1 /* regexp.h 2 * 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, 4 * 2005, 2006, 2007, by Larry Wall and others 5 * 6 * You may distribute under the terms of either the GNU General Public 7 * License or the Artistic License, as specified in the README file. 8 * 9 */ 10 11 /* 12 * Definitions etc. for regexp(3) routines. 13 * 14 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], 15 * not the System V one. 16 */ 17 #ifndef PLUGGABLE_RE_EXTENSION 18 /* we don't want to include this stuff if we are inside of 19 an external regex engine based on the core one - like re 'debug'*/ 20 21 struct regnode { 22 U8 flags; 23 U8 type; 24 U16 next_off; 25 }; 26 27 typedef struct regnode regnode; 28 29 struct reg_substr_data; 30 31 struct reg_data; 32 33 struct regexp_engine; 34 struct regexp; 35 36 struct reg_substr_datum { 37 I32 min_offset; 38 I32 max_offset; 39 SV *substr; /* non-utf8 variant */ 40 SV *utf8_substr; /* utf8 variant */ 41 I32 end_shift; 42 }; 43 struct reg_substr_data { 44 struct reg_substr_datum data[3]; /* Actual array */ 45 }; 46 47 #ifdef PERL_OLD_COPY_ON_WRITE 48 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */ 49 #else 50 #define SV_SAVED_COPY 51 #endif 52 53 typedef struct regexp_paren_pair { 54 I32 start; 55 I32 end; 56 } regexp_paren_pair; 57 58 /* 59 The regexp/REGEXP struct, see L<perlreapi> for further documentation 60 on the individual fields. The struct is ordered so that the most 61 commonly used fields are placed at the start. 62 63 Any patch that adds items to this struct will need to include 64 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c> 65 (C<pregfree()>). This involves freeing or cloning items in the 66 regexp's data array based on the data item's type. 67 */ 68 69 typedef struct regexp { 70 /* what engine created this regexp? */ 71 const struct regexp_engine* engine; 72 struct regexp* mother_re; /* what re is this a lightweight copy of? */ 73 74 /* Information about the match that the perl core uses to manage things */ 75 U32 extflags; /* Flags used both externally and internally */ 76 I32 minlen; /* mininum possible length of string to match */ 77 I32 minlenret; /* mininum possible length of $& */ 78 U32 gofs; /* chars left of pos that we search from */ 79 struct reg_substr_data *substrs; /* substring data about strings that must appear 80 in the final match, used for optimisations */ 81 U32 nparens; /* number of capture buffers */ 82 83 /* private engine specific data */ 84 U32 intflags; /* Engine Specific Internal flags */ 85 void *pprivate; /* Data private to the regex engine which 86 created this object. */ 87 88 /* Data about the last/current match. These are modified during matching*/ 89 U32 lastparen; /* last open paren matched */ 90 U32 lastcloseparen; /* last close paren matched */ 91 regexp_paren_pair *swap; /* Swap copy of *offs */ 92 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */ 93 94 char *subbeg; /* saved or original string 95 so \digit works forever. */ 96 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */ 97 I32 sublen; /* Length of string pointed by subbeg */ 98 99 100 /* Information about the match that isn't often used */ 101 I32 prelen; /* length of precomp */ 102 const char *precomp; /* pre-compilation regular expression */ 103 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */ 104 char *wrapped; /* wrapped version of the pattern */ 105 I32 wraplen; /* length of wrapped */ 106 I32 seen_evals; /* number of eval groups in the pattern - for security checks */ 107 HV *paren_names; /* Optional hash of paren names */ 108 109 /* Refcount of this regexp */ 110 I32 refcnt; /* Refcount of this regexp */ 111 } regexp; 112 113 /* used for high speed searches */ 114 typedef struct re_scream_pos_data_s 115 { 116 char **scream_olds; /* match pos */ 117 I32 *scream_pos; /* Internal iterator of scream. */ 118 } re_scream_pos_data; 119 120 /* regexp_engine structure. This is the dispatch table for regexes. 121 * Any regex engine implementation must be able to build one of these. 