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 */
49 #ifdef PERL_OLD_COPY_ON_WRITE
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 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
123 /* Array of offsets for (@-) and (@+) */ \
124 regexp_paren_pair *offs; \
125 /* saved or original string so \digit works forever. */ \
127 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
128 I32 sublen; /* Length of string pointed by subbeg */ \
129 I32 suboffset; /* byte offset of subbeg from logical start of str */ \
130 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
131 /* Information about the match that isn't often used */ \
132 /* offset from wrapped to the start of precomp */ \
133 PERL_BITFIELD32 pre_prefix:4; \
134 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
136 typedef struct regexp {
141 #define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
143 /* used for high speed searches */
144 typedef struct re_scream_pos_data_s
146 char **scream_olds; /* match pos */
147 I32 *scream_pos; /* Internal iterator of scream. */
148 } re_scream_pos_data;
150 /* regexp_engine structure. This is the dispatch table for regexes.
151 * Any regex engine implementation must be able to build one of these.
153 typedef struct regexp_engine {
154 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
155 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
156 char* strbeg, I32 minend, SV* screamer,
157 void* data, U32 flags);
158 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
159 char *strend, const U32 flags,
160 re_scream_pos_data *data);
161 SV* (*checkstr) (pTHX_ REGEXP * const rx);
162 void (*free) (pTHX_ REGEXP * const rx);
163 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
165 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
166 SV const * const value);
167 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
169 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
170 SV * const value, const U32 flags);
171 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
173 SV* (*qr_package)(pTHX_ REGEXP * const rx);
175 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
177 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
178 OP *expr, const struct regexp_engine* eng,
180 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
184 These are passed to the numbered capture variable callbacks as the
185 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
188 #define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
189 #define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
190 #define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
191 #define RX_BUFF_IDX_PREMATCH -2 /* $` */
192 #define RX_BUFF_IDX_POSTMATCH -1 /* $' */
193 #define RX_BUFF_IDX_FULLMATCH 0 /* $& */
196 Flags that are passed to the named_buff and named_buff_iter
197 callbacks above. Those routines are called from universal.c via the
198 Tie::Hash::NamedCapture interface for %+ and %- and the re::
199 functions in the same file.
202 /* The Tie::Hash::NamedCapture operation this is part of, if any */
203 #define RXapif_FETCH 0x0001
204 #define RXapif_STORE 0x0002
205 #define RXapif_DELETE 0x0004
206 #define RXapif_CLEAR 0x0008
207 #define RXapif_EXISTS 0x0010
208 #define RXapif_SCALAR 0x0020
209 #define RXapif_FIRSTKEY 0x0040
210 #define RXapif_NEXTKEY 0x0080
212 /* Whether %+ or %- is being operated on */
213 #define RXapif_ONE 0x0100 /* %+ */
214 #define RXapif_ALL 0x0200 /* %- */
216 /* Whether this is being called from a re:: function */
217 #define RXapif_REGNAME 0x0400
218 #define RXapif_REGNAMES 0x0800
219 #define RXapif_REGNAMES_COUNT 0x1000
222 =head1 REGEXP Functions
224 =for apidoc Am|REGEXP *|SvRX|SV *sv
226 Convenience macro to get the REGEXP from a SV. This is approximately
227 equivalent to the following snippet:
232 sv = MUTABLE_SV(SvRV(sv));
233 if (SvTYPE(sv) == SVt_REGEXP)
236 NULL will be returned if a REGEXP* is not found.
238 =for apidoc Am|bool|SvRXOK|SV* sv
240 Returns a boolean indicating whether the SV (or the one it references)
243 If you want to do something with the REGEXP* later use SvRX instead
249 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
250 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
253 /* Flags stored in regexp->extflags
254 * These are used by code external to the regexp engine
256 * Note that the flags whose names start with RXf_PMf_ are defined in
257 * op_reg_common.h, being copied from the parallel flags of op_pmflags
259 * NOTE: if you modify any RXf flags you should run regen.pl or
260 * regen/regcomp.pl so that regnodes.h is updated with the changes.
264 #include "op_reg_common.h"
266 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
268 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
269 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
270 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
271 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
272 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
274 /* Note, includes charset ones, assumes 0 is the default for them */
275 #define STD_PMMOD_FLAGS_CLEAR(pmfl) \
276 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
278 /* chars and strings used as regex pattern modifiers
279 * Singular is a 'c'har, plural is a "string"
281 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
282 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
283 * for its own uses. So 'k' is out as well.
285 #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
286 #define EXEC_PAT_MOD 'e'
287 #define KEEPCOPY_PAT_MOD 'p'
288 #define ONCE_PAT_MOD 'o'
289 #define GLOBAL_PAT_MOD 'g'
290 #define CONTINUE_PAT_MOD 'c'
291 #define MULTILINE_PAT_MOD 'm'
292 #define SINGLE_PAT_MOD 's'
293 #define IGNORE_PAT_MOD 'i'
294 #define XTENDED_PAT_MOD 'x'
295 #define NONDESTRUCT_PAT_MOD 'r'
296 #define LOCALE_PAT_MOD 'l'
297 #define UNICODE_PAT_MOD 'u'
298 #define DEPENDS_PAT_MOD 'd'
299 #define ASCII_RESTRICT_PAT_MOD 'a'
301 #define ONCE_PAT_MODS "o"
302 #define KEEPCOPY_PAT_MODS "p"
303 #define EXEC_PAT_MODS "e"
304 #define LOOP_PAT_MODS "gc"
305 #define NONDESTRUCT_PAT_MODS "r"
306 #define LOCALE_PAT_MODS "l"
307 #define UNICODE_PAT_MODS "u"
308 #define DEPENDS_PAT_MODS "d"
309 #define ASCII_RESTRICT_PAT_MODS "a"
310 #define ASCII_MORE_RESTRICT_PAT_MODS "aa"
312 /* This string is expected by regcomp.c to be ordered so that the first
313 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
314 * character is bit +1, etc. */
315 #define STD_PAT_MODS "msix"
317 #define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
319 /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
320 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
321 * extflags; the next character is in bit +1, etc. */
322 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
324 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
325 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
326 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
327 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
330 * NOTE: if you modify any RXf flags you should run regen.pl or
331 * regen/regcomp.pl so that regnodes.h is updated with the changes.
335 /* Leave some space, so future bit allocations can go either in the shared or
336 * unshared area without affecting binary compatibility */
337 #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
339 /* Manually decorate this function with gcc-style attributes just to
340 * avoid having to restructure the header files and their called order,
341 * as proto.h would have to be included before this file, and isn't */
343 PERL_STATIC_INLINE const char *
344 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
345 __attribute__warn_unused_result__;
347 #define MAX_CHARSET_NAME_LENGTH 2
349 PERL_STATIC_INLINE const char *
350 get_regex_charset_name(const U32 flags, STRLEN* const lenp)
352 /* Returns a string that corresponds to the name of the regex character set
353 * given by 'flags', and *lenp is set the length of that string, which
354 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
357 switch (get_regex_charset(flags)) {
358 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
359 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
360 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
361 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
362 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
364 return ASCII_MORE_RESTRICT_PAT_MODS;
366 return "?"; /* Unknown */
371 Two flags no longer used.
372 RXf_SPLIT used to be set in Perl_pmruntime if op_flags & OPf_SPECIAL,
373 i.e., split. It was used by the regex engine to check whether it should
374 set RXf_SKIPWHITE. Regexp plugins on CPAN also have done the same thing
375 historically, so we leave these flags defined.
379 # define RXf_SKIPWHITE 0
383 /* Anchor and GPOS related stuff */
384 #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
385 #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
386 #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
387 #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
388 #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
389 #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
391 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
392 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
393 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
395 /* What we have seen */
396 #define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
397 #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
398 #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
401 #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
402 #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
405 #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
408 #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
409 #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
410 #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
411 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
413 #define RXf_MODIFIES_VARS (1<<(RXf_BASE_SHIFT+15))
415 /* Copy and tainted info */
416 #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
418 /* during execution: pattern temporarily tainted by executing locale ops;
419 * post-execution: $1 et al are tainted */
420 #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
421 /* this pattern was tainted during compilation */
422 #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
424 /* Flags indicating special patterns */
425 #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
426 #define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
427 #define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
428 #if RXf_BASE_SHIFT+22 > 31
429 # error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
433 * NOTE: if you modify any RXf flags you should run regen.pl or
434 * regen/regcomp.pl so that regnodes.h is updated with the changes.
438 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
439 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
440 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
441 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
442 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
443 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
444 ? RX_MATCH_TAINTED_on(prog) \
445 : RX_MATCH_TAINTED_off(prog))
447 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
448 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
449 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
450 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
451 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
452 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
453 #define RX_MATCH_COPIED_set(prog,t) ((t) \
454 ? RX_MATCH_COPIED_on(prog) \
455 : RX_MATCH_COPIED_off(prog))
457 #define RXp_EXTFLAGS(rx) ((rx)->extflags)
459 /* For source compatibility. We used to store these explicitly. */
460 #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
461 #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
462 /* FIXME? Are we hardcoding too much here and constraining plugin extension
463 writers? Specifically, the value 1 assumes that the wrapped version always
464 has exactly one character at the end, a ')'. Will that always be true? */
465 #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
466 #define RX_WRAPPED(prog) SvPVX(prog)
467 #define RX_WRAPPED_const(prog) SvPVX_const(prog)
468 #define RX_WRAPLEN(prog) SvCUR(prog)
469 #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
470 #define RX_REFCNT(prog) SvREFCNT(prog)
471 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
472 # define RX_EXTFLAGS(prog) \
474 const REGEXP *const _rx_extflags = (prog); \
475 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
476 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
478 # define RX_ENGINE(prog) \
480 const REGEXP *const _rx_engine = (prog); \
481 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
482 &SvANY(_rx_engine)->engine; \
484 # define RX_SUBBEG(prog) \
486 const REGEXP *const _rx_subbeg = (prog); \
487 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
488 &SvANY(_rx_subbeg)->subbeg; \
490 # define RX_SUBOFFSET(prog) \
492 const REGEXP *const _rx_suboffset = (prog); \
493 assert(SvTYPE(_rx_suboffset) == SVt_REGEXP); \
494 &SvANY(_rx_suboffset)->suboffset; \
496 # define RX_SUBCOFFSET(prog) \
498 const REGEXP *const _rx_subcoffset = (prog); \
499 assert(SvTYPE(_rx_subcoffset) == SVt_REGEXP); \
500 &SvANY(_rx_subcoffset)->subcoffset; \
502 # define RX_OFFS(prog) \
504 const REGEXP *const _rx_offs = (prog); \
505 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
506 &SvANY(_rx_offs)->offs; \
508 # define RX_NPARENS(prog) \
510 const REGEXP *const _rx_nparens = (prog); \
511 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
512 &SvANY(_rx_nparens)->nparens; \
515 # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
516 # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
517 # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
518 # define RX_SUBOFFSET(prog) (((struct regexp *)SvANY(prog))->suboffset)
519 # define RX_SUBCOFFSET(prog) (((struct regexp *)SvANY(prog))->subcoffset)
520 # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
521 # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
523 #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
524 #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
525 #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
526 #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
527 #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
528 #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
529 #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
531 #endif /* PLUGGABLE_RE_EXTENSION */
533 /* Stuff that needs to be included in the pluggable extension goes below here */
535 #ifdef PERL_OLD_COPY_ON_WRITE
536 #define RX_MATCH_COPY_FREE(rx) \
537 STMT_START {if (RX_SAVED_COPY(rx)) { \
538 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
540 if (RX_MATCH_COPIED(rx)) { \
541 Safefree(RX_SUBBEG(rx)); \
542 RX_MATCH_COPIED_off(rx); \
545 #define RX_MATCH_COPY_FREE(rx) \
546 STMT_START {if (RX_MATCH_COPIED(rx)) { \
547 Safefree(RX_SUBBEG(rx)); \
548 RX_MATCH_COPIED_off(rx); \
552 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
553 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
554 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
555 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
556 #define RX_MATCH_UTF8_set(prog, t) ((t) \
557 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
558 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
560 /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
561 #define RX_UTF8(prog) SvUTF8(prog)
563 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
564 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
565 #define REXEC_SCREAM 0x04 /* use scream table. */
566 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
567 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
568 /* under REXEC_COPY_STR, it's ok for the
569 * engine (modulo PL_sawamperand etc)
570 * to skip copying ... */
571 #define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
572 #define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
574 #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
575 # define ReREFCNT_inc(re) \
577 /* This is here to generate a casting warning if incorrect. */ \
578 REGEXP *const _rerefcnt_inc = (re); \
579 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
580 SvREFCNT_inc(_rerefcnt_inc); \
583 # define ReREFCNT_dec(re) \
585 /* This is here to generate a casting warning if incorrect. */ \
586 REGEXP *const _rerefcnt_dec = (re); \
587 SvREFCNT_dec(_rerefcnt_dec); \
590 # define ReREFCNT_dec(re) SvREFCNT_dec(re)
591 # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
594 /* FIXME for plugins. */
596 #define FBMcf_TAIL_DOLLAR 1
597 #define FBMcf_TAIL_DOLLARM 2
598 #define FBMcf_TAIL_Z 4
599 #define FBMcf_TAIL_z 8
600 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
602 #define FBMrf_MULTILINE 1
604 /* some basic information about the current match that is created by
605 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
617 /* structures for holding and saving the state maintained by regmatch() */
619 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
620 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
623 typedef I32 CHECKPOINT;
625 typedef struct regmatch_state {
626 int resume_state; /* where to jump to on return */
627 char *locinput; /* where to backtrack in string on failure */
631 /* this is a fake union member that matches the first element
632 * of each member that needs to store positive backtrack
635 struct regmatch_state *prev_yes_state;
638 /* branchlike members */
639 /* this is a fake union member that matches the first elements
640 * of each member that needs to behave like a branch */
642 /* this first element must match u.yes */
643 struct regmatch_state *prev_yes_state;
651 /* the first elements must match u.branchlike */
652 struct regmatch_state *prev_yes_state;
657 regnode *next_branch; /* next branch node */
661 /* the first elements must match u.branchlike */
662 struct regmatch_state *prev_yes_state;
667 U32 accepted; /* how many accepting states left */
668 bool longfold;/* saw a fold with a 1->n char mapping */
669 U16 *jump; /* positive offsets from me */
670 regnode *me; /* Which node am I - needed for jump tries*/
671 U8 *firstpos;/* pos in string of first trie match */
672 U32 firstchars;/* len in chars of firstpos from start */
673 U16 nextword;/* next word to try */
674 U16 topword; /* longest accepted word */
677 /* special types - these members are used to store state for special
678 regops like eval, if/then, lookaround and the markpoint state */
680 /* this first element must match u.yes */
681 struct regmatch_state *prev_yes_state;
682 struct regmatch_state *prev_eval;
683 struct regmatch_state *prev_curlyx;
685 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
686 flip when transitioning between
687 inner and outer rexen */
688 CHECKPOINT cp; /* remember current savestack indexes */
690 U32 close_paren; /* which close bracket is our end */
691 regnode *B; /* the node following us */
695 /* this first element must match u.yes */
696 struct regmatch_state *prev_yes_state;
698 I32 logical; /* saved copy of 'logical' var */
699 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
700 } ifmatch; /* and SUSPEND/UNLESSM */
703 /* this first element must match u.yes */
704 struct regmatch_state *prev_yes_state;
705 struct regmatch_state *prev_mark;
714 /* quantifiers - these members are used for storing state for
715 for the regops used to implement quantifiers */
717 /* this first element must match u.yes */
718 struct regmatch_state *prev_yes_state;
719 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
720 regnode *me; /* the CURLYX node */
721 regnode *B; /* the B node in /A*B/ */
722 CHECKPOINT cp; /* remember current savestack index */
724 int parenfloor;/* how far back to strip paren data */
726 /* these two are modified by WHILEM */
727 int count; /* how many instances of A we've matched */
728 char *lastloc;/* where previous A matched (0-len detect) */
732 /* this first element must match u.yes */
733 struct regmatch_state *prev_yes_state;
734 struct regmatch_state *save_curlyx;
735 CHECKPOINT cp; /* remember current savestack indexes */
737 char *save_lastloc; /* previous curlyx.lastloc */
743 /* this first element must match u.yes */
744 struct regmatch_state *prev_yes_state;
745 int c1, c2; /* case fold search */
749 I32 alen; /* length of first-matched A string */
752 regnode *A, *B; /* the nodes corresponding to /A*B/ */
753 regnode *me; /* the curlym node */
754 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
755 U8 c2_utf8[UTF8_MAXBYTES+1];
763 int c1, c2; /* case fold search */
764 char *maxpos; /* highest possible point in string to match */
765 char *oldloc; /* the previous locinput */
767 int min, max; /* {m,n} */
768 regnode *A, *B; /* the nodes corresponding to /A*B/ */
769 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
770 U8 c2_utf8[UTF8_MAXBYTES+1];
771 } curly; /* and CURLYN/PLUS/STAR */
776 /* how many regmatch_state structs to allocate as a single slab.
777 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
778 * pointers, plus 1 for any mythical malloc overhead. */
780 #define PERL_REGMATCH_SLAB_SLOTS \
781 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
783 typedef struct regmatch_slab {
784 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
785 struct regmatch_slab *prev, *next;
788 #define PL_reg_flags PL_reg_state.re_state_reg_flags
789 #define PL_bostr PL_reg_state.re_state_bostr
790 #define PL_regeol PL_reg_state.re_state_regeol
791 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
792 #define PL_reg_magic PL_reg_state.re_state_reg_magic
793 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
794 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
795 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
796 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
797 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
798 #define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset
799 #define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset
800 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
801 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
802 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
803 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
804 #define PL_regsize PL_reg_state.re_state_regsize
805 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
806 #define PL_nrs PL_reg_state.re_state_nrs
808 struct re_save_state {
809 U32 re_state_reg_flags; /* from regexec.c */
810 bool re_state_eval_setup_done; /* from regexec.c */
811 bool re_state_reg_match_utf8; /* from regexec.c */
812 bool re_reparsing; /* runtime (?{}) fed back into parser */
813 char *re_state_bostr;
814 char *re_state_regeol; /* End of input, for $ check. */
815 MAGIC *re_state_reg_magic; /* from regexec.c */
816 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
817 PMOP *re_state_reg_curpm; /* from regexec.c */
818 char *re_state_reg_oldsaved; /* old saved substr during match */
819 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
820 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */
821 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */
822 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
823 I32 re_state_reg_oldpos; /* from regexec.c */
824 I32 re_state_reg_maxiter; /* max wait until caching pos */
825 I32 re_state_reg_leftiter; /* wait until caching pos */
826 U32 re_state_regsize; /* from regexec.c */
827 char *re_state_reg_poscache; /* cache of pos of WHILEM */
828 char *re_state_reg_starttry; /* from regexec.c */
829 #ifdef PERL_OLD_COPY_ON_WRITE
830 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 */
834 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
835 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
839 * c-indentation-style: bsd
841 * indent-tabs-mode: nil
844 * ex: set ts=8 sts=4 sw=4 et: