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 #define _REGEXP_COMMON \ 70 /* what engine created this regexp? */ \ 71 const struct regexp_engine* engine; \ 72 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \ 73 /* Information about the match that the perl core uses to */ \ 74 /* 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 /* substring data about strings that must appear in the */ \ 80 /* final match, used for optimisations */ \ 81 struct reg_substr_data *substrs; \ 82 U32 nparens; /* number of capture buffers */ \ 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 /* Data about the last/current match. These are modified */ \ 88 /* during matching */ \ 89 U32 lastparen; /* last open paren matched */ \ 90 U32 lastcloseparen; /* last close paren matched */ \ 91 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \ 92 /* Array of offsets for (@-) and (@+) */ \ 93 regexp_paren_pair *offs; \ 94 /* saved or original string so \digit works forever. */ \ 95 char *subbeg; \ 96 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\ 97 I32 sublen; /* Length of string pointed by subbeg */ \ 98 /* Information about the match that isn't often used */ \ 99 /* offset from wrapped to the start of precomp */ \ 100 PERL_BITFIELD32 pre_prefix:4; \ 101 /* number of eval groups in the pattern - for security checks */\ 102 PERL_BITFIELD32 seen_evals:28 103 104 typedef struct regexp { 105 _XPV_HEAD; 106 _XPVMG_HEAD; 107 _REGEXP_COMMON; 108 } regexp; 109 110 /* HV *paren_names; Optional hash of paren names 111 now stored in the IV union */ 112 113 #define RXp_PAREN_NAMES(rx) ((rx)->xiv_u.xivu_hv) 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_ SV * const pattern, 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 /* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME 234 * If you change these you need to change the equivalent flags in op.h, and 235 * vice versa. */ 236 #define RXf_PMf_MULTILINE 0x00000001 /* /m */ 237 #define RXf_PMf_SINGLELINE 0x00000002 /* /s */ 238 #define RXf_PMf_FOLD 0x00000004 /* /i */ 239 #define RXf_PMf_EXTENDED 0x00000008 /* /x */ 240 #define RXf_PMf_KEEPCOPY 0x00000010 /* /p */ 241 #define RXf_PMf_LOCALE 0x00000020 /* use locale */ 242 /* these flags are transfered from the PMOP->op_pmflags member during compilation */ 243 #define RXf_PMf_STD_PMMOD_SHIFT 0 244 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) 245 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) 246 247 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ 248 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ 249 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ 250 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ 251 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break 252 253 /* chars and strings used as regex pattern modifiers 254 * Singlular is a 'c'har, plural is a "string" 255 * 256 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve 257 * for compatibility reasons with Regexp::Common which highjacked (?k:...) 258 * for its own uses. So 'k' is out as well. 259 */ 260 #define EXEC_PAT_MOD 'e' 261 #define KEEPCOPY_PAT_MOD 'p' 262 #define ONCE_PAT_MOD 'o' 263 #define GLOBAL_PAT_MOD 'g' 264 #define CONTINUE_PAT_MOD 'c' 265 #define MULTILINE_PAT_MOD 'm' 266 #define SINGLE_PAT_MOD 's' 267 #define IGNORE_PAT_MOD 'i' 268 #define XTENDED_PAT_MOD 'x' 269 270 #define ONCE_PAT_MODS "o" 271 #define KEEPCOPY_PAT_MODS "p" 272 #define EXEC_PAT_MODS "e" 273 #define LOOP_PAT_MODS "gc" 274 275 #define STD_PAT_MODS "msix" 276 277 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS 278 279 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS 280 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS 281 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS 282 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS 283 284 /* 285 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 286 * so that regnodes.h is updated with the changes. 287 * 288 */ 289 290 /* Anchor and GPOS related stuff */ 291 #define RXf_ANCH_BOL 0x00000100 292 #define RXf_ANCH_MBOL 0x00000200 293 #define RXf_ANCH_SBOL 0x00000400 294 #define RXf_ANCH_GPOS 0x00000800 295 #define RXf_GPOS_SEEN 0x00001000 296 #define RXf_GPOS_FLOAT 0x00002000 297 /* two bits here */ 298 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) 299 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) 300 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) 301 302 /* What we have seen */ 303 #define RXf_LOOKBEHIND_SEEN 0x00004000 304 #define RXf_EVAL_SEEN 0x00008000 305 #define RXf_CANY_SEEN 0x00010000 306 307 /* Special */ 308 #define RXf_NOSCAN 0x00020000 309 #define RXf_CHECK_ALL 0x00040000 310 311 /* UTF8 related */ 312 #define RXf_MATCH_UTF8 0x00100000 313 314 /* Intuit related */ 315 #define RXf_USE_INTUIT_NOML 0x00200000 316 #define RXf_USE_INTUIT_ML 0x00400000 317 #define RXf_INTUIT_TAIL 0x00800000 318 319 /* 320 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will 321 be used by regex engines to check whether they should set 322 RXf_SKIPWHITE 323 */ 324 #define RXf_SPLIT 0x01000000 325 326 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) 327 328 /* Copy and tainted info */ 329 #define RXf_COPY_DONE 0x02000000 330 #define RXf_TAINTED_SEEN 0x04000000 331 #define RXf_TAINTED 0x08000000 /* this pattern is tainted */ 332 333 /* Flags indicating special patterns */ 334 #define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */ 335 #define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */ 336 #define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */ 337 #define RXf_NULL 0x80000000 /* Pattern is // */ 338 339 /* 340 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl 341 * so that regnodes.h is updated with the changes. 342 * 343 */ 344 345 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) 346 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 347 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 348 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) 349 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) 350 #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 351 ? RX_MATCH_TAINTED_on(prog) \ 352 : RX_MATCH_TAINTED_off(prog)) 353 354 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) 355 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) 356 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) 357 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) 358 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 359 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 360 #define RX_MATCH_COPIED_set(prog,t) ((t) \ 361 ? RX_MATCH_COPIED_on(prog) \ 362 : RX_MATCH_COPIED_off(prog)) 363 364 #define RXp_EXTFLAGS(rx) ((rx)->extflags) 365 366 /* For source compatibility. We used to store these explicitly. */ 367 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) 368 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) 369 /* FIXME? Are we hardcoding too much here and constraining plugin extension 370 writers? Specifically, the value 1 assumes that the wrapped version always 371 has exactly one character at the end, a ')'. Will that always be true? */ 372 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1) 373 #define RX_WRAPPED(prog) SvPVX(prog) 374 #define RX_WRAPPED_const(prog) SvPVX_const(prog) 375 #define RX_WRAPLEN(prog) SvCUR(prog) 376 #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr) 377 #define RX_REFCNT(prog) SvREFCNT(prog) 378 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) 379 # define RX_EXTFLAGS(prog) \ 380 (*({ \ 381 const REGEXP *const _rx_extflags = (prog); \ 382 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \ 383 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \ 384 })) 385 # define RX_ENGINE(prog) \ 386 (*({ \ 387 const REGEXP *const _rx_engine = (prog); \ 388 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \ 389 &SvANY(_rx_engine)->engine; \ 390 })) 391 # define RX_SUBBEG(prog) \ 392 (*({ \ 393 const REGEXP *const _rx_subbeg = (prog); \ 394 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \ 395 &SvANY(_rx_subbeg)->subbeg; \ 396 })) 397 # define RX_OFFS(prog) \ 398 (*({ \ 399 const REGEXP *const _rx_offs = (prog); \ 400 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \ 401 &SvANY(_rx_offs)->offs; \ 402 })) 403 # define RX_NPARENS(prog) \ 404 (*({ \ 405 const REGEXP *const _rx_nparens = (prog); \ 406 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \ 407 &SvANY(_rx_nparens)->nparens; \ 408 })) 409 #else 410 # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog)) 411 # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine) 412 # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg) 413 # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs) 414 # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens) 415 #endif 416 #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen) 417 #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen) 418 #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret) 419 #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs) 420 #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen) 421 #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen) 422 #define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals) 423 #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy) 424 425 #endif /* PLUGGABLE_RE_EXTENSION */ 426 427 /* Stuff that needs to be included in the plugable extension goes below here */ 428 429 #ifdef PERL_OLD_COPY_ON_WRITE 430 #define RX_MATCH_COPY_FREE(rx) \ 431 STMT_START {if (RX_SAVED_COPY(rx)) { \ 432 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \ 433 } \ 434 if (RX_MATCH_COPIED(rx)) { \ 435 Safefree(RX_SUBBEG(rx)); \ 436 RX_MATCH_COPIED_off(rx); \ 437 }} STMT_END 438 #else 439 #define RX_MATCH_COPY_FREE(rx) \ 440 STMT_START {if (RX_MATCH_COPIED(rx)) { \ 441 Safefree(RX_SUBBEG(rx)); \ 442 RX_MATCH_COPIED_off(rx); \ 443 }} STMT_END 444 #endif 445 446 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) 447 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) 448 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) 449 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) 450 #define RX_MATCH_UTF8_set(prog, t) ((t) \ 451 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 452 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 453 454 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ 455 #define RX_UTF8(prog) SvUTF8(prog) 456 457 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 458 #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 459 #define REXEC_SCREAM 0x04 /* use scream table. */ 460 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 461 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 462 463 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) 464 # define ReREFCNT_inc(re) \ 465 ({ \ 466 /* This is here to generate a casting warning if incorrect. */ \ 467 REGEXP *const _rerefcnt_inc = (re); \ 468 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \ 469 SvREFCNT_inc(_rerefcnt_inc); \ 470 _rerefcnt_inc; \ 471 }) 472 # define ReREFCNT_dec(re) \ 473 ({ \ 474 /* This is here to generate a casting warning if incorrect. */ \ 475 REGEXP *const _rerefcnt_dec = (re); \ 476 SvREFCNT_dec(_rerefcnt_dec); \ 477 }) 478 #else 479 # define ReREFCNT_dec(re) SvREFCNT_dec(re) 480 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re)) 481 #endif 482 483 /* FIXME for plugins. */ 484 485 #define FBMcf_TAIL_DOLLAR 1 486 #define FBMcf_TAIL_DOLLARM 2 487 #define FBMcf_TAIL_Z 4 488 #define FBMcf_TAIL_z 8 489 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 490 491 #define FBMrf_MULTILINE 1 492 493 /* an accepting state/position*/ 494 struct _reg_trie_accepted { 495 U8 *endpos; 496 U16 wordnum; 497 }; 498 typedef struct _reg_trie_accepted reg_trie_accepted; 499 500 /* some basic information about the current match that is created by 501 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ 502 503 typedef struct { 504 REGEXP *prog; 505 char *bol; 506 char *till; 507 SV *sv; 508 char *ganch; 509 char *cutpoint; 510 } regmatch_info; 511 512 513 /* structures for holding and saving the state maintained by regmatch() */ 514 515 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH 516 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 517 #endif 518 519 typedef I32 CHECKPOINT; 520 521 typedef struct regmatch_state { 522 int resume_state; /* where to jump to on return */ 523 char *locinput; /* where to backtrack in string on failure */ 524 525 union { 526 527 /* this is a fake union member that matches the first element 528 * of each member that needs to store positive backtrack 529 * information */ 530 struct { 531 struct regmatch_state *prev_yes_state; 532 } yes; 533 534 /* branchlike members */ 535 /* this is a fake union member that matches the first elements 536 * of each member that needs to behave like a branch */ 537 struct { 538 /* this first element must match u.yes */ 539 struct regmatch_state *prev_yes_state; 540 U32 lastparen; 541 CHECKPOINT cp; 542 543 } branchlike; 544 545 struct { 546 /* the first elements must match u.branchlike */ 547 struct regmatch_state *prev_yes_state; 548 U32 lastparen; 549 CHECKPOINT cp; 550 551 regnode *next_branch; /* next branch node */ 552 } branch; 553 554 struct { 555 /* the first elements must match u.branchlike */ 556 struct regmatch_state *prev_yes_state; 557 U32 lastparen; 558 CHECKPOINT cp; 559 560 reg_trie_accepted *accept_buff; /* accepting states we have seen */ 561 U32 accepted; /* how many accepting states we have seen */ 562 U16 *jump; /* positive offsets from me */ 563 regnode *B; /* node following the trie */ 564 regnode *me; /* Which node am I - needed for jump tries*/ 565 } trie; 566 567 /* special types - these members are used to store state for special 568 regops like eval, if/then, lookaround and the markpoint state */ 569 struct { 570 /* this first element must match u.yes */ 571 struct regmatch_state *prev_yes_state; 572 struct regmatch_state *prev_eval; 573 struct regmatch_state *prev_curlyx; 574 REGEXP *prev_rex; 575 U32 toggle_reg_flags; /* what bits in PL_reg_flags to 576 flip when transitioning between 577 inner and outer rexen */ 578 CHECKPOINT cp; /* remember current savestack indexes */ 579 CHECKPOINT lastcp; 580 U32 close_paren; /* which close bracket is our end */ 581 regnode *B; /* the node following us */ 582 } eval; 583 584 struct { 585 /* this first element must match u.yes */ 586 struct regmatch_state *prev_yes_state; 587 I32 wanted; 588 I32 logical; /* saved copy of 'logical' var */ 589 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ 590 } ifmatch; /* and SUSPEND/UNLESSM */ 591 592 struct { 593 /* this first element must match u.yes */ 594 struct regmatch_state *prev_yes_state; 595 struct regmatch_state *prev_mark; 596 SV* mark_name; 597 char *mark_loc; 598 } mark; 599 600 struct { 601 int val; 602 } keeper; 603 604 /* quantifiers - these members are used for storing state for 605 for the regops used to implement quantifiers */ 606 struct { 607 /* this first element must match u.yes */ 608 struct regmatch_state *prev_yes_state; 609 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ 610 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 611 CHECKPOINT cp; /* remember current savestack index */ 612 bool minmod; 613 int parenfloor;/* how far back to strip paren data */ 614 int min; /* the minimal number of A's to match */ 615 int max; /* the maximal number of A's to match */ 616 617 /* these two are modified by WHILEM */ 618 int count; /* how many instances of A we've matched */ 619 char *lastloc;/* where previous A matched (0-len detect) */ 620 } curlyx; 621 622 struct { 623 /* this first element must match u.yes */ 624 struct regmatch_state *prev_yes_state; 625 struct regmatch_state *save_curlyx; 626 CHECKPOINT cp; /* remember current savestack indexes */ 627 CHECKPOINT lastcp; 628 char *save_lastloc; /* previous curlyx.lastloc */ 629 I32 cache_offset; 630 I32 cache_mask; 631 } whilem; 632 633 struct { 634 /* this first element must match u.yes */ 635 struct regmatch_state *prev_yes_state; 636 I32 c1, c2; /* case fold search */ 637 CHECKPOINT cp; 638 I32 alen; /* length of first-matched A string */ 639 I32 count; 640 bool minmod; 641 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 642 regnode *me; /* the curlym node */ 643 } curlym; 644 645 struct { 646 U32 paren; 647 CHECKPOINT cp; 648 I32 c1, c2; /* case fold search */ 649 char *maxpos; /* highest possible point in string to match */ 650 char *oldloc; /* the previous locinput */ 651 int count; 652 int min, max; /* {m,n} */ 653 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 654 } curly; /* and CURLYN/PLUS/STAR */ 655 656 } u; 657 } regmatch_state; 658 659 /* how many regmatch_state structs to allocate as a single slab. 660 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev 661 * pointers, plus 1 for any mythical malloc overhead. */ 662 663 #define PERL_REGMATCH_SLAB_SLOTS \ 664 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) 665 666 typedef struct regmatch_slab { 667 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; 668 struct regmatch_slab *prev, *next; 669 } regmatch_slab; 670 671 #define PL_reg_flags PL_reg_state.re_state_reg_flags 672 #define PL_bostr PL_reg_state.re_state_bostr 673 #define PL_reginput PL_reg_state.re_state_reginput 674 #define PL_regeol PL_reg_state.re_state_regeol 675 #define PL_regoffs PL_reg_state.re_state_regoffs 676 #define PL_reglastparen PL_reg_state.re_state_reglastparen 677 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen 678 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp 679 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl 680 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set 681 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 682 #define PL_reg_magic PL_reg_state.re_state_reg_magic 683 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos 684 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm 685 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm 686 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved 687 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen 688 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter 689 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter 690 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache 691 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size 692 #define PL_regsize PL_reg_state.re_state_regsize 693 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry 694 #define PL_nrs PL_reg_state.re_state_nrs 695 696 struct re_save_state { 697 U32 re_state_reg_flags; /* from regexec.c */ 698 U32 re_state_reg_start_tmpl; /* from regexec.c */ 699 I32 re_state_reg_eval_set; /* from regexec.c */ 700 bool re_state_reg_match_utf8; /* from regexec.c */ 701 char *re_state_bostr; 702 char *re_state_reginput; /* String-input pointer. */ 703 char *re_state_regeol; /* End of input, for $ check. */ 704 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ 705 U32 *re_state_reglastparen; /* Similarly for lastparen. */ 706 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ 707 char **re_state_reg_start_tmp; /* from regexec.c */ 708 MAGIC *re_state_reg_magic; /* from regexec.c */ 709 PMOP *re_state_reg_oldcurpm; /* from regexec.c */ 710 PMOP *re_state_reg_curpm; /* from regexec.c */ 711 char *re_state_reg_oldsaved; /* old saved substr during match */ 712 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ 713 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ 714 I32 re_state_reg_oldpos; /* from regexec.c */ 715 I32 re_state_reg_maxiter; /* max wait until caching pos */ 716 I32 re_state_reg_leftiter; /* wait until caching pos */ 717 U32 re_state_regsize; /* from regexec.c */ 718 char *re_state_reg_poscache; /* cache of pos of WHILEM */ 719 char *re_state_reg_starttry; /* from regexec.c */ 720 #ifdef PERL_OLD_COPY_ON_WRITE 721 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 */ 722 #endif 723 }; 724 725 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ 726 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) 727 728 /* 729 * Local variables: 730 * c-indentation-style: bsd 731 * c-basic-offset: 4 732 * indent-tabs-mode: t 733 * End: 734 * 735 * ex: set ts=8 sts=4 sw=4 noet: 736 */ 737