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 #if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C) 59 #define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output) 60 #define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output) 61 62 /* Subtracting b from a leaves in a everything that was there that isn't in b, 63 * that is the intersection of a with b's complement */ 64 #define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output) 65 #endif 66 67 /* 68 The regexp/REGEXP struct, see L<perlreapi> for further documentation 69 on the individual fields. The struct is ordered so that the most 70 commonly used fields are placed at the start. 71 72 Any patch that adds items to this struct will need to include 73 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c> 74 (C<pregfree()>). This involves freeing or cloning items in the 75 regexp's data array based on the data item's type. 76 */ 77 78 #define _REGEXP_COMMON \ 79 /* what engine created this regexp? */ \ 80 const struct regexp_engine* engine; \ 81 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \ 82 HV *paren_names; /* Optional hash of paren names */ \ 83 /* Information about the match that the perl core uses to */ \ 84 /* manage things */ \ 85 U32 extflags; /* Flags used both externally and internally */ \ 86 I32 minlen; /* mininum possible length of string to match */\ 87 I32 minlenret; /* mininum possible length of $& */ \ 88 U32 gofs; /* chars left of pos that we search from */ \ 89 /* substring data about strings that must appear in the */ \ 90 /* final match, used for optimisations */ \ 91 struct reg_substr_data *substrs; \ 92 U32 nparens; /* number of capture buffers */ \ 93 /* private engine specific data */ \ 94 U32 intflags; /* Engine Specific Internal flags */ \ 95 void *pprivate; /* Data private to the regex engine which */ \ 96 /* created this object. */ \ 97 /* Data about the last/current match. These are modified */ \ 98 /* during matching */ \ 99 U32 lastparen; /* last open paren matched */ \ 100 U32 lastcloseparen; /* last close paren matched */ \ 101 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \ 102 /* Array of offsets for (@-) and (@+) */ \ 103 regexp_paren_pair *offs; \ 104 /* saved or original string so \digit works forever. */ \ 105 char *subbeg; \ 106 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\ 107 I32 sublen; /* Length of string pointed by subbeg */ \ 108 /* Information about the match that isn't often used */ \ 109 /* offset from wrapped to the start of precomp */ \ 110 PERL_BITFIELD32 pre_prefix:4; \ 111 /* number of eval groups in the pattern - for security checks */\ 112 PERL_BITFIELD32 seen_evals:28 113 114 typedef struct regexp { 115 _XPV_HEAD; 116 _REGEXP_COMMON; 117 } regexp; 118 119 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names) 120 121 /* used for high speed searches */ 122 typedef struct re_scream_pos_data_s 123 { 124 char **scream_olds; /* match pos */ 125 I32 *scream_pos; /* Internal iterator of scream. */ 126 } re_scream_pos_data; 127 128 /* regexp_engine structure. This is the dispatch table for regexes. 129 * Any regex engine implementation must be able to build one of these. 130 */ 131 typedef struct regexp_engine { 132 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags); 133 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, 134 char* strbeg, I32 minend, SV* screamer, 135 void* data, U32 flags); 136 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, 137 char *strend, const U32 flags, 138 re_scream_pos_data *data); 139 SV* (*checkstr) (pTHX_ REGEXP * const rx); 140 void (*free) (pTHX_ REGEXP * const rx); 141 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, 142 SV * const sv); 143 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, 144 SV const * const value); 145 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, 146 const I32 paren); 147 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, 148 SV * const value, const U32 flags); 149 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, 150 const U32 flags); 151 SV* (*qr_package)(pTHX_ REGEXP * const rx); 152 #ifdef USE_ITHREADS 153 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); 154 #endif 155 } regexp_engine; 156 157 /* 158 These are passed to the numbered capture variable callbacks as the 159 paren name. >= 1 is reserved for actual numbered captures, i.e. $1, 160 $2 etc. 161 */ 162 #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ 163 #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ 164 #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ 165 166 /* 167 Flags that are passed to the