argument and its return value used as the substitution.
 355  *
 356  * If @r is NULL, returns the number of substitutions made in the
 357  * previous call to mpdm_sregex() (can be zero if none was done).
 358  *
 359  * The global variables @mpdm_regex_offset and @mpdm_regex_size are
 360  * set to the offset of the matched string and the size of the
 361  * replaced string, respectively.
 362  *
 363  * Returns the modified string, or the original one if no substitutions
 364  * were done.
 365  * [Regular Expressions]
 366  */
 367 mpdm_t mpdm_sregex(mpdm_t r, const mpdm_t v, const mpdm_t s, int offset)
 368 {
 369         mpdm_t cr;
 370         mpdm_t o = v;
 371
 372         if (r == NULL) {
 373                 /* return last count */
 374                 return MPDM_I(mpdm_sregex_count);
 375         }
 376
 377         if (v == NULL)
 378                 return NULL;
 379
 380         /* compile the regex */
 381         if ((cr = mpdm_regcomp(r)) != NULL) {
 382                 char *ptr;
 383                 int f, i = 0;
 384                 wchar_t *global;
 385                 mpdm_t t;
 386
 387                 /* takes pointer to global flag */
 388                 if ((global = regex_flags(r)) !=NULL)
 389                         global = wcschr(global, 'g');
 390
 391                 /* store the first part */
 392                 o = MPDM_NS(v->data, offset);
 393
 394                 /* convert to mbs */
 395                 if ((ptr = mpdm_wcstombs((wchar_t *) v->data + offset, NULL)) == NULL)
 396                         return NULL;
 397
 398                 /* reset count */
 399                 mpdm_sregex_count = 0;
 400                 mpdm_regex_offset = -1;
 401
 402                 do {
 403                         regmatch_t rm;
 404
 405                         /* try match */
 406                         f = !regexec((regex_t *) cr->data, ptr + i,
 407                                      1, &rm, offset > 0 ? REG_NOTBOL : 0);
 408
 409                         if (f) {
 410                                 /* creates a string from the beginning
 411                                    to the start of the match */
 412                                 t = MPDM_NMBS(ptr + i, rm.rm_so);
 413                                 o = mpdm_strcat(o, t);
 414
 415                                 /* store offset of substitution */
 416                                 mpdm_regex_offset = mpdm_size(t) + offset;
 417
 418                                 /* get the matched part */
 419                                 t = MPDM_NMBS(ptr + i + rm.rm_so, rm.rm_eo - rm.rm_so);
 420
 421                                 /* is s an executable value? */
 422                                 if (MPDM_IS_EXEC(s)) {
 423                                         /* protect o from sweeping */
 424                                         mpdm_ref(o);
 425
 426                                         /* execute s, with t as argument */
 427                                         t = mpdm_exec_1(s, t);
 428
 429                                         mpdm_unref(o);
 430                                 }
 431                                 else
 432                                         /* is s a hash? use match as key */
 433                                 if (MPDM_IS_HASH(s)) {
 434                                         mpdm_t v = mpdm_hget(s, t);
 435
 436                                         /* is the value executable? */
 437                                         if (MPDM_IS_EXEC(v))
 438                                                 v = mpdm_exec_1(v, t);
 439
 440                                         t = v;
 441                                 }
 442                                 else
 443                                         t = expand_ampersands(s, t);
 444
 445                                 /* appends the substitution string */
 446                                 o = mpdm_strcat(o, t);
 447
 448                                 /* store size of substitution */
 449                                 mpdm_regex_size = mpdm_size(t);
 450
 451                                 i += rm.rm_eo;
 452
 453                                 /* one more substitution */
 454                                 mpdm_sregex_count++;
 455                         }
 456
 457                 } while (f && global);
 458
 459                 /* no (more) matches; convert and append the rest */
 460                 t = MPDM_MBS(ptr + i);
 461                 o = mpdm_strcat(o, t);
 462
 463                 free(ptr);
 464         }
 465
 466         return o;
 467 }