3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
4 * 2005, 2006, 2007, 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 typedef struct regexp_paren_pair {
59 The regexp/REGEXP struct, see L<perlreapi> for further documentation
60 on the individual fields. The struct is ordered so that the most
61 commonly used fields are placed at the start.
63 Any patch that adds items to this struct will need to include
64 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
65 (C<pregfree()>). This involves freeing or cloning items in the
66 regexp's data array based on the data item's type.
69 typedef struct regexp {
72 /* what engine created this regexp? */
73 const struct regexp_engine* engine;
74 REGEXP *mother_re; /* what re is this a lightweight copy of? */
76 /* Information about the match that the perl core uses to manage things */
77 U32 extflags; /* Flags used both externally and internally */
78 I32 minlen; /* mininum possible length of string to match */
79 I32 minlenret; /* mininum possible length of $& */
80 U32 gofs; /* chars left of pos that we search from */
81 struct reg_substr_data *substrs; /* substring data about strings that must appear
82 in the final match, used for optimisations */
83 U32 nparens; /* number of capture buffers */
85 /* private engine specific data */
86 U32 intflags; /* Engine Specific Internal flags */
87 void *pprivate; /* Data private to the regex engine which
88 created this object. */
90 /* Data about the last/current match. These are modified during matching*/
91 U32 lastparen; /* last open paren matched */
92 U32 lastcloseparen; /* last close paren matched */
93 regexp_paren_pair *swap; /* Swap copy of *offs */
94 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
96 char *subbeg; /* saved or original string
97 so \digit works forever. */
98 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
99 I32 sublen; /* Length of string pointed by subbeg */
102 /* Information about the match that isn't often used */
103 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
104 char *wrapped; /* wrapped version of the pattern */
105 I32 wraplen; /* length of wrapped */
106 unsigned pre_prefix:4; /* offset from wrapped to the start of precomp */
107 unsigned seen_evals:28; /* number of eval groups in the pattern - for security checks */
108 HV *paren_names; /* Optional hash of paren names */
111 /* used for high speed searches */
112 typedef struct re_scream_pos_data_s
114 char **scream_olds; /* match pos */
115 I32 *scream_pos; /* Internal iterator of scream. */
116 } re_scream_pos_data;
118 /* regexp_engine structure. This is the dispatch table for regexes.
119 * Any regex engine implementation must be able to build one of these.
121 typedef struct regexp_engine {
122 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
123 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
124 char* strbeg, I32 minend, SV* screamer,
125 void* data, U32 flags);
126 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
127 char *strend, const U32 flags,
128 re_scream_pos_data *data);
129 SV* (*checkstr) (pTHX_ REGEXP * const rx);
130 void (*free) (pTHX_ REGEXP * const rx);
131 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
133 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
134 SV const * const value);
135 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
137 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
138 SV * const value, const U32 flags);
139 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
141 SV* (*qr_package)(pTHX_ REGEXP * const rx);
143 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
148 These are passed to the numbered capture variable callbacks as the
149 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
152 #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
153 #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
154 #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
157 Flags that are passed to the named_buff and named_buff_iter
158 callbacks above. Those routines are called from universal.c via the
159 Tie::Hash::NamedCapture interface for %+ and %- and the re::
160 functions in the same file.
163 /* The Tie::Hash::NamedCapture operation this is part of, if any */
164 #define RXapif_FETCH 0x0001
165 #define RXapif_STORE 0x0002
166 #define RXapif_DELETE 0x0004
167 #define RXapif_CLEAR 0x0008
168 #define RXapif_EXISTS 0x0010
169 #define RXapif_SCALAR 0x0020
170 #define RXapif_FIRSTKEY 0x0040
171 #define RXapif_NEXTKEY 0x0080
173 /* Whether %+ or %- is being operated on */
174 #define RXapif_ONE 0x0100 /* %+ */
175 #define RXapif_ALL 0x0200 /* %- */
177 /* Whether this is being called from a re:: function */
178 #define RXapif_REGNAME 0x0400
179 #define RXapif_REGNAMES 0x0800
180 #define RXapif_REGNAMES_COUNT 0x1000
183 =head1 REGEXP Functions
185 =for apidoc Am|REGEXP *|SvRX|SV *sv
187 Convenience macro to get the REGEXP from a SV. This is approximately
188 equivalent to the following snippet:
193 (tmpsv = (SV*)SvRV(sv)) &&
194 SvTYPE(tmpsv) == SVt_PVMG &&
195 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
197 return (REGEXP *)tmpmg->mg_obj;
200 NULL will be returned if a REGEXP* is not found.
202 =for apidoc Am|bool|SvRXOK|SV* sv
204 Returns a boolean indicating whether the SV contains qr magic
207 If you want to do something with the REGEXP* later use SvRX instead
213 #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
214 #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
217 /* Flags stored in regexp->extflags
218 * These are used by code external to the regexp engine
220 * Note that flags starting with RXf_PMf_ have exact equivalents
221 * stored in op_pmflags and which are defined in op.h, they are defined
222 * numerically here only for clarity.
224 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
225 * so that regnodes.h is updated with the changes.
229 /* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
230 * If you change these you need to change the equivalent flags in op.h, and
232 #define RXf_PMf_MULTILINE 0x00000001 /* /m */
233 #define RXf_PMf_SINGLELINE 0x00000002 /* /s */
234 #define RXf_PMf_FOLD 0x00000004 /* /i */
235 #define RXf_PMf_EXTENDED 0x00000008 /* /x */
236 #define RXf_PMf_KEEPCOPY 0x00000010 /* /p */
237 #define RXf_PMf_LOCALE 0x00000020 /* use locale */
238 /* these flags are transfered from the PMOP->op_pmflags member during compilation */
239 #define RXf_PMf_STD_PMMOD_SHIFT 0
240 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
241 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
243 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
244 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
245 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
246 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
247 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
249 /* chars and strings used as regex pattern modifiers
250 * Singlular is a 'c'har, plural is a "string"
252 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
253 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
254 * for its own uses. So 'k' is out as well.
256 #define EXEC_PAT_MOD 'e'
257 #define KEEPCOPY_PAT_MOD 'p'
258 #define ONCE_PAT_MOD 'o'
259 #define GLOBAL_PAT_MOD 'g'
260 #define CONTINUE_PAT_MOD 'c'
261 #define MULTILINE_PAT_MOD 'm'
262 #define SINGLE_PAT_MOD 's'
263 #define IGNORE_PAT_MOD 'i'
264 #define XTENDED_PAT_MOD 'x'
266 #define ONCE_PAT_MODS "o"
267 #define KEEPCOPY_PAT_MODS "p"
268 #define EXEC_PAT_MODS "e"
269 #define LOOP_PAT_MODS "gc"
271 #define STD_PAT_MODS "msix"
273 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
275 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
276 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
277 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
278 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
281 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
282 * so that regnodes.h is updated with the changes.
286 /* Anchor and GPOS related stuff */
287 #define RXf_ANCH_BOL 0x00000100
288 #define RXf_ANCH_MBOL 0x00000200
289 #define RXf_ANCH_SBOL 0x00000400
290 #define RXf_ANCH_GPOS 0x00000800
291 #define RXf_GPOS_SEEN 0x00001000
292 #define RXf_GPOS_FLOAT 0x00002000
294 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
295 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
296 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
298 /* What we have seen */
299 #define RXf_LOOKBEHIND_SEEN 0x00004000
300 #define RXf_EVAL_SEEN 0x00008000
301 #define RXf_CANY_SEEN 0x00010000
304 #define RXf_NOSCAN 0x00020000
305 #define RXf_CHECK_ALL 0x00040000
308 #define RXf_UTF8 0x00080000
309 #define RXf_MATCH_UTF8 0x00100000
312 #define RXf_USE_INTUIT_NOML 0x00200000
313 #define RXf_USE_INTUIT_ML 0x00400000
314 #define RXf_INTUIT_TAIL 0x00800000
317 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
318 be used by regex engines to check whether they should set
321 #define RXf_SPLIT 0x01000000
323 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
325 /* Copy and tainted info */
326 #define RXf_COPY_DONE 0x02000000
327 #define RXf_TAINTED_SEEN 0x04000000
328 #define RXf_TAINTED 0x08000000 /* this pattern is tainted */
330 /* Flags indicating special patterns */
331 #define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */
332 #define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */
333 #define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */
334 #define RXf_NULL 0x80000000 /* Pattern is // */
337 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
338 * so that regnodes.h is updated with the changes.
