3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
4 * 2005, 2006, 2007, 2008 by Larry Wall and others
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.
12 * Definitions etc. for regexp(3) routines.
14 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
15 * not the System V one.
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'*/
29 typedef struct regnode regnode;
31 struct reg_substr_data;
38 struct reg_substr_datum {
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
45 struct reg_substr_data {
46 struct reg_substr_datum data[3]; /* Actual array */
50 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
55 /* offsets within a string of a particular /(.)/ capture */
57 typedef struct regexp_paren_pair {
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
63 * "abc" =~ /(.(?{print "[$1]"}))+/
65 * This field is not part of the API. */
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)
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)
78 /* record the position of a (?{...}) within a pattern */
80 struct reg_code_block {
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.
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.
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. */ \
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/(?{..})/ */
138 typedef struct regexp {
143 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
145 /* used for high speed searches */
146 typedef struct re_scream_pos_data_s
148 char **scream_olds; /* match pos */
149 I32 *scream_pos; /* Internal iterator of scream. */
150 } re_scream_pos_data;
152 /* regexp_engine structure. This is the dispatch table for regexes.
153 * Any regex engine implementation must be able to build one of these.
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_
163 const char * const strbeg,
167 re_scream_pos_data *data);
168 SV* (*checkstr) (pTHX_ REGEXP * const rx);
169 void (*free) (pTHX_ REGEXP * const rx);
170 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
172 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
173 SV const * const value);
174 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
176 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
177 SV * const value, const U32 flags);
178 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
180 SV* (*qr_package)(pTHX_ REGEXP * const rx);
182 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
184 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
185 OP *expr, const struct regexp_engine* eng,
187 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
191 These are passed to the numbered capture variable callbacks as the
192 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
195 #define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
196 #define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
197 #define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
198 #define RX_BUFF_IDX_PREMATCH -2 /* $` */
199 #define RX_BUFF_IDX_POSTMATCH -1 /* $' */
200 #define RX_BUFF_IDX_FULLMATCH 0 /* $& */
203 Flags that are passed to the named_buff and named_buff_iter
204 callbacks above. Those routines are called from universal.c via the
205 Tie::Hash::NamedCapture interface for %+ and %- and the re::
206 functions in the same file.
209 /* The Tie::Hash::NamedCapture operation this is part of, if any */
210 #define RXapif_FETCH 0x0001
211 #define RXapif_STORE 0x0002
212 #define RXapif_DELETE 0x0004
213 #define RXapif_CLEAR 0x0008
214 #define RXapif_EXISTS 0x0010
215 #define RXapif_SCALAR 0x0020
216 #define RXapif_FIRSTKEY 0x0040
217 #define RXapif_NEXTKEY 0x0080
219 /* Whether %+ or %- is being operated on */
220 #define RXapif_ONE 0x0100 /* %+ */
221 #define RXapif_ALL 0x0200 /* %- */
223 /* Whether this is being called from a re:: function */
224 #define RXapif_REGNAME 0x0400
225 #define RXapif_REGNAMES 0x0800
226 #define RXapif_REGNAMES_COUNT 0x1000
229 =head1 REGEXP Functions
231 =for apidoc Am|REGEXP *|SvRX|SV *sv
233 Convenience macro to get the REGEXP from a SV. This is approximately
234 equivalent to the following snippet:
239 sv = MUTABLE_SV(SvRV(sv));
240 if (SvTYPE(sv) == SVt_REGEXP)
243 NULL will be returned if a REGEXP* is not found.
245 =for apidoc Am|bool|SvRXOK|SV* sv
247 Returns a boolean indicating whether the SV (or the one it references)
250 If you want to do something with the REGEXP* later use SvRX instead
256 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
257 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
260 /* Flags stored in regexp->extflags
261 * These are used by code external to the regexp engine
263 * Note that the flags whose names start with RXf_PMf_ are defined in
264 * op_reg_common.h, being copied from the parallel flags of op_pmflags
266 * NOTE: if you modify any RXf flags you should run regen.pl or
267 * regen/regcomp.pl so that regnodes.h is updated with the changes.
271 #include "op_reg_common.h"
273 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
275 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
276 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
277 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
278 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
279 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
281 /* Note, includes charset ones, assumes 0 is the default for them */
282 #define STD_PMMOD_FLAGS_CLEAR(pmfl) \
283 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
285 /* chars and strings used as regex pattern modifiers
286 * Singular is a 'c'har, plural is a "string"
288 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
289 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
290 * for its own uses. So 'k' is out as well.
292 #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
293 #define EXEC_PAT_MOD 'e'
294 #define KEEPCOPY_PAT_MOD 'p'
295 #define ONCE_PAT_MOD 'o'
296 #define GLOBAL_PAT_MOD 'g'
297 #define CONTINUE_PAT_MOD 'c'
298 #define MULTILINE_PAT_MOD 'm'
299 #define SINGLE_PAT_MOD 's'
300 #define IGNORE_PAT_MOD 'i'
301 #define XTENDED_PAT_MOD 'x'
302 #define NONDESTRUCT_PAT_MOD 'r'
303 #define LOCALE_PAT_MOD 'l'
304 #define UNICODE_PAT_MOD 'u'
305 #define DEPENDS_PAT_MOD 'd'
306 #define ASCII_RESTRICT_PAT_MOD 'a'
308 #define ONCE_PAT_MODS "o"
309 #define KEEPCOPY_PAT_MODS "p"
310 #define EXEC_PAT_MODS "e"
311 #define LOOP_PAT_MODS "gc"
312 #define NONDESTRUCT_PAT_MODS "r"
313 #define LOCALE_PAT_MODS "l"
314 #define UNICODE_PAT_MODS "u"
315 #define DEPENDS_PAT_MODS "d"
316 #define ASCII_RESTRICT_PAT_MODS "a"
317 #define ASCII_MORE_RESTRICT_PAT_MODS "aa"
319 /* This string is expected by regcomp.c to be ordered so that the first
320 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
321 * character is bit +1, etc. */
322 #define STD_PAT_MODS "msix"
324 #define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
326 /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
327 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
328 * extflags; the next character is in bit +1, etc. */
329 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
331 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
332 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
333 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
334 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
337 * NOTE: if you modify any RXf flags you should run regen.pl or
338 * regen/regcomp.pl so that regnodes.h is updated with the changes.
342 /* Leave some space, so future bit allocations can go either in the shared or
343 * unshared area without affecting binary compatibility */
344 #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT)
347 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
348 be used by regex engines to check whether they should set
351 #define RXf_SPLIT (1<<(RXf_BASE_SHIFT-1))
352 #if RXf_SPLIT != RXf_PMf_SPLIT
353 # error "RXf_SPLIT does not match RXf_PMf_SPLIT"
356 /* Manually decorate this function with gcc-style attributes just to
357 * avoid having to restructure the header files and their called order,
358 * as proto.h would have to be included before this file, and isn't */
360 PERL_STATIC_INLINE const char *
361 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
362 __attribute__warn_unused_result__;
364 #define MAX_CHARSET_NAME_LENGTH 2
366 PERL_STATIC_INLINE const char *
367 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
369 /* Returns a string that corresponds to the name of the regex character set
370 * given by 'flags', and *lenp is set the length of that string, which
371 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
374 switch (get_regex_charset(flags)) {
375 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
376 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
377 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
378 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
379 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
381 return ASCII_MORE_RESTRICT_PAT_MODS;
383 return "?"; /* Unknown */
387 /* Anchor and GPOS related stuff */
388 #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
389 #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
390 #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
391 #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
392 #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
393 #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
395 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
396 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
397 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
399 /* What we have seen */
400 #define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+6))
401 #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
402 #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
405 #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
406 #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
409 #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) /* $1 etc are utf8 */
412 #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
413 #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
414 #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
415 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
417 /* Copy and tainted info */
418 #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
420 /* during execution: pattern temporarily tainted by executing locale ops;
421 * post-execution: $1 et al are tainted */
422 #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
423 /* this pattern was tainted during compilation */
424 #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
426 /* Flags indicating special patterns */
427 #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
428 #define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split " " */
429 #define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
430 #define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
431 #if RXf_BASE_SHIFT+22 > 31
432 # error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
436 * NOTE: if you modify any RXf flags you should run regen.pl or
437 * regen/regcomp.pl so that regnodes.h is updated with the changes.