named_buff and named_buff_iter 168 callbacks above. Those routines are called from universal.c via the 169 Tie::Hash::NamedCapture interface for %+ and %- and the re:: 170 functions in the same file. 171 */ 172 173 /* The Tie::Hash::NamedCapture operation this is part of, if any */ 174 #define RXapif_FETCH 0x0001 175 #define RXapif_STORE 0x0002 176 #define RXapif_DELETE 0x0004 177 #define RXapif_CLEAR 0x0008 178 #define RXapif_EXISTS 0x0010 179 #define RXapif_SCALAR 0x0020 180 #define RXapif_FIRSTKEY 0x0040 181 #define RXapif_NEXTKEY 0x0080 182 183 /* Whether %+ or %- is being operated on */ 184 #define RXapif_ONE 0x0100 /* %+ */ 185 #define RXapif_ALL 0x0200 /* %- */ 186 187 /* Whether this is being called from a re:: function */ 188 #define RXapif_REGNAME 0x0400 189 #define RXapif_REGNAMES 0x0800 190 #define RXapif_REGNAMES_COUNT 0x1000 191 192 /* 193 =head1 REGEXP Functions 194 195 =for apidoc Am|REGEXP *|SvRX|SV *sv 196 197 Convenience macro to get the REGEXP from a SV. This is approximately 198 equivalent to the following snippet: 199 200 if (SvMAGICAL(sv)) 201 mg_get(sv); 202 if (SvROK(sv)) 203 sv = MUTABLE_SV(SvRV(sv)); 204 if (SvTYPE(sv) == SVt_REGEXP) 205 return (REGEXP*) sv; 206 207 NULL will be returned if a REGEXP* is not found. 208 209 =for apidoc Am|bool|SvRXOK|SV* sv 210 211 Returns a boolean indicating whether the SV (or the one it references) 212 is a REGEXP. 213 214 If you want to do something with the REGEXP* later use SvRX instead 215 and check for NULL. 216 217 =cut 218 */ 219 220 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) 221 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) 222 223 224 /* Flags stored in regexp->extflags 225 * These are used by code external to the regexp engine 226 * 227 * Note that the flags whose names start with RXf_PMf_ are defined in 228 * op_reg_common.h, being copied from the parallel flags of op_pmflags 229 * 230 * NOTE: if you modify any RXf flags you should run regen.pl or 231 * regen/regcomp.pl so that regnodes.h is updated with the changes. 232 * 233 */ 234 235 #include "op_reg_common.h" 236 237 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) 238 239 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ 240 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ 241 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ 242 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ 243 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break 244 245 /* Note, includes charset ones, assumes 0 is the default for them */ 246 #define STD_PMMOD_FLAGS_CLEAR(pmfl) \ 247 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET) 248 249 /* chars and strings used as regex pattern modifiers 250 * Singular is a 'c'har, plural is a "string" 251 * 252 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve 253 * for compatibility reasons with Regexp::Common which highjacked (?k:...) 254 * for its own uses. So 'k' is out as well. 255 */ 256 #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */ 257 #define EXEC_PAT_MOD 'e' 258 #define KEEPCOPY_PAT_MOD 'p' 259 #define ONCE_PAT_MOD 'o' 260 #define GLOBAL_PAT_MOD 'g' 261 #define CONTINUE_PAT_MOD 'c' 262 #define MULTILINE_PAT_MOD 'm' 263 #define SINGLE_PAT_MOD 's' 264 #define IGNORE_PAT_MOD 'i' 265 #define XTENDED_PAT_MOD 'x' 266 #define NONDESTRUCT_PAT_MOD 'r' 267 #define LOCALE_PAT_MOD 'l' 268 #define UNICODE_PAT_MOD 'u' 269 #define DEPENDS_PAT_MOD 'd' 270 #define ASCII_RESTRICT_PAT_MOD 'a' 271 272 #define ONCE_PAT_MODS "o" 273 #define KEEPCOPY_PAT_MODS "p" 274 #define EXEC_PAT_MODS "e" 275 #define LOOP_PAT_MODS "gc" 276 #define NONDESTRUCT_PAT_MODS "r" 277 #define LOCALE_PAT_MODS "l" 278 #define UNICODE_PAT_MODS "u" 279 #define DEPENDS_PAT_MODS "d" 280 #define ASCII_RESTRICT_PAT_MODS "a" 281 #define ASCII_MORE_RESTRICT_PAT_MODS "aa" 282 283 /* This string is expected by regcomp.c to be ordered so that the first 284 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next 285 * character is bit +1, etc. */ 286 #define STD_PAT_MODS "msix" 287 288 #define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS 289 290 /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered 291 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of 292 * extflags; the next character is in bit +1, etc. */ 293 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS 294 295 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS 296 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS 297 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS 298 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS 299 300 /* 301 * NOTE: if you modify any RXf flags you should run regen.pl or 302 * regen/regcomp.pl so that regnodes.h is updated with the changes. 303 * 304 */ 305 306 /* Leave some space, so future bit allocations can go either in the shared or 307 * unshared area without affecting binary compatibility */ 308 #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1) 309 310 /* Manually decorate this function with gcc-style attributes just to 311 * avoid having to restructure the header files and their called order, 312 * as proto.h would have to be included before this file, and isn't */ 313 314 PERL_STATIC_INLINE const char * 315 get_regex_charset_name(const U32 flags, STRLEN* const lenp) 316 __attribute__warn_unused_result__; 317 318 #define MAX_CHARSET_NAME_LENGTH 2 319 320 PERL_STATIC_INLINE const char * 321 get_regex_charset_name(const U32 flags, STRLEN* const lenp) 322 { 323 /* Returns a string that corresponds to the name of the regex character set 324 * given by 'flags', and *lenp is set the length of that string, which 325 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */ 326 327 *lenp = 1; 328 switch (get_regex_charset(flags)) { 329 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS; 330 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS; 331 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS; 332 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS; 333 case REGEX_ASCII_MORE_RESTRICTED_CHARSET: 334 *lenp = 2; 335 return ASCII_MORE_RESTRICT_PAT_MODS; 336 default: 337 return "?"; /* Unknown */ 338 } 339 } 340 341 /* Anchor and GPOS related stuff */ 342 #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0)) 343 #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1)) 344 #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2)) 345 #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3)) 346 #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4)) 347 #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5)) 348 /* two bits here */ 349 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) 350 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) 351 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) 352 353 /* What we have seen */ 354 #define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6)) 355 #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7)) 356 #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8)) 357 358 /* Special */ 359 #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9)) 360 #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10)) 361 362 /* UTF8 related */ 363 #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) 364 365 /* Intuit related */ 366 #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12)) 367 #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13)) 368 #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14)) 369 370 /* 371 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will 372 be used by regex engines to check whether they should set 373 RXf_SKIPWHITE 374 */ 375 #define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15)) 376 377 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) 378 379 /* Copy and tainted info */ 380 #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16)) 381 382 /* during execution: pattern temporarily tainted by executing locale ops; 383 * post-execution: $1 et al are tainted */ 384 #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17)) 385 /* this pattern was tainted during compilation */ 386 #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18)) 387 388 /* Flags indicating special patterns */ 389 #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */ 390 #define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */ 391 #define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */ 392 #define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */ 393 #if RXf_BASE_SHIFT+22 > 31 394 # error Too many RXf_PMf bits used. See regnodes.h for any spare in middle 395 #endif 396 397 /* 398 * NOTE: if you modify any RXf flags you should run regen.pl or 399 * regen/regcomp.pl so that regnodes.h is updated with the changes. 400 * 401 */ 402 403 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) 404 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 405 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 406 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) 407 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) 408 #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 409 ? RX_MATCH_TAINTED_on(prog) \ 410 : RX_MATCH_TAINTED_off(prog)) 411 412 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) 413 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) 414 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) 415 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) 416 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 417 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 418 #define RX_MATCH_COPIED_set(prog,t) ((t) \ 419 ? RX_MATCH_COPIED_on(prog) \ 420 : RX_MATCH_COPIED_off(prog)) 421 422 #define RXp_EXTFLAGS(rx) ((rx)->extflags) 423 424 /* For source compatibility. We used to store these explicitly. */ 425 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) 426 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) 427 /* FIXME? Are we hardcoding too much here and constraining plugin extension 428 writers? Specifically, the value 1 assumes that the wrapped version always 429 has exactly one character at the end, a ')'. Will that always be true? */ 430 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1) 431 #define RX_WRAPPED(prog) SvPVX(prog) 432 #define RX_WRAPPED_const(prog) SvPVX_const(prog) 433 #define RX_WRAPLEN(prog) SvCUR(prog) 434 #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr) 435 #define RX_REFCNT(prog) SvREFCNT(prog) 436 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) 437 # define RX_EXTFLAGS(prog) \ 438 (*({ \ 439 const REGEXP *const _rx_extflags = (prog); \ 440 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \ 441 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \ 442 })) 443 # define RX_ENGINE(prog) \ 444 (*({ \ 445 const REGEXP *const _rx_engine = (prog); \ 446 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \ 447 &SvANY(_rx_engine)->engine; \ 448 })) 449 # define RX_SUBBEG(prog) \ 450 (*({ \ 451 const REGEXP *const _rx_subbeg = (prog); \ 452 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \ 453 &SvANY(_rx_subbeg)->subbeg; \ 454 })) 455 # define RX_OFFS(prog) \ 456 (*({ \ 457 const REGEXP *const _rx_offs = (prog); \ 458 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \ 459 &SvANY(_rx_offs)->offs; \ 460 })) 461 # define RX_NPARENS(prog) \ 462 (*({ \ 463 const REGEXP *const _rx_nparens = (prog); \ 464 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \ 465 &SvANY(_rx_nparens)->nparens; \ 466 })) 467 #else 468 # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog)) 469 # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine) 470 # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg) 471 # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs) 472 # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens) 473 #endif 474 #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen) 475 #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen) 476 #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret) 477 #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs) 478 #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen) 479 #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen) 480 #define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals) 481 #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy) 482 483 #endif /* PLUGGABLE_RE_EXTENSION */ 484 485 /* Stuff that needs to be included in the pluggable extension goes below here */ 486 487 #ifdef PERL_OLD_COPY_ON_WRITE 488 #define RX_MATCH_COPY_FREE(rx) \ 489 STMT_START {if (RX_SAVED_COPY(rx)) { \ 490 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \ 491 } \ 492 if (RX_MATCH_COPIED(rx)) { \ 493 Safefree(RX_SUBBEG(rx)); \ 494 RX_MATCH_COPIED_off(rx); \ 495 }} STMT_END 496 #else 497 #define RX_MATCH_COPY_FREE(rx) \ 498 STMT_START {if (RX_MATCH_COPIED(rx)) { \ 499 Safefree(RX_SUBBEG(rx)); \ 500 RX_MATCH_COPIED_off(rx); \ 501 }} STMT_END 502 #endif 503 504 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) 505 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) 506 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) 507 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) 508 #define RX_MATCH_UTF8_set(prog, t) ((t) \ 509 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 510 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 511 512 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ 513 #define RX_UTF8(prog) SvUTF8(prog) 514 515 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 516 #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 517 #define REXEC_SCREAM 0x04 /* use scream table. */ 518 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 519 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 520 521 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) 522 # define ReREFCNT_inc(re) \ 523 ({ \ 524 /* This is here to generate a casting warning if incorrect. */ \ 525 REGEXP *const _rerefcnt_inc = (re); \ 526 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \ 527 SvREFCNT_inc(_rerefcnt_inc); \ 528 _rerefcnt_inc; \ 529 }) 530 # define ReREFCNT_dec(re) \ 531 ({ \ 532 /* This is here to generate a casting warning if incorrect. */ \ 533 REGEXP *const _rerefcnt_dec = (re); \ 534 SvREFCNT_dec(_rerefcnt_dec); \ 535 }) 536 #else 537 # define ReREFCNT_dec(re) SvREFCNT_dec(re) 538 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re)) 539 #endif 540 541 /* FIXME for plugins. */ 542 543 #define FBMcf_TAIL_DOLLAR 1 544 #define FBMcf_TAIL_DOLLARM 2 545 #define FBMcf_TAIL_Z 4 546 #define FBMcf_TAIL_z 8 547 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 548 549 #define FBMrf_MULTILINE 1 550 551 /* some basic information about the current match that is created by 552 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ 553 554 typedef struct { 555 REGEXP *prog; 556 char *bol; 557 char *till; 558 SV *sv; 559 char *ganch; 560 char *cutpoint; 561 } regmatch_info; 562 563 564 /* structures for holding and saving the state maintained by regmatch() */ 565 566 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH 567 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 568 #endif 569 570 typedef I32 CHECKPOINT; 571 572 typedef struct regmatch_state { 573 int resume_state; /* where to jump to on return */ 574 char *locinput; /* where to backtrack in string on failure */ 575 576 union { 577 578 /* this is a fake union member that matches the first element 579 * of each member that needs to store positive backtrack 580 * information */ 581 struct { 582 struct regmatch_state *prev_yes_state; 583 } yes; 584 585 /* branchlike members */ 586 /* this is a fake union member that matches the first elements 587 * of each member that needs to behave like a branch */ 588 struct { 589 /* this first element must match u.yes */ 590 struct regmatch_state *prev_yes_state; 591 U32 lastparen; 592 CHECKPOINT cp; 593 594 } branchlike; 595 596 struct { 597 /* the first elements must match u.branchlike */ 598 struct regmatch_state *prev_yes_state; 599 U32 lastparen; 600 CHECKPOINT cp; 601 602 regnode *next_branch; /* next branch node */ 603 } branch; 604 605 struct { 606 /* the first elements must match u.branchlike */ 607 struct regmatch_state *prev_yes_state; 608 U32 lastparen; 609 CHECKPOINT cp; 610 611 U32 accepted; /* how many accepting states left */ 612 U16 *jump; /* positive offsets from me */ 613 regnode *B; /* node following the trie */ 614 regnode *me; /* Which node am I - needed for jump tries*/ 615 U8 *firstpos;/* pos in string of first trie match */ 616 U32 firstchars;/* len in chars of firstpos from start */ 617 U16 nextword;/* next word to try */ 618 U16 topword; /* longest accepted word */ 619 bool longfold;/* saw a fold with a 1->n char mapping */ 620 } trie; 621 622 /* special types - these members are used to store state for special 623 regops like eval, if/then, lookaround and the markpoint state */ 624 struct { 625 /* this first element must match u.yes */ 626 struct regmatch_state *prev_yes_state; 627 struct regmatch_state *prev_eval; 628 struct regmatch_state *prev_curlyx; 629 REGEXP *prev_rex; 630 U32 toggle_reg_flags; /* what bits in PL_reg_flags to 631 flip when transitioning between 632 inner and outer rexen */ 633 CHECKPOINT cp; /* remember current savestack indexes */ 634 CHECKPOINT lastcp; 635 U32 close_paren; /* which close bracket is our end */ 636 regnode *B; /* the node following us */ 637 } eval; 638 639 struct { 640 /* this first element must match u.yes */ 641 struct regmatch_state *prev_yes_state; 642 I32 wanted; 643 I32 logical; /* saved copy of 'logical' var */ 644 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ 645 } ifmatch; /* and SUSPEND/UNLESSM */ 646 647 struct { 648 /* this first element must match u.yes */ 649 struct regmatch_state *prev_yes_state; 650 struct regmatch_state *prev_mark; 651 SV* mark_name; 652 char *mark_loc; 653 } mark; 654 655 struct { 656 int val; 657 } keeper; 658 659 /* quantifiers - these members are used for storing state for 660 for the regops used to implement quantifiers */ 661 struct { 662 /* this first element must match u.yes */ 663 struct regmatch_state *prev_yes_state; 664 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ 665 regnode *me; /* the CURLYX node */ 666 regnode *B; /* the B node in /A*B/ */ 667 CHECKPOINT cp; /* remember current savestack index */ 668 bool minmod; 669 int parenfloor;/* how far back to strip paren data */ 670 671 /* these two are modified by WHILEM */ 672 int count; /* how many instances of A we've matched */ 673 char *lastloc;/* where previous A matched (0-len detect) */ 674 } curlyx; 675 676 struct { 677 /* this first element must match u.yes */ 678 struct regmatch_state *prev_yes_state; 679 struct regmatch_state *save_curlyx; 680 CHECKPOINT cp; /* remember current savestack indexes */ 681 CHECKPOINT lastcp; 682 char *save_lastloc; /* previous curlyx.lastloc */ 683 I32 cache_offset; 684 I32 cache_mask; 685 } whilem; 686 687 struct { 688 /* this first element must match u.yes */ 689 struct regmatch_state *prev_yes_state; 690 I32 c1, c2; /* case fold search */ 691 CHECKPOINT cp; 692 I32 alen; /* length of first-matched A string */ 693 I32 count; 694 bool minmod; 695 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 696 regnode *me; /* the curlym node */ 697 } curlym; 698 699 struct { 700 U32 paren; 701 CHECKPOINT cp; 702 I32 c1, c2; /* case fold search */ 703 char *maxpos; /* highest possible point in string to match */ 704 char *oldloc; /* the previous locinput */ 705 int count; 706 int min, max; /* {m,n} */ 707 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 708 } curly; /* and CURLYN/PLUS/STAR */ 709 710 } u; 711 } regmatch_state; 712 713 /* how many regmatch_state structs to allocate as a single slab. 714 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev 715 * pointers, plus 1 for any mythical malloc overhead. */ 716 717 #define PERL_REGMATCH_SLAB_SLOTS \ 718 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) 719 720 typedef struct regmatch_slab { 721 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; 722 struct regmatch_slab *prev, *next; 723 } regmatch_slab; 724 725 #define PL_reg_flags PL_reg_state.re_state_reg_flags 726 #define PL_bostr PL_reg_state.re_state_bostr 727 #define PL_reginput PL_reg_state.re_state_reginput 728 #define PL_regeol PL_reg_state.re_state_regeol 729 #define PL_regoffs PL_reg_state.re_state_regoffs 730 #define PL_reglastparen PL_reg_state.re_state_reglastparen 731 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen 732 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp 733 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl 734 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set 735 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 736 #define PL_reg_magic PL_reg_state.re_state_reg_magic 737 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos 738 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm 739 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm 740 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved 741 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen 742 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter 743 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter 744 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache 745 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size 746 #define PL_regsize PL_reg_state.re_state_regsize 747 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry 748 #define PL_nrs PL_reg_state.re_state_nrs 749 750 struct re_save_state { 751 U32 re_state_reg_flags; /* from regexec.c */ 752 U32 re_state_reg_start_tmpl; /* from regexec.c */ 753 I32 re_state_reg_eval_set; /* from regexec.c */ 754 bool re_state_reg_match_utf8; /* from regexec.c */ 755 char *re_state_bostr; 756 char *re_state_reginput; /* String-input pointer. */ 757 char *re_state_regeol; /* End of input, for $ check. */ 758 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ 759 U32 *re_state_reglastparen; /* Similarly for lastparen. */ 760 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ 761 char **re_state_reg_start_tmp; /* from regexec.c */ 762 MAGIC *re_state_reg_magic; /* from regexec.c */ 763 PMOP *re_state_reg_oldcurpm; /* from regexec.c */ 764 PMOP *re_state_reg_curpm; /* from regexec.c */ 765 char *re_state_reg_oldsaved; /* old saved substr during match */ 766 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ 767 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ 768 I32 re_state_reg_oldpos; /* from regexec.c */ 769 I32 re_state_reg_maxiter; /* max wait until caching pos */ 770 I32 re_state_reg_leftiter; /* wait until caching pos */ 771 U32 re_state_regsize; /* from regexec.c */ 772 char *re_state_reg_poscache; /* cache of pos of WHILEM */ 773 char *re_state_reg_starttry; /* from regexec.c */ 774 #ifdef PERL_OLD_COPY_ON_WRITE 775 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 */ 776 #endif 777 }; 778 779 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ 780 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) 781 782 /* 783 * Local variables: 784 * c-indentation-style: bsd 785 * c-basic-offset: 4 786 * indent-tabs-mode: t 787 * End: 788 * 789 * ex: set ts=8 sts=4 sw=4 noet: 790 */ 791