mpdm_r.c

   1 /*
   2
   3     MPDM - Minimum Profit Data Manager
   4     Copyright (C) 2003/2010 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 /* matching of the last regex */
  51
  52 int mpdm_regex_offset = -1;
  53 int mpdm_regex_size = 0;
  54
  55 /* number of substitutions in last sregex */
  56
  57 int mpdm_sregex_count = 0;
  58
  59
  60 /** code **/
  61
  62 static wchar_t *regex_flags(const mpdm_t r)
  63 {
  64         return wcsrchr((wchar_t *) r->data, *(wchar_t *) r->data);
  65 }
  66
  67
  68 static mpdm_t mpdm_regcomp(mpdm_t r)
  69 {
  70         mpdm_t c = NULL;
  71         mpdm_t regex_cache = NULL;
  72
  73         /* if cache does not exist, create it */
  74         if ((regex_cache = mpdm_hget_s(mpdm_root(), L"__REGEX_CACHE__")) == NULL) {
  75                 regex_cache = MPDM_H(0);
  76                 mpdm_hset_s(mpdm_root(), L"__REGEX_CACHE__", regex_cache);
  77         }
  78
  79         /* search the regex in the cache */
  80         if ((c = mpdm_hget(regex_cache, r)) == NULL) {
  81                 mpdm_t rmb;
  82                 regex_t re;
  83                 char *regex;
  84                 char *flags;
  85                 int f = REG_EXTENDED;
  86
  87                 /* not found; regex must be compiled */
  88
  89                 /* convert to mbs */
  90                 rmb = mpdm_ref(MPDM_2MBS(r->data));
  91                 regex = (char *) rmb->data;
  92
  93                 if ((flags = strrchr(regex, *regex)) != NULL) {
  94
  95                 if (strchr(flags, 'i') != NULL)
  96                         f |= REG_ICASE;
  97                 if (strchr(flags, 'm') != NULL)
  98                         f |= REG_NEWLINE;
  99
 100                 regex++;
 101                 *flags = '\0';
 102
 103                 if (!regcomp(&re, regex, f)) {
 104                         void *ptr;
 105
 106                         if ((ptr = malloc(sizeof(regex_t))) != NULL) {
 107                                 /* copies */
 108                                 memcpy(ptr, &re, sizeof(regex_t));
 109
 110                                 /* create value */
 111                                 c = mpdm_new(MPDM_FREE, ptr, sizeof(regex_t));
 112
 113                                 /* stores */
 114                                 mpdm_hset(regex_cache, r, c);
 115                         }
 116                 }
 117         }
 118
 119         mpdm_unref(rmb);
 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(0);
 248
 249                 for (n = 0; n < mpdm_size(r); n++) {
 250                         t = mpdm_regex(mpdm_aget(r, n), v, offset);
 251
 252                         if (t == NULL)
 253                                 break;
 254
 255                         /* found; store and move forward */
 256                         mpdm_push(w, t);
 257                         offset = mpdm_regex_offset + mpdm_regex_size;
 258                 }
 259         }
 260         else {
 261                 wchar_t *global;
 262
 263                 /* takes pointer to 'global' flag */
 264                 if ((global = regex_flags(r)) != NULL)
 265                         global = wcschr(global, 'g');
 266
 267                 if (global != NULL) {
 268                         mpdm_t t;
 269
 270                         /* match sequentially until done */
 271                         w = MPDM_A(0);
 272
 273                         while ((t = regex1(r, v, offset)) != NULL) {
 274                                 mpdm_push(w, t);
 275
 276                                 offset = mpdm_regex_offset + mpdm_regex_size;
 277                         }
 278                 }
 279                 else
 280                         w = regex1(r, v, offset);
 281         }
 282
 283         return w;
 284 }
 285
 286
 287 static mpdm_t expand_ampersands(const mpdm_t s, const mpdm_t t)
 288 /* substitutes all unescaped ampersands in s with t */
 289 {
 290         const wchar_t *sptr = mpdm_string(s);
 291         wchar_t *wptr;
 292     wchar_t *optr = NULL;
 293     int osize = 0;
 294         mpdm_t r = NULL;
 295
 296         if (s == NULL)
 297                 return s;
 298
 299         while ((wptr = wcschr(sptr, L'\\')) != NULL ||
 300                 (wptr = wcschr(sptr, L'&')) != NULL) {
 301                 int n = wptr - sptr;
 302
 303                 /* add the leading part */
 304         optr = mpdm_pokewsn(optr, &osize, sptr, n);
 305
 306                 if (*wptr == L'\\') {
 307                         if (*(wptr + 1) == L'&' || *(wptr + 1) == L'\\')
 308                                 wptr++;
 309
 310             optr = mpdm_pokewsn(optr, &osize, wptr, 1);
 311                 }
 312                 else
 313                 if (*wptr == '&')
 314             optr = mpdm_pokev(optr, &osize, t);
 315
 316                 sptr = wptr + 1;
 317         }
 318
 319     /* add the rest of the string */
 320     optr = mpdm_pokews(optr, &osize, sptr);
 321     optr = mpdm_pokewsn(optr, &osize, L"", 1);
 322     r = MPDM_ENS(optr, osize - 1);
 323
 324         return r;
 325 }
 326
 327
 328 /**
 329  * mpdm_sregex - Matches and substitutes a regular expression.
 330  * @r: the regular expression
 331  * @v: the value to be matched
 332  * @s: the substitution string, hash or code
 333  * @offset: offset from the start of v->data
 334  *
 335  * Matches a regular expression against a value, and substitutes the
 336  * found substring with @s. Valid flags are 'i', for case-insensitive
 337  * matching, and 'g', for global replacements (all ocurrences in @v
 338  * will be replaced, instead of just the first found one).
 339  *
 340  * If @s is executable, it's executed with the matched part as
 341  * the only argument and its return value is used as the
 342  * substitution string.
 343  *
 344  * If @s is a hash, the matched string is used as a key to it and
 345  * its value used as the substitution. If this value itself is
 346  * executable, it's executed with the matched string as its only
 347  * argument and its return value used as the substitution.
 348  *
 349  * If @r is NULL, returns the number of substitutions made in the
 350  * previous call to mpdm_sregex() (can be zero if none was done).
 351  *
 352  * The global variables @mpdm_regex_offset and @mpdm_regex_size are
 353  * set to the offset of the matched string and the size of the
 354  * replaced string, respectively.
 355  *
 356  * Always returns a new string (either modified or an exact copy).
 357  * [Regular Expressions]
 358  */
 359 mpdm_t mpdm_sregex(mpdm_t r, const mpdm_t v, const mpdm_t s, int offset)
 360 {
 361         mpdm_t cr;
 362     wchar_t *optr = NULL;
 363     int osize = 0;
 364
 365         if (r == NULL) {
 366                 /* return last count */
 367                 return MPDM_I(mpdm_sregex_count);
 368         }
 369
 370         if (v == NULL)
 371                 return NULL;
 372
 373         /* compile the regex */
 374         if ((cr = mpdm_regcomp(r)) != NULL) {
 375                 char *ptr;
 376                 int f, i = 0;
 377                 wchar_t *global;
 378                 mpdm_t t;
 379
 380                 /* takes pointer to global flag */
 381                 if ((global = regex_flags(r)) !=NULL)
 382                         global = wcschr(global, 'g');
 383
 384                 /* store the first part */
 385         optr = mpdm_pokewsn(optr, &osize, v->data, offset);
 386
 387                 /* convert to mbs */
 388                 if ((ptr = mpdm_wcstombs((wchar_t *) v->data + offset, NULL)) == NULL)
 389                         return NULL;
 390
 391                 /* reset count */
 392                 mpdm_sregex_count = 0;
 393                 mpdm_regex_offset = -1;
 394
 395                 do {
 396                         regmatch_t rm;
 397
 398                         /* try match */
 399                         f = !regexec((regex_t *) cr->data, ptr + i,
 400                                      1, &rm, offset > 0 ? REG_NOTBOL : 0);
 401
 402                         if (f) {
 403                                 /* creates a string from the beginning
 404                                    to the start of the match */
 405                                 t = mpdm_ref(MPDM_NMBS(ptr + i, rm.rm_so));
 406                 optr = mpdm_pokev(optr, &osize, t);
 407
 408                                 /* store offset of substitution */
 409                                 mpdm_regex_offset = mpdm_size(t) + offset;
 410
 411                 mpdm_unref(t);
 412
 413                                 /* get the matched part */
 414                                 t = MPDM_NMBS(ptr + i + rm.rm_so, rm.rm_eo - rm.rm_so);
 415
 416                                 /* is s an executable value? */
 417                                 if (MPDM_IS_EXEC(s)) {
 418                     mpdm_t v = mpdm_ref(t);
 419
 420                                         /* execute s, with t as argument */
 421                                         t = mpdm_exec_1(s, v);
 422
 423                                         mpdm_unref(v);
 424                                 }
 425                                 else
 426                                         /* is s a hash? use match as key */
 427                                 if (MPDM_IS_HASH(s)) {
 428                                         mpdm_t v = mpdm_hget(s, t);
 429
 430                     mpdm_ref(t);
 431
 432                                         /* is the value executable? */
 433                                         if (MPDM_IS_EXEC(v)) {
 434                         mpdm_t w = mpdm_ref(v);
 435
 436                                                 v = mpdm_exec_1(w, t);
 437
 438                         mpdm_unref(w);
 439                     }
 440
 441                     mpdm_unref(t);
 442
 443                                         t = v;
 444                                 }
 445                                 else {
 446                     mpdm_t v = mpdm_ref(t);
 447                                         t = expand_ampersands(s, v);
 448                     mpdm_unref(v);
 449                 }
 450
 451                                 /* appends the substitution string */
 452                 mpdm_ref(t);
 453
 454                 optr = mpdm_pokev(optr, &osize, t);
 455
 456                                 /* store size of substitution */
 457                                 mpdm_regex_size = mpdm_size(t);
 458
 459                 mpdm_unref(t);
 460
 461                                 i += rm.rm_eo;
 462
 463                                 /* one more substitution */
 464                                 mpdm_sregex_count++;
 465                         }
 466
 467                 } while (f && global);
 468
 469                 /* no (more) matches; convert and append the rest */
 470                 t = mpdm_ref(MPDM_MBS(ptr + i));
 471         optr = mpdm_pokev(optr, &osize, t);
 472         mpdm_unref(t);
 473
 474                 free(ptr);
 475         }
 476
 477     /* NULL-terminate */
 478     optr = mpdm_pokewsn(optr, &osize, L"", 1);
 479
 480     return MPDM_ENS(optr, osize - 1);
 481 }