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'*/
27 typedef struct regnode regnode;
29 struct reg_substr_data;
36 struct reg_substr_datum {
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
43 struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
47 #ifdef PERL_OLD_COPY_ON_WRITE
48 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
53 /* offsets within a string of a particular /(.)/ capture */
55 typedef struct regexp_paren_pair {
58 /* 'start_tmp' records a new opening position before the matching end
59 * has been found, so that the old start and end values are still
61 * "abc" =~ /(.(?{print "[$1]"}))+/
63 * This field is not part of the API. */
67 #if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
68 #define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
69 #define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
71 /* Subtracting b from a leaves in a everything that was there that isn't in b,
72 * that is the intersection of a with b's complement */
73 #define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
76 /* record the position of a (?{...}) within a pattern */
78 struct reg_code_block {
87 The regexp/REGEXP struct, see L<perlreapi> for further documentation
88 on the individual fields. The struct is ordered so that the most
89 commonly used fields are placed at the start.
91 Any patch that adds items to this struct will need to include
92 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
93 (C<pregfree()>). This involves freeing or cloning items in the
94 regexp's data array based on the data item's type.
97 #define _REGEXP_COMMON \
98 /* what engine created this regexp? */ \
99 const struct regexp_engine* engine; \
100 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
101 HV *paren_names; /* Optional hash of paren names */ \
102 /* Information about the match that the perl core uses to */ \
103 /* manage things */ \
104 U32 extflags; /* Flags used both externally and internally */ \
105 I32 minlen; /* mininum possible length of string to match */\
106 I32 minlenret; /* mininum possible length of $& */ \
107 U32 gofs; /* chars left of pos that we search from */ \
108 /* substring data about strings that must appear in the */ \
109 /* final match, used for optimisations */ \
110 struct reg_substr_data *substrs; \
111 U32 nparens; /* number of capture buffers */ \
112 /* private engine specific data */ \
113 U32 intflags; /* Engine Specific Internal flags */ \
114 void *pprivate; /* Data private to the regex engine which */ \
115 /* created this object. */ \
116 /* Data about the last/current match. These are modified */ \
117 /* during matching */ \
118 U32 lastparen; /* last open paren matched */ \
119 U32 lastcloseparen; /* last close paren matched */ \
120 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
121 /* Array of offsets for (@-) and (@+) */ \
122 regexp_paren_pair *offs; \
123 /* saved or original string so \digit works forever. */ \
125 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
126 I32 sublen; /* Length of string pointed by subbeg */ \
127 I32 suboffset; /* byte offset of subbeg from logical start of str */ \
128 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
129 /* Information about the match that isn't often used */ \
130 /* offset from wrapped to the start of precomp */ \
131 PERL_BITFIELD32 pre_prefix:4; \
132 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
134 typedef struct regexp {
139 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
141 /* used for high speed searches */
142 typedef struct re_scream_pos_data_s
144 char **scream_olds; /* match pos */
145 I32 *scream_pos; /* Internal iterator of scream. */
146 } re_scream_pos_data;
148 /* regexp_engine structure. This is the dispatch table for regexes.
149 * Any regex engine implementation must be able to build one of these.
151 typedef struct regexp_engine {
152 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
153 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
154 char* strbeg, I32 minend, SV* screamer,
155 void* data, U32 flags);
156 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
157 char *strend, const U32 flags,
158 re_scream_pos_data *data);
159 SV* (*checkstr) (pTHX_ REGEXP * const rx);
160 void (*free) (pTHX_ REGEXP * const rx);
161 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
163 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
164 SV const * const value);
165 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
167 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
168 SV * const value, const U32 flags);
169 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
171 SV* (*qr_package)(pTHX_ REGEXP * const rx);
173 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
175 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
176 OP *expr, const struct regexp_engine* eng,
178 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
182 These are passed to the numbered capture variable callbacks as the
183 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
186 #define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
187 #define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
188 #define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
189 #define RX_BUFF_IDX_PREMATCH -2 /* $` */
190 #define RX_BUFF_IDX_POSTMATCH -1 /* $' */
191 #define RX_BUFF_IDX_FULLMATCH 0 /* $& */
194 Flags that are passed to the named_buff and named_buff_iter
195 callbacks above. Those routines are called from universal.c via the
196 Tie::Hash::NamedCapture interface for %+ and %- and the re::
197 functions in the same file.
200 /* The Tie::Hash::NamedCapture operation this is part of, if any */
201 #define RXapif_FETCH 0x0001
202 #define RXapif_STORE 0x0002
203 #define RXapif_DELETE 0x0004
204 #define RXapif_CLEAR 0x0008
205 #define RXapif_EXISTS 0x0010
206 #define RXapif_SCALAR 0x0020
207 #define RXapif_FIRSTKEY 0x0040
208 #define RXapif_NEXTKEY 0x0080
210 /* Whether %+ or %- is being operated on */
211 #define RXapif_ONE 0x0100 /* %+ */
212 #define RXapif_ALL 0x0200 /* %- */
214 /* Whether this is being called from a re:: function */
215 #define RXapif_REGNAME 0x0400
216 #define RXapif_REGNAMES 0x0800
217 #define RXapif_REGNAMES_COUNT 0x1000
220 =head1 REGEXP Functions
222 =for apidoc Am|REGEXP *|SvRX|SV *sv
224 Convenience macro to get the REGEXP from a SV. This is approximately
225 equivalent to the following snippet:
230 sv = MUTABLE_SV(SvRV(sv));
231 if (SvTYPE(sv) == SVt_REGEXP)
234 NULL will be returned if a REGEXP* is not found.
236 =for apidoc Am|bool|SvRXOK|SV* sv
238 Returns a boolean indicating whether the SV (or the one it references)
241 If you want to do something with the REGEXP* later use SvRX instead
247 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
248 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
251 /* Flags stored in regexp->extflags
252 * These are used by code external to the regexp engine
254 * Note that the flags whose names start with RXf_PMf_ are defined in
255 * op_reg_common.h, being copied from the parallel flags of op_pmflags
257 * NOTE: if you modify any RXf flags you should run regen.pl or
258 * regen/regcomp.pl so that regnodes.h is updated with the changes.
262 #include "op_reg_common.h"
264 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
266 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
267 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
268 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
269 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
270 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
272 /* Note, includes charset ones, assumes 0 is the default for them */
273 #define STD_PMMOD_FLAGS_CLEAR(pmfl) \
274 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
276 /* chars and strings used as regex pattern modifiers
277 * Singular is a 'c'har, plural is a "string"
279 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
280 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
281 * for its own uses. So 'k' is out as well.
283 #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
284 #define EXEC_PAT_MOD 'e'
285 #define KEEPCOPY_PAT_MOD 'p'
286 #define ONCE_PAT_MOD 'o'
287 #define GLOBAL_PAT_MOD 'g'
288 #define CONTINUE_PAT_MOD 'c'
289 #define MULTILINE_PAT_MOD 'm'
290 #define SINGLE_PAT_MOD 's'
291 #define IGNORE_PAT_MOD 'i'
292 #define XTENDED_PAT_MOD 'x'
293 #define NONDESTRUCT_PAT_MOD 'r'
294 #define LOCALE_PAT_MOD 'l'
295 #define UNICODE_PAT_MOD 'u'
296 #define DEPENDS_PAT_MOD 'd'
297 #define ASCII_RESTRICT_PAT_MOD 'a'
299 #define ONCE_PAT_MODS "o"
300 #define KEEPCOPY_PAT_MODS "p"
301 #define EXEC_PAT_MODS "e"
302 #define LOOP_PAT_MODS "gc"
303 #define NONDESTRUCT_PAT_MODS "r"
304 #define LOCALE_PAT_MODS "l"
305 #define UNICODE_PAT_MODS "u"
306 #define DEPENDS_PAT_MODS "d"
307 #define ASCII_RESTRICT_PAT_MODS "a"
308 #define ASCII_MORE_RESTRICT_PAT_MODS "aa"
310 /* This string is expected by regcomp.c to be ordered so that the first
311 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
312 * character is bit +1, etc. */
313 #define STD_PAT_MODS "msix"
315 #define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
317 /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
318 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
319 * extflags; the next character is in bit +1, etc. */
320 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
322 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
323 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
324 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
325 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
328 * NOTE: if you modify any RXf flags you should run regen.pl or
329 * regen/regcomp.pl so that regnodes.h is updated with the changes.
333 /* Leave some space, so future bit allocations can go either in the shared or
334 * unshared area without affecting binary compatibility */
335 #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
337 /* Manually decorate this function with gcc-style attributes just to
338 * avoid having to restructure the header files and their called order,
339 * as proto.h would have to be included before this file, and isn't */
341 PERL_STATIC_INLINE const char *
342 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
343 __attribute__warn_unused_result__;
345 #define MAX_CHARSET_NAME_LENGTH 2
347 PERL_STATIC_INLINE const char *
348 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
350 /* Returns a string that corresponds to the name of the regex character set
351 * given by 'flags', and *lenp is set the length of that string, which
352 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
355 switch (get_regex_charset(flags)) {
356 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
357 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
358 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
359 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
360 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
362 return ASCII_MORE_RESTRICT_PAT_MODS;
364 return "?"; /* Unknown */
368 /* Anchor and GPOS related stuff */
369 #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
370 #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
371 #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
372 #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
373 #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
374 #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
376 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
377 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
378 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
380 /* What we have seen */
381 #define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
382 #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
383 #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
386 #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
387 #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
390 #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
393 #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
394 #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
395 #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
398 This used to be set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e.
399 split. It was used by the regex engine to check whether it should set
400 RXf_SKIPWHITE. Regexp plugins on CPAN also have done the same thing
401 historically, so we leave this flag defined, even though it is never set.
403 #if !defined(PERL_CORE) || defined(PERL_IN_DUMP_C)
404 # define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
407 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
409 /* Copy and tainted info */
410 #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
412 /* during execution: pattern temporarily tainted by executing locale ops;
413 * post-execution: $1 et al are tainted */
414 #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
415 /* this pattern was tainted during compilation */
416 #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
418 /* Flags indicating special patterns */
419 #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
420 /* No longer used, but CPAN modules still mention it. */
421 #if !defined(PERL_CORE) || defined(PERL_IN_DUMP_C)
422 # 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
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.
436 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
437 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
438 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
439 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
440 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
441 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
442 ? RX_MATCH_TAINTED_on(prog) \
443 : RX_MATCH_TAINTED_off(prog))
445 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
446 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
447 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
448 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
449 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
450 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
451 #define RX_MATCH_COPIED_set(prog,t) ((t) \
452 ? RX_MATCH_COPIED_on(prog) \
453 : RX_MATCH_COPIED_off(prog))
455 #define RXp_EXTFLAGS(rx) ((rx)->extflags)
457 /* For source compatibility. We used to store these explicitly. */
458 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
459 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
460 /* FIXME? Are we hardcoding too much here and constraining plugin extension
461 writers? Specifically, the value 1 assumes that the wrapped version always
462 has exactly one character at the end, a ')'. Will that always be true? */
463 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
464 #define RX_WRAPPED(prog) SvPVX(prog)
465 #define RX_WRAPPED_const(prog) SvPVX_const(prog)
466 #define RX_WRAPLEN(prog) SvCUR(prog)
467 #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
468 #define RX_REFCNT(prog) SvREFCNT(prog)
469 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
470 # define RX_EXTFLAGS(prog) \
472 const REGEXP *const _rx_extflags = (prog); \
473 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
474 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
476 # define RX_ENGINE(prog) \
478 const REGEXP *const _rx_engine = (prog); \
479 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
480 &SvANY(_rx_engine)->engine; \
482 # define RX_SUBBEG(prog) \
484 const REGEXP *const _rx_subbeg = (prog); \
485 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
486 &SvANY(_rx_subbeg)->subbeg; \
488 # define RX_SUBOFFSET(prog) \
490 const REGEXP *const _rx_suboffset = (prog); \
491 assert(SvTYPE(_rx_suboffset) == SVt_REGEXP); \
492 &SvANY(_rx_suboffset)->suboffset; \
494 # define RX_SUBCOFFSET(prog) \
496 const REGEXP *const _rx_subcoffset = (prog); \
497 assert(SvTYPE(_rx_subcoffset) == SVt_REGEXP); \
498 &SvANY(_rx_subcoffset)->subcoffset; \
500 # define RX_OFFS(prog) \
502 const REGEXP *const _rx_offs = (prog); \
503 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
504 &SvANY(_rx_offs)->offs; \
506 # define RX_NPARENS(prog) \
508 const REGEXP *const _rx_nparens = (prog); \
509 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
510 &SvANY(_rx_nparens)->nparens; \
513 # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
514 # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
515 # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
516 # define RX_SUBOFFSET(prog) (((struct regexp *)SvANY(prog))->suboffset)
517 # define RX_SUBCOFFSET(prog) (((struct regexp *)SvANY(prog))->subcoffset)
518 # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
519 # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
521 #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
522 #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
523 #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
524 #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
525 #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
526 #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
527 #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
529 #endif /* PLUGGABLE_RE_EXTENSION */
531 /* Stuff that needs to be included in the pluggable extension goes below here */
533 #ifdef PERL_OLD_COPY_ON_WRITE
534 #define RX_MATCH_COPY_FREE(rx) \
535 STMT_START {if (RX_SAVED_COPY(rx)) { \
536 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
538 if (RX_MATCH_COPIED(rx)) { \
539 Safefree(RX_SUBBEG(rx)); \
540 RX_MATCH_COPIED_off(rx); \
543 #define RX_MATCH_COPY_FREE(rx) \
544 STMT_START {if (RX_MATCH_COPIED(rx)) { \
545 Safefree(RX_SUBBEG(rx)); \
546 RX_MATCH_COPIED_off(rx); \
550 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
551 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
552 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
553 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
554 #define RX_MATCH_UTF8_set(prog, t) ((t) \
555 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
556 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
558 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
559 #define RX_UTF8(prog) SvUTF8(prog)
561 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
562 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
563 #define REXEC_SCREAM 0x04 /* use scream table. */
564 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
565 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
566 /* under REXEC_COPY_STR, it's ok for the
567 * engine (modulo PL_sawamperand etc)
568 * to skip copying ... */
569 #define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
570 #define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
572 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
573 # define ReREFCNT_inc(re) \
575 /* This is here to generate a casting warning if incorrect. */ \
576 REGEXP *const _rerefcnt_inc = (re); \
577 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
578 SvREFCNT_inc(_rerefcnt_inc); \
581 # define ReREFCNT_dec(re) \
583 /* This is here to generate a casting warning if incorrect. */ \
584 REGEXP *const _rerefcnt_dec = (re); \
585 SvREFCNT_dec(_rerefcnt_dec); \
588 # define ReREFCNT_dec(re) SvREFCNT_dec(re)
589 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
592 /* FIXME for plugins. */
594 #define FBMcf_TAIL_DOLLAR 1
595 #define FBMcf_TAIL_DOLLARM 2
596 #define FBMcf_TAIL_Z 4
597 #define FBMcf_TAIL_z 8
598 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
600 #define FBMrf_MULTILINE 1
602 /* some basic information about the current match that is created by
603 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
615 /* structures for holding and saving the state maintained by regmatch() */
617 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
618 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
621 typedef I32 CHECKPOINT;
623 typedef struct regmatch_state {
624 int resume_state; /* where to jump to on return */
625 char *locinput; /* where to backtrack in string on failure */
629 /* this is a fake union member that matches the first element
630 * of each member that needs to store positive backtrack
633 struct regmatch_state *prev_yes_state;
636 /* branchlike members */
637 /* this is a fake union member that matches the first elements
638 * of each member that needs to behave like a branch */
640 /* this first element must match u.yes */
641 struct regmatch_state *prev_yes_state;
649 /* the first elements must match u.branchlike */
650 struct regmatch_state *prev_yes_state;
655 regnode *next_branch; /* next branch node */
659 /* the first elements must match u.branchlike */
660 struct regmatch_state *prev_yes_state;
665 U32 accepted; /* how many accepting states left */
666 bool longfold;/* saw a fold with a 1->n char mapping */
667 U16 *jump; /* positive offsets from me */
668 regnode *me; /* Which node am I - needed for jump tries*/
669 U8 *firstpos;/* pos in string of first trie match */
670 U32 firstchars;/* len in chars of firstpos from start */
671 U16 nextword;/* next word to try */
672 U16 topword; /* longest accepted word */
675 /* special types - these members are used to store state for special
676 regops like eval, if/then, lookaround and the markpoint state */
678 /* this first element must match u.yes */
679 struct regmatch_state *prev_yes_state;
680 struct regmatch_state *prev_eval;
681 struct regmatch_state *prev_curlyx;
683 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
684 flip when transitioning between
685 inner and outer rexen */
686 CHECKPOINT cp; /* remember current savestack indexes */
688 U32 close_paren; /* which close bracket is our end */
689 regnode *B; /* the node following us */
693 /* this first element must match u.yes */
694 struct regmatch_state *prev_yes_state;
696 I32 logical; /* saved copy of 'logical' var */
697 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
698 } ifmatch; /* and SUSPEND/UNLESSM */
701 /* this first element must match u.yes */
702 struct regmatch_state *prev_yes_state;
703 struct regmatch_state *prev_mark;
712 /* quantifiers - these members are used for storing state for
713 for the regops used to implement quantifiers */
715 /* this first element must match u.yes */
716 struct regmatch_state *prev_yes_state;
717 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
718 regnode *me; /* the CURLYX node */
719 regnode *B; /* the B node in /A*B/ */
720 CHECKPOINT cp; /* remember current savestack index */
722 int parenfloor;/* how far back to strip paren data */
724 /* these two are modified by WHILEM */
725 int count; /* how many instances of A we've matched */
726 char *lastloc;/* where previous A matched (0-len detect) */
730 /* this first element must match u.yes */
731 struct regmatch_state *prev_yes_state;
732 struct regmatch_state *save_curlyx;
733 CHECKPOINT cp; /* remember current savestack indexes */
735 char *save_lastloc; /* previous curlyx.lastloc */
741 /* this first element must match u.yes */
742 struct regmatch_state *prev_yes_state;
743 I32 c1, c2; /* case fold search */
747 I32 alen; /* length of first-matched A string */
750 regnode *A, *B; /* the nodes corresponding to /A*B/ */
751 regnode *me; /* the curlym node */
759 I32 c1, c2; /* case fold search */
760 char *maxpos; /* highest possible point in string to match */
761 char *oldloc; /* the previous locinput */
763 int min, max; /* {m,n} */
764 regnode *A, *B; /* the nodes corresponding to /A*B/ */
765 } curly; /* and CURLYN/PLUS/STAR */
770 /* how many regmatch_state structs to allocate as a single slab.
771 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
772 * pointers, plus 1 for any mythical malloc overhead. */
774 #define PERL_REGMATCH_SLAB_SLOTS \
775 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
777 typedef struct regmatch_slab {
778 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
779 struct regmatch_slab *prev, *next;
782 #define PL_reg_flags PL_reg_state.re_state_reg_flags
783 #define PL_bostr PL_reg_state.re_state_bostr
784 #define PL_regeol PL_reg_state.re_state_regeol
785 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
786 #define PL_reg_magic PL_reg_state.re_state_reg_magic
787 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
788 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
789 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
790 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
791 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
792 #define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset
793 #define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset
794 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
795 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
796 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
797 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
798 #define PL_regsize PL_reg_state.re_state_regsize
799 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
800 #define PL_nrs PL_reg_state.re_state_nrs
802 struct re_save_state {
803 U32 re_state_reg_flags; /* from regexec.c */
804 bool re_state_eval_setup_done; /* from regexec.c */
805 bool re_state_reg_match_utf8; /* from regexec.c */
806 bool re_reparsing; /* runtime (?{}) fed back into parser */
807 char *re_state_bostr;
808 char *re_state_regeol; /* End of input, for $ check. */
809 MAGIC *re_state_reg_magic; /* from regexec.c */
810 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
811 PMOP *re_state_reg_curpm; /* from regexec.c */
812 char *re_state_reg_oldsaved; /* old saved substr during match */
813 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
814 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */
815 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */
816 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
817 I32 re_state_reg_oldpos; /* from regexec.c */
818 I32 re_state_reg_maxiter; /* max wait until caching pos */
819 I32 re_state_reg_leftiter; /* wait until caching pos */
820 U32 re_state_regsize; /* from regexec.c */
821 char *re_state_reg_poscache; /* cache of pos of WHILEM */
822 char *re_state_reg_starttry; /* from regexec.c */
823 #ifdef PERL_OLD_COPY_ON_WRITE
824 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 */
828 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
829 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
833 * c-indentation-style: bsd
835 * indent-tabs-mode: nil
838 * ex: set ts=8 sts=4 sw=4 et: