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 #include "utf8.h" 22 23 struct regnode { 24 U8 flags; 25 U8 type; 26 U16 next_off; 27 }; 28 29 typedef struct regnode regnode; 30 31 struct reg_substr_data; 32 33 struct reg_data; 34 35 struct regexp_engine; 36 struct regexp; 37 38 struct reg_substr_datum { 39 I32 min_offset; 40 I32 max_offset; 41 SV *substr; /* non-utf8 variant */ 42 SV *utf8_substr; /* utf8 variant */ 43 I32 end_shift; 44 }; 45 struct reg_substr_data { 46 struct reg_substr_datum data[3]; /* Actual array */ 47 }; 48 49 #ifdef PERL_ANY_COW 50 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */ 51 #else 52 #define SV_SAVED_COPY 53 #endif 54 55 /* offsets within a string of a particular /(.)/ capture */ 56 57 typedef struct regexp_paren_pair { 58 I32 start; 59 I32 end; 60 /* 'start_tmp' records a new opening position before the matching end 61 * has been found, so that the old start and end values are still 62 * valid, e.g. 63 * "abc" =~ /(.(?{print "[$1]"}))+/ 64 *outputs [][a][b] 65 * This field is not part of the API. */ 66 I32 start_tmp; 67 } regexp_paren_pair; 68 69 #if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C) 70 #define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output) 71 #define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output) 72 73 /* Subtracting b from a leaves in a everything that was there that isn't in b, 74 * that is the intersection of a with b's complement */ 75 #define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output) 76 #endif 77 78 /* record the position of a (?{...}) within a pattern */ 79 80 struct reg_code_block { 81 STRLEN start; 82 STRLEN end; 83 OP *block; 84 REGEXP *src_regex; 85 }; 86 87 88 /* 89 The regexp/REGEXP struct, see L<perlreapi> for further documentation 90 on the individual fields. The struct is ordered so that the most 91 commonly used fields are placed at the start. 92 93 Any patch that adds items to this struct will need to include 94 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c> 95 (C<pregfree()>). This involves freeing or cloning items in the 96 regexp's data array based on the data item's type. 97 */ 98 99 #define _REGEXP_COMMON \ 100 /* what engine created this regexp? */ \ 101 const struct regexp_engine* engine; \ 102 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \ 103 HV *paren_names; /* Optional hash of paren names */ \ 104 /* Information about the match that the perl core uses to */ \ 105 /* manage things */ \ 106 U32 extflags; /* Flags used both externally and internally */ \ 107 I32 minlen; /* mininum possible number of chars in string to match */\ 108 I32 minlenret; /* mininum possible number of chars in $& */ \ 109 U32 gofs; /* chars left of pos that we search from */ \ 110 /* substring data about strings that must appear in the */ \ 111 /* final match, used for optimisations */ \ 112 struct reg_substr_data *substrs; \ 113 U32 nparens; /* number of capture buffers */ \ 114 /* private engine specific data */ \ 115 U32 intflags; /* Engine Specific Internal flags */ \ 116 void *pprivate; /* Data private to the regex engine which */ \ 117 /* created this object. */ \ 118 /* Data about the last/current match. These are modified */ \ 119 /* during matching */ \ 120 U32 lastparen; /* last open paren matched */ \ 121 U32 lastcloseparen; /* last close paren matched */ \ 122 /* Array of offsets for (@-) and (@+) */ \ 123 regexp_paren_pair *offs; \ 124 /* saved or original string so \digit works forever. */ \ 125 char *subbeg; \ 126 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\ 127 I32 sublen; /* Length of string pointed by subbeg */ \ 128 I32 suboffset; /* byte offset of subbeg from logical start of str */ \ 129 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \ 130 /* Information about the match that isn't often used */ \ 131 /* offset from wrapped to the start of precomp */ \ 132 PERL_BITFIELD32 pre_prefix:4; \ 133 /* original flags used to compile the pattern, may differ */ \ 134 /* from extflags in various ways */ \ 135 PERL_BITFIELD32 compflags:9; \ 136 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */ 137 138 typedef struct regexp { 139 _XPV_HEAD; 140 _REGEXP_COMMON; 141 } regexp; 142 143 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names) 144 145 /* used for high speed searches */ 146 typedef struct re_scream_pos_data_s 147 { 148 char **scream_olds; /* match pos */ 149 I32 *scream_pos; /* Internal iterator of scream. */ 150 } re_scream_pos_data; 151 152 /* regexp_engine structure. This is the dispatch table for regexes. 153 * Any regex engine implementation must be able to build one of these. 154 */ 155 typedef struct regexp_engine { 156 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags); 157 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, 158 char* strbeg, I32 minend, SV* screamer, 159 void* data, U32 flags); 160 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, 161 char *strend, const U32 flags, 162 re_scream_pos_data *data); 163 SV* (*checkstr) (pTHX_ REGEXP * const rx); 164 void (*free) (pTHX_ REGEXP * const rx); 165 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, 166 SV * const sv); 167 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, 168 SV const * const value); 169 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, 170 const I32 paren); 171 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, 172 SV * const value, const U32 flags); 173 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, 174 const U32 flags); 175 SV* (*qr_package)(pTHX_ REGEXP * const rx); 176 #ifdef USE_ITHREADS 177 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); 178 #endif 179 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count, 180 OP *expr, const struct regexp_engine* eng, 181 REGEXP *VOL old_re, 182 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags); 183 } regexp_engine; 184 185 /* 186 These are passed to the numbered capture variable callbacks as the 187 paren name. >= 1 is reserved for actual numbered captures, i.e. $1, 188 $2 etc. 189 */ 190 #define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */ 191 #define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */ 192 #define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */ 193 #define RX_BUFF_IDX_PREMATCH -2 /* $` */ 194 #define RX_BUFF_IDX_POSTMATCH -1 /* $' */ 195 #define RX_BUFF_IDX_FULLMATCH 0 /* $& */ 196 197 /* 198 Flags that are passed to the named_buff and named_buff_iter 199 callbacks above. Those routines are called from universal.c via the 200 Tie::Hash::NamedCapture interface for %+ and %- and the re:: 201 functions in the same file. 202 */ 203 204 /* The Tie::Hash::NamedCapture operation this is part of, if any */ 205 #define RXapif_FETCH 0x0001 206 #define RXapif_STORE 0x0002 207 #define RXapif_DELETE 0x0004 208 #define RXapif_CLEAR 0x0008 209 #define RXapif_EXISTS 0x0010 210 #define RXapif_SCALAR 0x0020 211 #define RXapif_FIRSTKEY 0x0040 212 #define RXapif_NEXTKEY 0x0080 213 214 /* Whether %+ or %- is being operated on */ 215 #define RXapif_ONE 0x0100 /* %+ */ 216 #define RXapif_ALL 0x0200 /* %- */ 217 218 /* Whether this is being called from a re:: function */ 219 #define RXapif_REGNAME 0x0400 220 #define RXapif_REGNAMES 0x0800 221 #define RXapif_REGNAMES_COUNT 0x1000 222 223 /* 224 =head1 REGEXP Functions 225 226 =for apidoc Am|REGEXP *|SvRX|SV *sv 227 228 Convenience macro to get the REGEXP from a SV. This is approximately 229 equivalent to the following snippet: 230 231 if (SvMAGICAL(sv)) 232 mg_get(sv); 233 if (SvROK(sv)) 234 sv = MUTABLE_SV(SvRV(sv)); 235 if (SvTYPE(sv) == SVt_REGEXP) 236 return (REGEXP*) sv; 237 238 NULL will be returned if a REGEXP* is not found. 239 240 =for apidoc Am|bool|SvRXOK|SV* sv 241 242 Returns a boolean indicating whether the SV (or the one it references) 243 is a REGEXP. 244 245 If you want to do something with the REGEXP* later use SvRX instead 246 and check for NULL. 247 248 =cut 249 */ 250 251 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) 252 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) 253 254 255 /* Flags stored in regexp->extflags 256 * These are used by code external to the regexp engine 257 * 258 * Note that the flags whose names start with RXf_PMf_ are defined in 259 * op_reg_common.h, being copied from the parallel flags of op_pmflags 260 * 261 * NOTE: if you modify any RXf flags you should run regen.pl or 262 * regen/regcomp.pl so that regnodes.h is updated with the changes. 263 * 264 */ 265 266 #include "op_reg_common.h" 267 268 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) 269 270 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ 271 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ 272 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ 273 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ 274 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break 275 276 /* Note, includes charset ones, assumes 0 is the default for them */ 277 #define STD_PMMOD_FLAGS_CLEAR(pmfl) \ 278 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET) 279 280 /* chars and strings used as regex pattern modifiers 281 * Singular is a 'c'har, plural is a "string" 282 * 283 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve 284 * for compatibility reasons with Regexp::Common which highjacked (?k:...) 285 * for its own uses. So 'k' is out as well. 286 */ 287 #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */ 288 #define EXEC_PAT_MOD 'e' 289 #define KEEPCOPY_PAT_MOD 'p' 290 #define ONCE_PAT_MOD 'o' 291 #define GLOBAL_PAT_MOD 'g' 292 #define CONTINUE_PAT_MOD 'c' 293 #define MULTILINE_PAT_MOD 'm' 294 #define SINGLE_PAT_MOD 's' 295 #define IGNORE_PAT_MOD 'i' 296 #define XTENDED_PAT_MOD 'x' 297 #define NONDESTRUCT_PAT_MOD 'r' 298 #define LOCALE_PAT_MOD 'l' 299 #define UNICODE_PAT_MOD 'u' 300 #define DEPENDS_PAT_MOD 'd' 301 #define ASCII_RESTRICT_PAT_MOD 'a' 302 303 #define ONCE_PAT_MODS "o" 304 #define KEEPCOPY_PAT_MODS "p" 305 #define EXEC_PAT_MODS "e" 306 #define LOOP_PAT_MODS "gc" 307 #define NONDESTRUCT_PAT_MODS "r" 308 #define LOCALE_PAT_MODS "l" 309 #define UNICODE_PAT_MODS "u" 310 #define DEPENDS_PAT_MODS "d" 311 #define ASCII_RESTRICT_PAT_MODS "a" 312 #define ASCII_MORE_RESTRICT_PAT_MODS "aa" 313 314 /* This string is expected by regcomp.c to be ordered so that the first 315 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next 316 * character is bit +1, etc. */ 317 #define STD_PAT_MODS "msix" 318 319 #define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS 320 321 /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered 322 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of 323 * extflags; the next character is in bit +1, etc. */ 324 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS 325 326 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS 327 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS 328 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS 329 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS 330 331 /* 332 * NOTE: if you modify any RXf flags you should run regen.pl or 333 * regen/regcomp.pl so that regnodes.h is updated with the changes. 334 * 335 */ 336 337 /* Leave some space, so future bit allocations can go either in the shared or 338 * unshared area without affecting binary compatibility */ 339 #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT) 340 341 /* 342 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will 343 be used by regex engines to check whether they should set 344 RXf_SKIPWHITE 345 */ 346 #define RXf_SPLIT (1<<(RXf_BASE_SHIFT-1)) 347 #if RXf_SPLIT != RXf_PMf_SPLIT 348 # error "RXf_SPLIT does not match RXf_PMf_SPLIT" 349 #endif 350 351 /* Manually decorate this function with gcc-style attributes just to 352 * avoid having to restructure the header files and their called order, 353 * as proto.h would have to be included before this file, and isn't */ 354 355 PERL_STATIC_INLINE const char * 356 get_regex_charset_name(const U32 flags, STRLEN* const lenp) 357 __attribute__warn_unused_result__; 358 359 #define MAX_CHARSET_NAME_LENGTH 2 360 361 PERL_STATIC_INLINE const char * 362 get_regex_charset_name(const U32 flags, STRLEN* const lenp) 363 { 364 /* Returns a string that corresponds to the name of the regex character set 365 * given by 'flags', and *lenp is set the length of that string, which 366 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */ 367 368 *lenp = 1; 369 switch (get_regex_charset(flags)) { 370 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS; 371 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS; 372 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS; 373 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS; 374 case REGEX_ASCII_MORE_RESTRICTED_CHARSET: 375 *lenp = 2; 376 return ASCII_MORE_RESTRICT_PAT_MODS; 377 default: 378 return "?"; /* Unknown */ 379 } 380 } 381 382 /* Anchor and GPOS related stuff */ 383 #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0)) 384 #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1)) 385 #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2)) 386 #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3)) 387 #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4)) 388 #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5)) 389 /* two bits here */ 390 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) 391 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) 392 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) 393 394 /* What we have seen */ 395 #define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+6)) 396 #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7)) 397 #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8)) 398 399 /* Special */ 400 #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9)) 401 #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10)) 402 403 /* UTF8 related */ 404 #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) 405 406 /* Intuit related */ 407 #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12)) 408 #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13)) 409 #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14)) 410 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) 411 412 /* Copy and tainted info */ 413 #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16)) 414 415 /* during execution: pattern temporarily tainted by executing locale ops; 416 * post-execution: $1 et al are tainted */ 417 #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17)) 418 /* this pattern was tainted during compilation */ 419 #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18)) 420 421 /* Flags indicating special patterns */ 422 #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */ 423 #define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split " " */ 424 #define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */ 425 #define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */ 426 #if RXf_BASE_SHIFT+22 > 31 427 # error Too many RXf_PMf bits used. See regnodes.h for any spare in middle 428 #endif 429 430 /* 431 * NOTE: if you modify any RXf flags you should run regen.pl or 432 * regen/regcomp.pl so that regnodes.h is updated with the changes. 433 * 434 */ 435 436 #if NO_TAINT_SUPPORT 437 # define RX_ISTAINTED(prog) 0 438 # define RX_TAINT_on(prog) NOOP 439 # define RXp_MATCH_TAINTED(prog) 0 440 # define RX_MATCH_TAINTED(prog) 0 441 # define RXp_MATCH_TAINTED_on(prog) NOOP 442 # define RX_MATCH_TAINTED_on(prog) NOOP 443 # define RX_MATCH_TAINTED_off(prog) NOOP 444 #else 445 # define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED) 446 # define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED) 447 # define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 448 # define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) 449 # define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) 450 # define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) 451 # define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) 452 #endif 453 454 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) 455 #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 456 ? RX_MATCH_TAINTED_on(prog) \ 457 : RX_MATCH_TAINTED_off(prog)) 458 459 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) 460 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) 461 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) 462 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) 463 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 464 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) 465 #define RX_MATCH_COPIED_set(prog,t) ((t) \ 466 ? RX_MATCH_COPIED_on(prog) \ 467 : RX_MATCH_COPIED_off(prog)) 468 469 #define RXp_EXTFLAGS(rx) ((rx)->extflags) 470 #define RXp_COMPFLAGS(rx) ((rx)->compflags) 471 472 /* For source compatibility. We used to store these explicitly. */ 473 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix) 474 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix) 475 /* FIXME? Are we hardcoding too much here and constraining plugin extension 476 writers? Specifically, the value 1 assumes that the wrapped version always 477 has exactly one character at the end, a ')'. Will that always be true? */ 478 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1) 479 #define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv 480 #define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog)) 481 #define RX_WRAPLEN(prog) SvCUR(prog) 482 #define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr) 483 #define RX_REFCNT(prog) SvREFCNT(prog) 484 #define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog)) 485 #define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog)) 486 #define RX_ENGINE(prog) (ReANY(prog)->engine) 487 #define RX_SUBBEG(prog) (ReANY(prog)->subbeg) 488 #define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset) 489 #define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset) 490 #define RX_OFFS(prog) (ReANY(prog)->offs) 491 #define RX_NPARENS(prog) (ReANY(prog)->nparens) 492 #define RX_SUBLEN(prog) (ReANY(prog)->sublen) 493 #define RX_MINLEN(prog) (ReANY(prog)->minlen) 494 #define RX_MINLENRET(prog) (ReANY(prog)->minlenret) 495 #define RX_GOFS(prog) (ReANY(prog)->gofs) 496 #define RX_LASTPAREN(prog) (ReANY(prog)->lastparen) 497 #define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen) 498 #define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy) 499 500 #endif /* PLUGGABLE_RE_EXTENSION */ 501 502 /* Stuff that needs to be included in the pluggable extension goes below here */ 503 504 #ifdef PERL_ANY_COW 505 #define RX_MATCH_COPY_FREE(rx) \ 506 STMT_START {if (RX_SAVED_COPY(rx)) { \ 507 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \ 508 } \ 509 if (RX_MATCH_COPIED(rx)) { \ 510 Safefree(RX_SUBBEG(rx)); \ 511 RX_MATCH_COPIED_off(rx); \ 512 }} STMT_END 513 #else 514 #define RX_MATCH_COPY_FREE(rx) \ 515 STMT_START {if (RX_MATCH_COPIED(rx)) { \ 516 Safefree(RX_SUBBEG(rx)); \ 517 RX_MATCH_COPIED_off(rx); \ 518 }} STMT_END 519 #endif 520 521 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) 522 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) 523 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) 524 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) 525 #define RX_MATCH_UTF8_set(prog, t) ((t) \ 526 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 527 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 528 529 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ 530 #define RX_UTF8(prog) SvUTF8(prog) 531 532 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 533 #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 534 #define REXEC_SCREAM 0x04 /* use scream table. */ 535 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 536 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 537 /* under REXEC_COPY_STR, it's ok for the 538 * engine (modulo PL_sawamperand etc) 539 * to skip copying ... */ 540 #define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */ 541 #define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */ 542 543 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) 544 # define ReREFCNT_inc(re) \ 545 ({ \ 546 /* This is here to generate a casting warning if incorrect. */ \ 547 REGEXP *const _rerefcnt_inc = (re); \ 548 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \ 549 SvREFCNT_inc(_rerefcnt_inc); \ 550 _rerefcnt_inc; \ 551 }) 552 # define ReREFCNT_dec(re) \ 553 ({ \ 554 /* This is here to generate a casting warning if incorrect. */ \ 555 REGEXP *const _rerefcnt_dec = (re); \ 556 SvREFCNT_dec(_rerefcnt_dec); \ 557 }) 558 #else 559 # define ReREFCNT_dec(re) SvREFCNT_dec(re) 560 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re)) 561 #endif 562 #define ReANY(re) S_ReANY((const REGEXP *)(re)) 563 564 /* FIXME for plugins. */ 565 566 #define FBMcf_TAIL_DOLLAR 1 567 #define FBMcf_TAIL_DOLLARM 2 568 #define FBMcf_TAIL_Z 4 569 #define FBMcf_TAIL_z 8 570 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 571 572 #define FBMrf_MULTILINE 1 573 574 /* some basic information about the current match that is created by 575 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ 576 577 typedef struct { 578 REGEXP *prog; 579 char *bol; 580 char *till; 581 SV *sv; 582 char *ganch; 583 char *cutpoint; 584 bool is_utf8_pat; 585 bool warned; /* we have issued a recursion warning; no need for more */ 586 } regmatch_info; 587 588 589 /* structures for holding and saving the state maintained by regmatch() */ 590 591 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH 592 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 593 #endif 594 595 typedef I32 CHECKPOINT; 596 597 typedef struct regmatch_state { 598 int resume_state; /* where to jump to on return */ 599 char *locinput; /* where to backtrack in string on failure */ 600 601 union { 602 603 /* this is a fake union member that matches the first element 604 * of each member that needs to store positive backtrack 605 * information */ 606 struct { 607 struct regmatch_state *prev_yes_state; 608 } yes; 609 610 /* branchlike members */ 611 /* this is a fake union member that matches the first elements 612 * of each member that needs to behave like a branch */ 613 struct { 614 /* this first element must match u.yes */ 615 struct regmatch_state *prev_yes_state; 616 U32 lastparen; 617 U32 lastcloseparen; 618 CHECKPOINT cp; 619 620 } branchlike; 621 622 struct { 623 /* the first elements must match u.branchlike */ 624 struct regmatch_state *prev_yes_state; 625 U32 lastparen; 626 U32 lastcloseparen; 627 CHECKPOINT cp; 628 629 regnode *next_branch; /* next branch node */ 630 } branch; 631 632 struct { 633 /* the first elements must match u.branchlike */ 634 struct regmatch_state *prev_yes_state; 635 U32 lastparen; 636 U32 lastcloseparen; 637 CHECKPOINT cp; 638 639 U32 accepted; /* how many accepting states left */ 640 bool longfold;/* saw a fold with a 1->n char mapping */ 641 U16 *jump; /* positive offsets from me */ 642 regnode *me; /* Which node am I - needed for jump tries*/ 643 U8 *firstpos;/* pos in string of first trie match */ 644 U32 firstchars;/* len in chars of firstpos from start */ 645 U16 nextword;/* next word to try */ 646 U16 topword; /* longest accepted word */ 647 } trie; 648 649 /* special types - these members are used to store state for special 650 regops like eval, if/then, lookaround and the markpoint state */ 651 struct { 652 /* this first element must match u.yes */ 653 struct regmatch_state *prev_yes_state; 654 struct regmatch_state *prev_eval; 655 struct regmatch_state *prev_curlyx; 656 REGEXP *prev_rex; 657 bool saved_utf8_pat; /* saved copy of is_utf8_pat */ 658 CHECKPOINT cp; /* remember current savestack indexes */ 659 CHECKPOINT lastcp; 660 U32 close_paren; /* which close bracket is our end */ 661 regnode *B; /* the node following us */ 662 } eval; 663 664 struct { 665 /* this first element must match u.yes */ 666 struct regmatch_state *prev_yes_state; 667 I32 wanted; 668 I32 logical; /* saved copy of 'logical' var */ 669 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ 670 } ifmatch; /* and SUSPEND/UNLESSM */ 671 672 struct { 673 /* this first element must match u.yes */ 674 struct regmatch_state *prev_yes_state; 675 struct regmatch_state *prev_mark; 676 SV* mark_name; 677 char *mark_loc; 678 } mark; 679 680 struct { 681 int val; 682 } keeper; 683 684 /* quantifiers - these members are used for storing state for 685 for the regops used to implement quantifiers */ 686 struct { 687 /* this first element must match u.yes */ 688 struct regmatch_state *prev_yes_state; 689 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ 690 regnode *me; /* the CURLYX node */ 691 regnode *B; /* the B node in /A*B/ */ 692 CHECKPOINT cp; /* remember current savestack index */ 693 bool minmod; 694 int parenfloor;/* how far back to strip paren data */ 695 696 /* these two are modified by WHILEM */ 697 int count; /* how many instances of A we've matched */ 698 char *lastloc;/* where previous A matched (0-len detect) */ 699 } curlyx; 700 701 struct { 702 /* this first element must match u.yes */ 703 struct regmatch_state *prev_yes_state; 704 struct regmatch_state *save_curlyx; 705 CHECKPOINT cp; /* remember current savestack indexes */ 706 CHECKPOINT lastcp; 707 char *save_lastloc; /* previous curlyx.lastloc */ 708 I32 cache_offset; 709 I32 cache_mask; 710 } whilem; 711 712 struct { 713 /* this first element must match u.yes */ 714 struct regmatch_state *prev_yes_state; 715 int c1, c2; /* case fold search */ 716 CHECKPOINT cp; 717 U32 lastparen; 718 U32 lastcloseparen; 719 I32 alen; /* length of first-matched A string */ 720 I32 count; 721 bool minmod; 722 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 723 regnode *me; /* the curlym node */ 724 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */ 725 U8 c2_utf8[UTF8_MAXBYTES+1]; 726 } curlym; 727 728 struct { 729 U32 paren; 730 CHECKPOINT cp; 731 U32 lastparen; 732 U32 lastcloseparen; 733 int c1, c2; /* case fold search */ 734 char *maxpos; /* highest possible point in string to match */ 735 char *oldloc; /* the previous locinput */ 736 int count; 737 int min, max; /* {m,n} */ 738 regnode *A, *B; /* the nodes corresponding to /A*B/ */ 739 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */ 740 U8 c2_utf8[UTF8_MAXBYTES+1]; 741 } curly; /* and CURLYN/PLUS/STAR */ 742 743 } u; 744 } regmatch_state; 745 746 /* how many regmatch_state structs to allocate as a single slab. 747 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev 748 * pointers, plus 1 for any mythical malloc overhead. */ 749 750 #define PERL_REGMATCH_SLAB_SLOTS \ 751 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) 752 753 typedef struct regmatch_slab { 754 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; 755 struct regmatch_slab *prev, *next; 756 } regmatch_slab; 757 758 #define PL_bostr PL_reg_state.re_state_bostr 759 #define PL_regeol PL_reg_state.re_state_regeol 760 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 761 #define PL_reg_magic PL_reg_state.re_state_reg_magic 762 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos 763 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm 764 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm 765 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved 766 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen 767 #define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset 768 #define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset 769 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter 770 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter 771 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache 772 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size 773 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry 774 #define PL_nrs PL_reg_state.re_state_nrs 775 776 struct re_save_state { 777 bool re_state_eval_setup_done; /* from regexec.c */ 778 bool re_state_reg_match_utf8; /* from regexec.c */ 779 /* Space for U8 */ 780 I32 re_state_reg_oldpos; /* from regexec.c */ 781 I32 re_state_reg_maxiter; /* max wait until caching pos */ 782 I32 re_state_reg_leftiter; /* wait until caching pos */ 783 char *re_state_bostr; 784 char *re_state_regeol; /* End of input, for $ check. */ 785 MAGIC *re_state_reg_magic; /* from regexec.c */ 786 PMOP *re_state_reg_oldcurpm; /* from regexec.c */ 787 PMOP *re_state_reg_curpm; /* from regexec.c */ 788 char *re_state_reg_oldsaved; /* old saved substr during match */ 789 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ 790 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */ 791 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */ 792 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ 793 char *re_state_reg_poscache; /* cache of pos of WHILEM */ 794 char *re_state_reg_starttry; /* from regexec.c */ 795 #ifdef PERL_ANY_COW 796 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 */ 797 #endif 798 }; 799 800 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ 801 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) 802 803 /* 804 * Local variables: 805 * c-indentation-style: bsd 806 * c-basic-offset: 4 807 * indent-tabs-mode: nil 808 * End: 809 * 810 * ex: set ts=8 sts=4 sw=4 et: 811 */ 812