442 # define RX_ISTAINTED(prog) 0
443 # define RX_TAINT_on(prog) NOOP
444 # define RXp_MATCH_TAINTED(prog) 0
445 # define RX_MATCH_TAINTED(prog) 0
446 # define RXp_MATCH_TAINTED_on(prog) NOOP
447 # define RX_MATCH_TAINTED_on(prog) NOOP
448 # define RX_MATCH_TAINTED_off(prog) NOOP
450 # define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
451 # define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
452 # define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
453 # define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
454 # define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
455 # define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
456 # define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
459 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
460 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
461 ? RX_MATCH_TAINTED_on(prog) \
462 : RX_MATCH_TAINTED_off(prog))
464 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
465 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
466 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
467 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
468 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
469 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
470 #define RX_MATCH_COPIED_set(prog,t) ((t) \
471 ? RX_MATCH_COPIED_on(prog) \
472 : RX_MATCH_COPIED_off(prog))
474 #define RXp_EXTFLAGS(rx) ((rx)->extflags)
475 #define RXp_COMPFLAGS(rx) ((rx)->compflags)
477 /* For source compatibility. We used to store these explicitly. */
478 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
479 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
480 /* FIXME? Are we hardcoding too much here and constraining plugin extension
481 writers? Specifically, the value 1 assumes that the wrapped version always
482 has exactly one character at the end, a ')'. Will that always be true? */
483 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
484 #define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
485 #define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
486 #define RX_WRAPLEN(prog) SvCUR(prog)
487 #define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
488 #define RX_REFCNT(prog) SvREFCNT(prog)
489 #define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
490 #define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog))
491 #define RX_ENGINE(prog) (ReANY(prog)->engine)
492 #define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
493 #define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
494 #define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
495 #define RX_OFFS(prog) (ReANY(prog)->offs)
496 #define RX_NPARENS(prog) (ReANY(prog)->nparens)
497 #define RX_SUBLEN(prog) (ReANY(prog)->sublen)
498 #define RX_MINLEN(prog) (ReANY(prog)->minlen)
499 #define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
500 #define RX_GOFS(prog) (ReANY(prog)->gofs)
501 #define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
502 #define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
503 #define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
505 #endif /* PLUGGABLE_RE_EXTENSION */
507 /* Stuff that needs to be included in the pluggable extension goes below here */
510 #define RX_MATCH_COPY_FREE(rx) \
511 STMT_START {if (RX_SAVED_COPY(rx)) { \
512 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
514 if (RX_MATCH_COPIED(rx)) { \
515 Safefree(RX_SUBBEG(rx)); \
516 RX_MATCH_COPIED_off(rx); \
519 #define RX_MATCH_COPY_FREE(rx) \
520 STMT_START {if (RX_MATCH_COPIED(rx)) { \
521 Safefree(RX_SUBBEG(rx)); \
522 RX_MATCH_COPIED_off(rx); \
526 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
527 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
528 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
529 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
530 #define RX_MATCH_UTF8_set(prog, t) ((t) \
531 ? RX_MATCH_UTF8_on(prog) \
532 : RX_MATCH_UTF8_off(prog))
534 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
535 #define RX_UTF8(prog) SvUTF8(prog)
537 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
538 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
539 #define REXEC_SCREAM 0x04 /* use scream table. */
540 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
541 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
542 /* under REXEC_COPY_STR, it's ok for the
543 * engine (modulo PL_sawamperand etc)
544 * to skip copying ... */
545 #define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
546 #define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
548 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
549 # define ReREFCNT_inc(re) \
551 /* This is here to generate a casting warning if incorrect. */ \
552 REGEXP *const _rerefcnt_inc = (re); \
553 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
554 SvREFCNT_inc(_rerefcnt_inc); \
557 # define ReREFCNT_dec(re) \
559 /* This is here to generate a casting warning if incorrect. */ \
560 REGEXP *const _rerefcnt_dec = (re); \
561 SvREFCNT_dec(_rerefcnt_dec); \
564 # define ReREFCNT_dec(re) SvREFCNT_dec(re)
565 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
567 #define ReANY(re) S_ReANY((const REGEXP *)(re))
569 /* FIXME for plugins. */
571 #define FBMcf_TAIL_DOLLAR 1
572 #define FBMcf_TAIL_DOLLARM 2
573 #define FBMcf_TAIL_Z 4
574 #define FBMcf_TAIL_z 8
575 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
577 #define FBMrf_MULTILINE 1
579 struct regmatch_state;
580 struct regmatch_slab;
582 /* like regmatch_info_aux, but contains extra fields only needed if the
583 * pattern contains (?{}). If used, is snuck into the second slot in the
584 * regmatch_state stack at the start of execution */
588 PMOP *curpm; /* saved PL_curpm */
590 SV *saved_copy; /* saved saved_copy field from rex */
592 char *subbeg; /* saved subbeg field from rex */
593 STRLEN sublen; /* saved sublen field from rex */
594 STRLEN suboffset; /* saved suboffset field from rex */
595 STRLEN subcoffset; /* saved subcoffset field from rex */
596 MAGIC *pos_magic; /* pos() magic attached to $_ */
597 I32 pos; /* the original value of pos() in pos_magic */
598 } regmatch_info_aux_eval;
601 /* fields that logically live in regmatch_info, but which need cleaning
602 * up on croak(), and so are instead are snuck into the first slot in
603 * the regmatch_state stack at the start of execution */
606 regmatch_info_aux_eval *info_aux_eval;
607 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
608 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
609 char *poscache; /* S-L cache of fail positions of WHILEMs */
613 /* some basic information about the current match that is created by
614 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
615 * It is allocated as a local var on the stack, so nothing should be
616 * stored in it that needs preserving or clearing up on croak().
617 * For that, see the aux_info and aux_info_eval members of the
618 * regmatch_state union. */
621 REGEXP *prog; /* the regex being executed */
622 const char * strbeg; /* real start of string */
623 char *strend; /* one byte beyond last char of match string */
624 char *till; /* matches shorter than this fail (see minlen arg) */
625 SV *sv; /* the SV string currently being matched */
626 char *ganch; /* position of \G anchor */
627 char *cutpoint; /* (*COMMIT) position (if any) */
628 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
629 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
630 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
631 I32 poscache_iter; /* current countdown from _maxiter to zero */
632 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
633 bool intuit; /* re_intuit_start() is the top-level caller */
634 bool is_utf8_pat; /* regex is utf8 */
635 bool is_utf8_target; /* string being matched is utf8 */
636 bool warned; /* we have issued a recursion warning; no need for more */
640 /* structures for holding and saving the state maintained by regmatch() */
642 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
643 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
646 typedef I32 CHECKPOINT;
648 typedef struct regmatch_state {
649 int resume_state; /* where to jump to on return */
650 char *locinput; /* where to backtrack in string on failure */
654 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
655 * the nest. They aren't saved backtrack state; rather they
656 * represent one or two extra chunks of data that need allocating
657 * at the start of a match. These fields would logically live in
658 * the regmatch_info struct, except that is allocated on the
659 * C stack, and these fields are all things that require cleanup
660 * after a croak(), when the stack is lost.
661 * As a convenience, we just use the first 1 or 2 regmatch_state
662 * slots to store this info, as we will be allocating a slab of
663 * these anyway. Otherwise we'd have to malloc and then free them,
664 * or allocate them on the save stack (where they will get
665 * realloced if the save stack grows).
666 * info_aux contains the extra fields that are always needed;
667 * info_aux_eval contains extra fields that only needed if
668 * the pattern contains code blocks
669 * We split them into two separate structs to avoid increasing
670 * the size of the union.
673 regmatch_info_aux info_aux;
675 regmatch_info_aux_eval info_aux_eval;
677 /* this is a fake union member that matches the first element
678 * of each member that needs to store positive backtrack
681 struct regmatch_state *prev_yes_state;
684 /* branchlike members */
685 /* this is a fake union member that matches the first elements
686 * of each member that needs to behave like a branch */
688 /* this first element must match u.yes */
689 struct regmatch_state *prev_yes_state;
697 /* the first elements must match u.branchlike */
698 struct regmatch_state *prev_yes_state;
703 regnode *next_branch; /* next branch node */
707 /* the first elements must match u.branchlike */
708 struct regmatch_state *prev_yes_state;
713 U32 accepted; /* how many accepting states left */
714 bool longfold;/* saw a fold with a 1->n char mapping */
715 U16 *jump; /* positive offsets from me */
716 regnode *me; /* Which node am I - needed for jump tries*/
717 U8 *firstpos;/* pos in string of first trie match */
718 U32 firstchars;/* len in chars of firstpos from start */
719 U16 nextword;/* next word to try */
720 U16 topword; /* longest accepted word */
723 /* special types - these members are used to store state for special
724 regops like eval, if/then, lookaround and the markpoint state */
726 /* this first element must match u.yes */
727 struct regmatch_state *prev_yes_state;
728 struct regmatch_state *prev_eval;
729 struct regmatch_state *prev_curlyx;
731 CHECKPOINT cp; /* remember current savestack indexes */
733 U32 close_paren; /* which close bracket is our end */
734 regnode *B; /* the node following us */
738 /* this first element must match u.yes */
739 struct regmatch_state *prev_yes_state;
741 I32 logical; /* saved copy of 'logical' var */
742 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
743 } ifmatch; /* and SUSPEND/UNLESSM */
746 /* this first element must match u.yes */
747 struct regmatch_state *prev_yes_state;
748 struct regmatch_state *prev_mark;
757 /* quantifiers - these members are used for storing state for
758 for the regops used to implement quantifiers */
760 /* this first element must match u.yes */
761 struct regmatch_state *prev_yes_state;
762 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
763 regnode *me; /* the CURLYX node */
764 regnode *B; /* the B node in /A*B/ */
765 CHECKPOINT cp; /* remember current savestack index */
767 int parenfloor;/* how far back to strip paren data */
769 /* these two are modified by WHILEM */
770 int count; /* how many instances of A we've matched */
771 char *lastloc;/* where previous A matched (0-len detect) */
775 /* this first element must match u.yes */
776 struct regmatch_state *prev_yes_state;
777 struct regmatch_state *save_curlyx;
778 CHECKPOINT cp; /* remember current savestack indexes */
780 char *save_lastloc; /* previous curlyx.lastloc */
786 /* this first element must match u.yes */
787 struct regmatch_state *prev_yes_state;
788 int c1, c2; /* case fold search */
792 I32 alen; /* length of first-matched A string */
795 regnode *A, *B; /* the nodes corresponding to /A*B/ */
796 regnode *me; /* the curlym node */
797 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
798 U8 c2_utf8[UTF8_MAXBYTES+1];
806 int c1, c2; /* case fold search */
807 char *maxpos; /* highest possible point in string to match */
808 char *oldloc; /* the previous locinput */
810 int min, max; /* {m,n} */
811 regnode *A, *B; /* the nodes corresponding to /A*B/ */
812 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
813 U8 c2_utf8[UTF8_MAXBYTES+1];
814 } curly; /* and CURLYN/PLUS/STAR */
819 /* how many regmatch_state structs to allocate as a single slab.
820 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
821 * pointers, plus 1 for any mythical malloc overhead. */
823 #define PERL_REGMATCH_SLAB_SLOTS \
824 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
826 typedef struct regmatch_slab {
827 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
828 struct regmatch_slab *prev, *next;
835 * c-indentation-style: bsd
837 * indent-tabs-mode: nil
840 * ex: set ts=8 sts=4 sw=4 et: