1 /* regexp.h 2 * 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, 4 * 2005, 2006, 2007, 2008 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 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names) 114 115 /* used for high speed searches */ 116 typedef struct re_scream_pos_data_s 117 { 118 char **scream_olds; /* match pos */ 119 I32 *scream_pos; /* Internal iterator of scream. */ 120 } re_scream_pos_data; 121 122 /* regexp_engine structure. This is the dispatch table for regexes. 123 * Any regex engine implementation must be able to build one of these. 124 */ 125 typedef struct regexp_engine { 126 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags); 127 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, 128 char* strbeg, I32 minend, SV* screamer, 129 void* data, U32 flags); 130 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, 131 char *strend, const U32 flags, 132 re_scream_pos_data *data); 133 SV* (*checkstr) (pTHX_ REGEXP * const rx); 134 void (*free) (pTHX_ REGEXP * const rx); 135 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, 136 SV * const sv); 137 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, 138 SV const * const value); 139 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, 140 const I32 paren); 141 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, 142 SV * const value, const U32 flags); 143 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, 144 const U32 flags); 145 SV* (*qr_package)(pTHX_ REGEXP * const rx); 146 #ifdef USE_ITHREADS 147 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); 148 #endif 149 } regexp_engine; 150 151 /* 152 These are passed to the numbered capture variable callbacks as the 153 paren name. >= 1 is reserved for actual numbered captures, i.e. $1, 154 $2 etc. 155 */ 156 #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ 157 #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ 158 #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ 159 160 /* 161 Flags that are passed to the named_buff and named_buff_iter 162 callbacks above. Those routines are called from universal.c via the 163 Tie::Hash::NamedCapture interface for %+ and %- and the re:: 164 functions in the same file. 165 */ 166 167 /* The Tie::Hash::NamedCapture operation this is part of, if any */ 168 #define RXapif_FETCH 0x0001 169 #define RXapif_STORE 0x0002 170 #define RXapif_DELETE 0x0004 171 #define RXapif_CLEAR 0x0008 172 #define RXapif_EXISTS 0x0010 173 #define RXapif_SCALAR 0x0020 174 #define RXapif_FIRSTKEY 0x0040 175 #define RXapif_NEXTKEY 0x0080 176 177 /* Whether %+ or %- is being operated on */ 178 #define RXapif_ONE 0x0100 /* %+ */ 179 #define RXapif_ALL 0x0200 /* %- */ 180 181 /* Whether this is being called from a re:: function */ 182 #define RXapif_REGNAME 0x0400 183 #define RXapif_REGNAMES 0x0800 184 #define RXapif_REGNAMES_COUNT 0x1000 185 186 /* 187 =head1 REGEXP Functions 188 189 =for apidoc Am|REGEXP *|SvRX|SV *sv 190 191 Convenience macro to get the REGEXP from a SV. This is approximately 192 equivalent to the following snippet: 193 194 if (SvMAGICAL(sv)) 195 mg_get(sv); 196 if (SvROK(sv) && 197 (tmpsv = (SV*)SvRV(sv)) && 198 SvTYPE(tmpsv) == SVt_PVMG && 199 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr))) 200 { 201 return (REGEXP *)tmpmg->mg_obj; 202 } 203 204 NULL will be returned if a REGEXP* is not found. 205 206 =for apidoc Am|bool|SvRXOK|SV* sv 207 208 Returns a boolean indicating whether the SV contains qr magic 209 (PERL_MAGIC_qr). 210 211 If you want to do something with the REGEXP* later use SvRX instead 212 and check for NULL. 213 214 =cut 215 */ 216 217 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) 218 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) 219 220 221 /* Flags stored in regexp->extflags 222 * These are used by code external to the regexp engine 223 * 224 * Note that flags starting with RXf_PMf_ have exact equivalents 225 * stored in op_pmflags and which are defined in op.h, they are defined 226 * numerically here only for clarity. 227 * 228 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 229 * so that regnodes.h is updated with the changes. 230 * 231 */ 232 233 /* Anchor and GPOS related stuff */ 234 #define RXf_ANCH_BOL 0x00000001 235 #define RXf_ANCH_MBOL 0x00000002 236 #define RXf_ANCH_SBOL 0x00000004 237 #define RXf_ANCH_GPOS 0x00000008 238 #define RXf_GPOS_SEEN 0x00000010 239 #define RXf_GPOS_FLOAT 0x00000020 240 /* two bits here */ 241 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) 242 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) 243 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) 244 245 /* Flags indicating special patterns */ 246 #define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */ 247 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ 248 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ 249 #define RXf_NULL 0x40000000 /* Pattern is // */ 250 251 /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME 252 * If you change these you need to change the equivalent flags in op.h, and 253 * vice versa. */ 254 #define RXf_PMf_LOCALE 0x00000800 /* use locale */ 255 #define RXf_PMf_MULTILINE 0x00001000 /* /m */ 256 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ 257 #define RXf_PMf_FOLD 0x00004000 /* /i */ 258 #define RXf_PMf_EXTENDED 0x00008000 /* /x */ 259 #define RXf_PMf_KEEPCOPY 0x00010000 /* /p */ 260 /* these flags are transfered from the PMOP->op_pmflags member during compilation */ 261 #define RXf_PMf_STD_PMMOD_SHIFT 12 262 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) 263 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) 264 265 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ 266 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ 267 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ 268 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ 269 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break 270 271 /* chars and strings used as regex pattern modifiers 272 * Singlular is a 'c'har, plural is a "string" 273 * 274 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve 275 * for compatibility reasons with Regexp::Common which highjacked (?k:...) 276 * for its own uses. So 'k' is out as well. 277 */ 278 #define EXEC_PAT_MOD 'e' 279 #define KEEPCOPY_PAT_MOD 'p' 280 #define ONCE_PAT_MOD 'o' 281 #define GLOBAL_PAT_MOD 'g' 282 #define CONTINUE_PAT_MOD 'c' 283 #define MULTILINE_PAT_MOD 'm' 284 #define SINGLE_PAT_MOD 's' 285 #define IGNORE_PAT_MOD 'i' 286 #define XTENDED_PAT_MOD 'x' 287 288 #define ONCE_PAT_MODS "o" 289 #define KEEPCOPY_PAT_MODS "p" 290 #define EXEC_PAT_MODS "e" 291 #define LOOP_PAT_MODS "gc" 292 293 #define STD_PAT_MODS "msix" 294 295 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS 296 297 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS 298 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS 299 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS 300 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS 301 302 /* 303 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 304 * so that regnodes.h is updated with the changes. 305 * 306 */ 307 308 /* What we have seen */ 309 #define RXf_LOOKBEHIND_SEEN 0x00020000 310 #define RXf_EVAL_SEEN 0x00040000 311 #define RXf_CANY_SEEN 0x00080000 312 313 /* Special */ 314 #define RXf_NOSCAN 0x00100000 315 #define RXf_CHECK_ALL 0x00200000 316 317 /* UTF8 related */ 318 #define RXf_UTF8 0x00400000 319 #define RXf_MATCH_UTF8 0x00800000 320 321 /* Intuit related */ 322 #define RXf_USE_INTUIT_NOML 0x01000000 323 #define RXf_USE_INTUIT_ML 0x02000000 324 #define RXf_INTUIT_TAIL 0x04000000 325 326 /* 327 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will 328 be used by regex engines to check whether they should set 329 RXf_SKIPWHITE 330 */ 331 #define RXf_SPLIT 0x08000000 332 333 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) 334 335 /* Copy and tainted info */ 336 #define RXf_COPY_DONE 0x10000000 337 #define RXf_TAINTED_SEEN 0x20000000 338 #define RXf_TAINTED 0x80000000 /* this pattern is tainted */ 339 340 /* 341 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 342 * so that regnodes.h is updated with the changes. 343 * 344 */ 345 346 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) 347 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 348 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 349 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) 350 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) 351 #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 352 ? RX_MATCH_TAINTED_on(prog) \ 353 : RX_MATCH_TAINTED_off(prog)) 354 355 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) 356 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) 357 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) 358 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) 359 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 360 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 361 #define RX_MATCH_COPIED_set(prog,t) ((t) \ 362 ? RX_MATCH_COPIED_on(prog) \ 363 : RX_MATCH_COPIED_off(prog)) 364 365 #define RXp_EXTFLAGS(rx) ((rx)->extflags) 366 367 #define RX_PRECOMP(prog) ((prog)->precomp) 368 /* *** 5.10.x-specific definition of RX_PRECOMP_const */ 369 #define RX_PRECOMP_const(prog) ((prog)->precomp) 370 #define RX_PRELEN(prog) ((prog)->prelen) 371 #define RX_WRAPPED(prog) ((prog)->wrapped) 372 #define RX_WRAPLEN(prog) ((prog)->wraplen) 373 #define RX_CHECK_SUBSTR(prog) ((prog)->check_substr) 374 #define RX_EXTFLAGS(prog) ((prog)->extflags) 375 #define RX_REFCNT(prog) ((prog)->refcnt) 376 #define RX_ENGINE(prog) ((prog)->engine) 377 #define RX_SUBBEG(prog) ((prog)->subbeg) 378 #define RX_OFFS(prog) ((prog)->offs) 379 #define RX_NPARENS(prog) ((prog)->nparens) 380 #define RX_SUBLEN(prog) ((prog)->sublen) 381 #define RX_SUBBEG(prog) ((prog)->subbeg) 382 #define RX_MINLEN(prog) ((prog)->minlen) 383 #define RX_MINLENRET(prog) ((prog)->minlenret) 384 #define RX_GOFS(prog) ((prog)->gofs) 385 #define RX_LASTPAREN(prog) ((prog)->lastparen) 386 #define RX_LASTCLOSEPAREN(prog) ((prog)->lastcloseparen) 387 #define RX_SEEN_EVALS(prog) ((prog)->seen_evals) 388 389 #endif /* PLUGGABLE_RE_EXTENSION */ 390 391 /* Stuff that needs to be included in the plugable extension goes below here */ 392 393 #ifdef PERL_OLD_COPY_ON_WRITE 394 #define RX_MATCH_COPY_FREE(rx) \ 395 STMT_START {if (rx->saved_copy) { \ 396 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ 397 } \ 398 if (RX_MATCH_COPIED(rx)) { \ 399 Safefree(RX_SUBBEG(rx)); \ 400 RX_MATCH_COPIED_off(rx); \ 401 }} STMT_END 402 #else 403 #define RX_MATCH_COPY_FREE(rx) \ 404 STMT_START {if (RX_MATCH_COPIED(rx)) { \ 405 Safefree(RX_SUBBEG(rx)); \ 406 RX_MATCH_COPIED_off(rx); \ 407 }} STMT_END 408 #endif 409 410 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) 411 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) 412 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) 413 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) 414 #define RX_MATCH_UTF8_set(prog, t) ((t) \ 415 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 416 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 417 418 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ 419 #define RX_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_UTF8) 420 421 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 422 #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 423 #define REXEC_SCREAM 0x04 /* use scream table. */ 424 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 425 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 426 427 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) 428 #define ReREFCNT_dec(re) CALLREGFREE(re) 429 430 #define FBMcf_TAIL_DOLLAR 1 431 #define FBMcf_TAIL_DOLLARM 2 432 #define FBMcf_TAIL_Z 4 433 #define FBMcf_TAIL_z 8 434 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 435 436 #define FBMrf_MULTILINE 1 437 438 /* an accepting state/position*/ 439 struct _reg_trie_accepted { 440 U8 *endpos; 441 U16 wordnum; 442 }; 443 typedef struct _reg_trie_accepted reg_trie_accepted; 444 445 /* some basic information about the current match that is created by 446 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ 447 448 typedef struct { 449 regexp *prog; 450 char *bol; 451 char *till; 452 SV *sv; 453 char *ganch; 454 char *cutpoint; 455 } regmatch_info; 456 457 458 /* structures for holding and saving the state maintained by regmatch() */ 459 460 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH 461 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 462 #endif 463 464 typedef I32 CHECKPOINT; 465 466 typedef struct regmatch_state { 467 int resume_state; /* where to jump to on return */ 468 char *locinput; /* where to backtrack in string on failure */ 469 470 union { 471 472 /* this is a fake union member that matches the first element 473 * of each member that needs to store positive backtrack 474 * information */ 475 struct { 476 struct regmatch_state *prev_yes_state; 477 } yes; 478 479 /* branchlike members */ 480 /* this is a fake union member that matches the first elements 481 * of each member that needs to behave like a branch */ 482 struct { 483 /* this first element must match u.yes */ 484 struct regmatch_state *prev_yes_state; 485 U32 lastparen; 486 CHECKPOINT cp; 487 488 } branchlike; 489 490 struct { 491 /* the first elements must match u.branchlike */ 492 struct regmatch_state *prev_yes_state; 493 U32 lastparen; 494 CHECKPOINT cp; 495 496 regnode *next_branch; /* next branch node */ 497 } branch; 498 499 struct { 500 /* the first elements must match u.branchlike */ 501 struct regmatch_state *prev_yes_state; 502 U32 lastparen; 503 CHECKPOINT cp; 504 505 reg_trie_accepted *accept_buff; /* accepting states we have seen */ 506 U32 accepted; /* how many accepting states we have seen */ 507 U16 *jump; /* positive offsets from me */ 508 regnode *B; /* node following the trie */ 509 regnode *me; /* Which node am I - needed for jump tries*/ 510 } trie; 511 512 /* special types - these members are used to store state for special 513 regops like eval, if/then, lookaround and the markpoint state */ 514 struct { 515 /* this first element must match u.yes */ 516 struct regmatch_state *prev_yes_state; 517 struct regmatch_state *prev_eval; 518 struct regmatch_state *prev_curlyx; 519 regexp *prev_rex; 520 U32 toggle_reg_flags; /* what bits in PL_reg_flags to 521 flip when transitioning between 522 inner and outer rexen */ 523 CHECKPOINT cp; /* remember current savestack indexes */ 524 CHECKPOINT lastcp; 525 U32 close_paren; /* which close bracket is our end */ 526 regnode *B; /* the node following us */ 527 } eval; 528 529 struct { 530 /* this first element must match u.yes */ 531 struct regmatch_state *prev_yes_state; 532 I32 wanted; 533 I32 logical; /* saved copy of 'logical' var */ 534 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ 535 } ifmatch; /* and SUSPEND/UNLESSM */ 536 537 struct { 538 /* this first element must match u.yes */ 539 struct regmatch_state *prev_yes_state; 540 struct regmatch_state *prev_mark; 541 SV* mark_name; 542 char *mark_loc; 543 } mark; 544 545 struct { 546 int val; 547 } keeper; 548 549 /* quantifiers - these members are used for storing state for 550 for the regops used to implement quantifiers */ 551 struct { 552 /* this first element must match u.yes */ 553 struct regmatch_state *prev_yes_state; 554 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ 555 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 556 CHECKPOINT cp; /* remember current savestack index */ 557 bool minmod; 558 int parenfloor;/* how far back to strip paren data */ 559 int min; /* the minimal number of A's to match */ 560 int max; /* the maximal number of A's to match */ 561 562 /* these two are modified by WHILEM */ 563 int count; /* how many instances of A we've matched */ 564 char *lastloc;/* where previous A matched (0-len detect) */ 565 } curlyx; 566 567 struct { 568 /* this first element must match u.yes */ 569 struct regmatch_state *prev_yes_state; 570 struct regmatch_state *save_curlyx; 571 CHECKPOINT cp; /* remember current savestack indexes */ 572 CHECKPOINT lastcp; 573 char *save_lastloc; /* previous curlyx.lastloc */ 574 I32 cache_offset; 575 I32 cache_mask; 576 } whilem; 577 578 struct { 579 /* this first element must match u.yes */ 580 struct regmatch_state *prev_yes_state; 581 I32 c1, c2; /* case fold search */ 582 CHECKPOINT cp; 583 I32 alen; /* length of first-matched A string */ 584 I32 count; 585 bool minmod; 586 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 587 regnode *me; /* the curlym node */ 588 } curlym; 589 590 struct { 591 U32 paren; 592 CHECKPOINT cp; 593 I32 c1, c2; /* case fold search */ 594 char *maxpos; /* highest possible point in string to match */ 595 char *oldloc; /* the previous locinput */ 596 int count; 597 int min, max; /* {m,n} */ 598 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 599 } curly; /* and CURLYN/PLUS/STAR */ 600 601 } u; 602 } regmatch_state; 603 604 /* how many regmatch_state structs to allocate as a single slab. 605 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev 606 * pointers, plus 1 for any mythical malloc overhead. */ 607 608 #define PERL_REGMATCH_SLAB_SLOTS \ 609 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) 610 611 typedef struct regmatch_slab { 612 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; 613 struct regmatch_slab *prev, *next; 614 } regmatch_slab; 615 616 #define PL_reg_flags PL_reg_state.re_state_reg_flags 617 #define PL_bostr PL_reg_state.re_state_bostr 618 #define PL_reginput PL_reg_state.re_state_reginput 619 #define PL_regeol PL_reg_state.re_state_regeol 620 #define PL_regoffs PL_reg_state.re_state_regoffs 621 #define PL_reglastparen PL_reg_state.re_state_reglastparen 622 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen 623 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp 624 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl 625 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set 626 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 627 #define PL_reg_magic PL_reg_state.re_state_reg_magic 628 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos 629 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm 630 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm 631 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved 632 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen 633 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter 634 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter 635 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache 636 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size 637 #define PL_regsize PL_reg_state.re_state_regsize 638 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry 639 #define PL_nrs PL_reg_state.re_state_nrs 640 641 struct re_save_state { 642 U32 re_state_reg_flags; /* from regexec.c */ 643 U32 re_state_reg_start_tmpl; /* from regexec.c */ 644 I32 re_state_reg_eval_set; /* from regexec.c */ 645 bool re_state_reg_match_utf8; /* from regexec.c */ 646 char *re_state_bostr; 647 char *re_state_reginput; /* String-input pointer. */ 648 char *re_state_regeol; /* End of input, for $ check. */ 649 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ 650 U32 *re_state_reglastparen; /* Similarly for lastparen. */ 651 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ 652 char **re_state_reg_start_tmp; /* from regexec.c */ 653 MAGIC *re_state_reg_magic; /* from regexec.c */ 654 PMOP *re_state_reg_oldcurpm; /* from regexec.c */ 655 PMOP *re_state_reg_curpm; /* from regexec.c */ 656 char *re_state_reg_oldsaved; /* old saved substr during match */ 657 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ 658 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ 659 I32 re_state_reg_oldpos; /* from regexec.c */ 660 I32 re_state_reg_maxiter; /* max wait until caching pos */ 661 I32 re_state_reg_leftiter; /* wait until caching pos */ 662 U32 re_state_regsize; /* from regexec.c */ 663 char *re_state_reg_poscache; /* cache of pos of WHILEM */ 664 char *re_state_reg_starttry; /* from regexec.c */ 665 #ifdef PERL_OLD_COPY_ON_WRITE 666 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 */ 667 #endif 668 }; 669 670 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ 671 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) 672 673 /* 674 * Local variables: 675 * c-indentation-style: bsd 676 * c-basic-offset: 4 677 * indent-tabs-mode: t 678 * End: 679 * 680 * ex: set ts=8 sts=4 sw=4 noet: 681 */ 682