122 */ 123 typedef struct regexp_engine { 124 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags); 125 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, 126 char* strbeg, I32 minend, SV* screamer, 127 void* data, U32 flags); 128 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, 129 char *strend, const U32 flags, 130 re_scream_pos_data *data); 131 SV* (*checkstr) (pTHX_ REGEXP * const rx); 132 void (*free) (pTHX_ REGEXP * const rx); 133 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, 134 SV * const sv); 135 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, 136 SV const * const value); 137 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, 138 const I32 paren); 139 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, 140 SV * const value, const U32 flags); 141 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, 142 const U32 flags); 143 SV* (*qr_package)(pTHX_ REGEXP * const rx); 144 #ifdef USE_ITHREADS 145 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); 146 #endif 147 } regexp_engine; 148 149 /* 150 These are passed to the numbered capture variable callbacks as the 151 paren name. >= 1 is reserved for actual numbered captures, i.e. $1, 152 $2 etc. 153 */ 154 #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ 155 #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ 156 #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ 157 158 /* 159 Flags that are passed to the named_buff and named_buff_iter 160 callbacks above. Those routines are called from universal.c via the 161 Tie::Hash::NamedCapture interface for %+ and %- and the re:: 162 functions in the same file. 163 */ 164 165 /* The Tie::Hash::NamedCapture operation this is part of, if any */ 166 #define RXapif_FETCH 0x0001 167 #define RXapif_STORE 0x0002 168 #define RXapif_DELETE 0x0004 169 #define RXapif_CLEAR 0x0008 170 #define RXapif_EXISTS 0x0010 171 #define RXapif_SCALAR 0x0020 172 #define RXapif_FIRSTKEY 0x0040 173 #define RXapif_NEXTKEY 0x0080 174 175 /* Whether %+ or %- is being operated on */ 176 #define RXapif_ONE 0x0100 /* %+ */ 177 #define RXapif_ALL 0x0200 /* %- */ 178 179 /* Whether this is being called from a re:: function */ 180 #define RXapif_REGNAME 0x0400 181 #define RXapif_REGNAMES 0x0800 182 #define RXapif_REGNAMES_COUNT 0x1000 183 184 /* 185 =head1 REGEXP Functions 186 187 =for apidoc Am|REGEXP *|SvRX|SV *sv 188 189 Convenience macro to get the REGEXP from a SV. This is approximately 190 equivalent to the following snippet: 191 192 if (SvMAGICAL(sv)) 193 mg_get(sv); 194 if (SvROK(sv) && 195 (tmpsv = (SV*)SvRV(sv)) && 196 SvTYPE(tmpsv) == SVt_PVMG && 197 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr))) 198 { 199 return (REGEXP *)tmpmg->mg_obj; 200 } 201 202 NULL will be returned if a REGEXP* is not found. 203 204 =for apidoc Am|bool|SvRXOK|SV* sv 205 206 Returns a boolean indicating whether the SV contains qr magic 207 (PERL_MAGIC_qr). 208 209 If you want to do something with the REGEXP* later use SvRX instead 210 and check for NULL. 211 212 =cut 213 */ 214 215 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) 216 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) 217 218 219 /* Flags stored in regexp->extflags 220 * These are used by code external to the regexp engine 221 * 222 * Note that flags starting with RXf_PMf_ have exact equivalents 223 * stored in op_pmflags and which are defined in op.h, they are defined 224 * numerically here only for clarity. 225 * 226 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 227 * so that regnodes.h is updated with the changes. 228 * 229 */ 230 231 /* Anchor and GPOS related stuff */ 232 #define RXf_ANCH_BOL 0x00000001 233 #define RXf_ANCH_MBOL 0x00000002 234 #define RXf_ANCH_SBOL 0x00000004 235 #define RXf_ANCH_GPOS 0x00000008 236 #define RXf_GPOS_SEEN 0x00000010 237 #define RXf_GPOS_FLOAT 0x00000020 238 /* two bits here */ 239 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) 240 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) 241 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) 242 243 /* Flags indicating special patterns */ 244 #define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */ 245 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ 246 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ 247 #define RXf_NULL 0x40000000 /* Pattern is // */ 248 249 /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */ 250 #define RXf_PMf_LOCALE 0x00000800 /* use locale */ 251 #define RXf_PMf_MULTILINE 0x00001000 /* /m */ 252 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ 253 #define RXf_PMf_FOLD 0x00004000 /* /i */ 254 #define RXf_PMf_EXTENDED 0x00008000 /* /x */ 255 #define RXf_PMf_KEEPCOPY 0x00010000 /* /p */ 256 /* these flags are transfered from the PMOP->op_pmflags member during compilation */ 257 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) 258 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) 259 260 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ 261 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ 262 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ 263 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ 264 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break 265 266 /* chars and strings used as regex pattern modifiers 267 * Singlular is a 'c'har, plural is a "string" 268 * 269 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve 270 * for compatibility reasons with Regexp::Common which highjacked (?k:...) 271 * for its own uses. So 'k' is out as well. 272 */ 273 #define EXEC_PAT_MOD 'e' 274 #define KEEPCOPY_PAT_MOD 'p' 275 #define ONCE_PAT_MOD 'o' 276 #define GLOBAL_PAT_MOD 'g' 277 #define CONTINUE_PAT_MOD 'c' 278 #define MULTILINE_PAT_MOD 'm' 279 #define SINGLE_PAT_MOD 's' 280 #define IGNORE_PAT_MOD 'i' 281 #define XTENDED_PAT_MOD 'x' 282 283 #define ONCE_PAT_MODS "o" 284 #define KEEPCOPY_PAT_MODS "p" 285 #define EXEC_PAT_MODS "e" 286 #define LOOP_PAT_MODS "gc" 287 288 #define STD_PAT_MODS "msix" 289 290 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS 291 292 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS 293 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS 294 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS 295 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS 296 297 /* 298 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 299 * so that regnodes.h is updated with the changes. 300 * 301 */ 302 303 /* What we have seen */ 304 #define RXf_LOOKBEHIND_SEEN 0x00020000 305 #define RXf_EVAL_SEEN 0x00040000 306 #define RXf_CANY_SEEN 0x00080000 307 308 /* Special */ 309 #define RXf_NOSCAN 0x00100000 310 #define RXf_CHECK_ALL 0x00200000 311 312 /* UTF8 related */ 313 #define RXf_UTF8 0x00400000 314 #define RXf_MATCH_UTF8 0x00800000 315 316 /* Intuit related */ 317 #define RXf_USE_INTUIT_NOML 0x01000000 318 #define RXf_USE_INTUIT_ML 0x02000000 319 #define RXf_INTUIT_TAIL 0x04000000 320 321 /* 322 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will 323 be used by regex engines to check whether they should set 324 RXf_SKIPWHITE 325 */ 326 #define RXf_SPLIT 0x08000000 327 328 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) 329 330 /* Copy and tainted info */ 331 #define RXf_COPY_DONE 0x10000000 332 #define RXf_TAINTED_SEEN 0x20000000 333 #define RXf_TAINTED 0x80000000 /* this pattern is tainted */ 334 335 /* 336 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 337 * so that regnodes.h is updated with the changes. 338 * 339 */ 340 341 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) 342 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN) 343 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN) 344 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN) 345 #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 346 ? RX_MATCH_TAINTED_on(prog) \ 347 : RX_MATCH_TAINTED_off(prog)) 348 349 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE) 350 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE) 351 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE) 352 #define RX_MATCH_COPIED_set(prog,t) ((t) \ 353 ? RX_MATCH_COPIED_on(prog) \ 354 : RX_MATCH_COPIED_off(prog)) 355 356 #endif /* PLUGGABLE_RE_EXTENSION */ 357 358 /* Stuff that needs to be included in the plugable extension goes below here */ 359 360 #ifdef PERL_OLD_COPY_ON_WRITE 361 #define RX_MATCH_COPY_FREE(rx) \ 362 STMT_START {if (rx->saved_copy) { \ 363 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ 364 } \ 365 if (RX_MATCH_COPIED(rx)) { \ 366 Safefree(rx->subbeg); \ 367 RX_MATCH_COPIED_off(rx); \ 368 }} STMT_END 369 #else 370 #define RX_MATCH_COPY_FREE(rx) \ 371 STMT_START {if (RX_MATCH_COPIED(rx)) { \ 372 Safefree(rx->subbeg); \ 373 RX_MATCH_COPIED_off(rx); \ 374 }} STMT_END 375 #endif 376 377 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8) 378 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8) 379 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8) 380 #define RX_MATCH_UTF8_set(prog, t) ((t) \ 381 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 382 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 383 384 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 385 #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 386 #define REXEC_SCREAM 0x04 /* use scream table. */ 387 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 388 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 389 390 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) 391 #define ReREFCNT_dec(re) CALLREGFREE(re) 392 393 #define FBMcf_TAIL_DOLLAR 1 394 #define FBMcf_TAIL_DOLLARM 2 395 #define FBMcf_TAIL_Z 4 396 #define FBMcf_TAIL_z 8 397 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 398 399 #define FBMrf_MULTILINE 1 400 401 /* an accepting state/position*/ 402 struct _reg_trie_accepted { 403 U8 *endpos; 404 U16 wordnum; 405 }; 406 typedef struct _reg_trie_accepted reg_trie_accepted; 407 408 /* some basic information about the current match that is created by 409 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ 410 411 typedef struct { 412 regexp *prog; 413 char *bol; 414 char *till; 415 SV *sv; 416 char *ganch; 417 char *cutpoint; 418 } regmatch_info; 419 420 421 /* structures for holding and saving the state maintained by regmatch() */ 422 423 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH 424 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 425 #endif 426 427 typedef I32 CHECKPOINT; 428 429 typedef struct regmatch_state { 430 int resume_state; /* where to jump to on return */ 431 char *locinput; /* where to backtrack in string on failure */ 432 433 union { 434 435 /* this is a fake union member that matches the first element 436 * of each member that needs to store positive backtrack 437 * information */ 438 struct { 439 struct regmatch_state *prev_yes_state; 440 } yes; 441 442 /* branchlike members */ 443 /* this is a fake union member that matches the first elements 444 * of each member that needs to behave like a branch */ 445 struct { 446 /* this first element must match u.yes */ 447 struct regmatch_state *prev_yes_state; 448 U32 lastparen; 449 CHECKPOINT cp; 450 451 } branchlike; 452 453 struct { 454 /* the first elements must match u.branchlike */ 455 struct regmatch_state *prev_yes_state; 456 U32 lastparen; 457 CHECKPOINT cp; 458 459 regnode *next_branch; /* next branch node */ 460 } branch; 461 462 struct { 463 /* the first elements must match u.branchlike */ 464 struct regmatch_state *prev_yes_state; 465 U32 lastparen; 466 CHECKPOINT cp; 467 468 reg_trie_accepted *accept_buff; /* accepting states we have seen */ 469 U32 accepted; /* how many accepting states we have seen */ 470 U16 *jump; /* positive offsets from me */ 471 regnode *B; /* node following the trie */ 472 regnode *me; /* Which node am I - needed for jump tries*/ 473 } trie; 474 475 /* special types - these members are used to store state for special 476 regops like eval, if/then, lookaround and the markpoint state */ 477 struct { 478 /* this first element must match u.yes */ 479 struct regmatch_state *prev_yes_state; 480 struct regmatch_state *prev_eval; 481 struct regmatch_state *prev_curlyx; 482 regexp *prev_rex; 483 U32 toggle_reg_flags; /* what bits in PL_reg_flags to 484 flip when transitioning between 485 inner and outer rexen */ 486 CHECKPOINT cp; /* remember current savestack indexes */ 487 CHECKPOINT lastcp; 488 U32 close_paren; /* which close bracket is our end */ 489 regnode *B; /* the node following us */ 490 } eval; 491 492 struct { 493 /* this first element must match u.yes */ 494 struct regmatch_state *prev_yes_state; 495 I32 wanted; 496 I32 logical; /* saved copy of 'logical' var */ 497 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ 498 } ifmatch; /* and SUSPEND/UNLESSM */ 499 500 struct { 501 /* this first element must match u.yes */ 502 struct regmatch_state *prev_yes_state; 503 struct regmatch_state *prev_mark; 504 SV* mark_name; 505 char *mark_loc; 506 } mark; 507 508 struct { 509 int val; 510 } keeper; 511 512 /* quantifiers - these members are used for storing state for 513 for the regops used to implement quantifiers */ 514 struct { 515 /* this first element must match u.yes */ 516 struct regmatch_state *prev_yes_state; 517 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ 518 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 519 CHECKPOINT cp; /* remember current savestack index */ 520 bool minmod; 521 int parenfloor;/* how far back to strip paren data */ 522 int min; /* the minimal number of A's to match */ 523 int max; /* the maximal number of A's to match */ 524 525 /* these two are modified by WHILEM */ 526 int count; /* how many instances of A we've matched */ 527 char *lastloc;/* where previous A matched (0-len detect) */ 528 } curlyx; 529 530 struct { 531 /* this first element must match u.yes */ 532 struct regmatch_state *prev_yes_state; 533 struct regmatch_state *save_curlyx; 534 CHECKPOINT cp; /* remember current savestack indexes */ 535 CHECKPOINT lastcp; 536 char *save_lastloc; /* previous curlyx.lastloc */ 537 I32 cache_offset; 538 I32 cache_mask; 539 } whilem; 540 541 struct { 542 /* this first element must match u.yes */ 543 struct regmatch_state *prev_yes_state; 544 I32 c1, c2; /* case fold search */ 545 CHECKPOINT cp; 546 I32 alen; /* length of first-matched A string */ 547 I32 count; 548 bool minmod; 549 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 550 regnode *me; /* the curlym node */ 551 } curlym; 552 553 struct { 554 U32 paren; 555 CHECKPOINT cp; 556 I32 c1, c2; /* case fold search */ 557 char *maxpos; /* highest possible point in string to match */ 558 char *oldloc; /* the previous locinput */ 559 int count; 560 int min, max; /* {m,n} */ 561 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 562 } curly; /* and CURLYN/PLUS/STAR */ 563 564 } u; 565 } regmatch_state; 566 567 /* how many regmatch_state structs to allocate as a single slab. 568 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev 569 * pointers, plus 1 for any mythical malloc overhead. */ 570 571 #define PERL_REGMATCH_SLAB_SLOTS \ 572 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) 573 574 typedef struct regmatch_slab { 575 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; 576 struct regmatch_slab *prev, *next; 577 } regmatch_slab; 578 579 #define PL_reg_flags PL_reg_state.re_state_reg_flags 580 #define PL_bostr PL_reg_state.re_state_bostr 581 #define PL_reginput PL_reg_state.re_state_reginput 582 #define PL_regeol PL_reg_state.re_state_regeol 583 #define PL_regoffs PL_reg_state.re_state_regoffs 584 #define PL_reglastparen PL_reg_state.re_state_reglastparen 585 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen 586 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp 587 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl 588 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set 589 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 590 #define PL_reg_magic PL_reg_state.re_state_reg_magic 591 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos 592 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm 593 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm 594 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved 595 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen 596 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter 597 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter 598 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache 599 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size 600 #define PL_regsize PL_reg_state.re_state_regsize 601 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry 602 #define PL_nrs PL_reg_state.re_state_nrs 603 604 struct re_save_state { 605 U32 re_state_reg_flags; /* from regexec.c */ 606 U32 re_state_reg_start_tmpl; /* from regexec.c */ 607 I32 re_state_reg_eval_set; /* from regexec.c */ 608 bool re_state_reg_match_utf8; /* from regexec.c */ 609 char *re_state_bostr; 610 char *re_state_reginput; /* String-input pointer. */ 611 char *re_state_regeol; /* End of input, for $ check. */ 612 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ 613 U32 *re_state_reglastparen; /* Similarly for lastparen. */ 614 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ 615 char **re_state_reg_start_tmp; /* from regexec.c */ 616 MAGIC *re_state_reg_magic; /* from regexec.c */ 617 PMOP *re_state_reg_oldcurpm; /* from regexec.c */ 618 PMOP *re_state_reg_curpm; /* from regexec.c */ 619 char *re_state_reg_oldsaved; /* old saved substr during match */ 620 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ 621 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ 622 I32 re_state_reg_oldpos; /* from regexec.c */ 623 I32 re_state_reg_maxiter; /* max wait until caching pos */ 624 I32 re_state_reg_leftiter; /* wait until caching pos */ 625 U32 re_state_regsize; /* from regexec.c */ 626 char *re_state_reg_poscache; /* cache of pos of WHILEM */ 627 char *re_state_reg_starttry; /* from regexec.c */ 628 #ifdef PERL_OLD_COPY_ON_WRITE 629 SV *re_state_nrs; /* was placeholder: unused since 5.8.0 (5.7.2 patch #12027 for bug ID 20010815.012). Used to save rx->saved_copy */ 630 #endif 631 }; 632 633 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ 634 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) 635 636 /* 637 * Local variables: 638 * c-indentation-style: bsd 639 * c-basic-offset: 4 640 * indent-tabs-mode: t 641 * End: 642 * 643 * ex: set ts=8 sts=4 sw=4 noet: 644 */ 645