342 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
343 #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
344 #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
345 #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
346 #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
347 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
348 ? RX_MATCH_TAINTED_on(prog) \
349 : RX_MATCH_TAINTED_off(prog))
351 #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
352 #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
353 #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
354 #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
355 #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
356 #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
357 #define RX_MATCH_COPIED_set(prog,t) ((t) \
358 ? RX_MATCH_COPIED_on(prog) \
359 : RX_MATCH_COPIED_off(prog))
361 #define RXp_PRECOMP(rx) ((rx)->wrapped + (rx)->pre_prefix)
362 /* FIXME? Are we hardcoding too much here and constraining plugin extension
363 writers? Specifically, the value 1 assumes that the wrapped version always
364 has exactly one character at the end, a ')'. Will that always be true? */
365 #define RXp_PRELEN(rx) ((rx)->wraplen - (rx)->pre_prefix - 1)
366 #define RXp_WRAPPED(rx) ((rx)->wrapped)
367 #define RXp_WRAPLEN(rx) ((rx)->wraplen)
368 #define RXp_EXTFLAGS(rx) ((rx)->extflags)
370 /* For source compatibility. We used to store these explicitly. */
371 #define RX_PRECOMP(prog) RXp_PRECOMP((struct regexp *)SvANY(prog))
372 #define RX_PRELEN(prog) RXp_PRELEN((struct regexp *)SvANY(prog))
373 #define RX_WRAPPED(prog) RXp_WRAPPED((struct regexp *)SvANY(prog))
374 #define RX_WRAPLEN(prog) RXp_WRAPLEN((struct regexp *)SvANY(prog))
375 #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
376 #define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
377 #define RX_REFCNT(prog) SvREFCNT(prog)
378 #define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
379 #define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
380 #define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
381 #define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
382 #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
383 #define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
384 #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
385 #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
386 #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
387 #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
388 #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
389 #define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
391 #endif /* PLUGGABLE_RE_EXTENSION */
393 /* Stuff that needs to be included in the plugable extension goes below here */
395 #ifdef PERL_OLD_COPY_ON_WRITE
396 #define RX_MATCH_COPY_FREE(rx) \
397 STMT_START {if (rx->saved_copy) { \
398 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
400 if (RX_MATCH_COPIED(rx)) { \
401 Safefree(RX_SUBBEG(rx)); \
402 RX_MATCH_COPIED_off(rx); \
405 #define RX_MATCH_COPY_FREE(rx) \
406 STMT_START {if (RX_MATCH_COPIED(rx)) { \
407 Safefree(RX_SUBBEG(rx)); \
408 RX_MATCH_COPIED_off(rx); \
412 #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
413 #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
414 #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
415 #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
416 #define RX_MATCH_UTF8_set(prog, t) ((t) \
417 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
418 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
420 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
421 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
422 #define REXEC_SCREAM 0x04 /* use scream table. */
423 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
424 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
426 #if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC)
427 # define ReREFCNT_inc(re) \
429 /* This is here to generate a casting warning if incorrect. */ \
430 REGEXP *const zwapp = (re); \
431 SvREFCNT_inc(zwapp); \
433 # define ReREFCNT_dec(re) \
435 /* This is here to generate a casting warning if incorrect. */ \
436 REGEXP *const boff = (re); \
437 SvREFCNT_dec(boff); \
440 # define ReREFCNT_dec(re) SvREFCNT_dec(re)
441 # define ReREFCNT_inc(re) SvREFCNT_inc(re)
444 /* FIXME for plugins. */
446 #define FBMcf_TAIL_DOLLAR 1
447 #define FBMcf_TAIL_DOLLARM 2
448 #define FBMcf_TAIL_Z 4
449 #define FBMcf_TAIL_z 8
450 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
452 #define FBMrf_MULTILINE 1
454 /* an accepting state/position*/
455 struct _reg_trie_accepted {
459 typedef struct _reg_trie_accepted reg_trie_accepted;
461 /* some basic information about the current match that is created by
462 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
474 /* structures for holding and saving the state maintained by regmatch() */
476 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
477 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
480 typedef I32 CHECKPOINT;
482 typedef struct regmatch_state {
483 int resume_state; /* where to jump to on return */
484 char *locinput; /* where to backtrack in string on failure */
488 /* this is a fake union member that matches the first element
489 * of each member that needs to store positive backtrack
492 struct regmatch_state *prev_yes_state;
495 /* branchlike members */
496 /* this is a fake union member that matches the first elements
497 * of each member that needs to behave like a branch */
499 /* this first element must match u.yes */
500 struct regmatch_state *prev_yes_state;
507 /* the first elements must match u.branchlike */
508 struct regmatch_state *prev_yes_state;
512 regnode *next_branch; /* next branch node */
516 /* the first elements must match u.branchlike */
517 struct regmatch_state *prev_yes_state;
521 reg_trie_accepted *accept_buff; /* accepting states we have seen */
522 U32 accepted; /* how many accepting states we have seen */
523 U16 *jump; /* positive offsets from me */
524 regnode *B; /* node following the trie */
525 regnode *me; /* Which node am I - needed for jump tries*/
528 /* special types - these members are used to store state for special
529 regops like eval, if/then, lookaround and the markpoint state */
531 /* this first element must match u.yes */
532 struct regmatch_state *prev_yes_state;
533 struct regmatch_state *prev_eval;
534 struct regmatch_state *prev_curlyx;
536 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
537 flip when transitioning between
538 inner and outer rexen */
539 CHECKPOINT cp; /* remember current savestack indexes */
541 U32 close_paren; /* which close bracket is our end */
542 regnode *B; /* the node following us */
546 /* this first element must match u.yes */
547 struct regmatch_state *prev_yes_state;
549 I32 logical; /* saved copy of 'logical' var */
550 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
551 } ifmatch; /* and SUSPEND/UNLESSM */
554 /* this first element must match u.yes */
555 struct regmatch_state *prev_yes_state;
556 struct regmatch_state *prev_mark;
565 /* quantifiers - these members are used for storing state for
566 for the regops used to implement quantifiers */
568 /* this first element must match u.yes */
569 struct regmatch_state *prev_yes_state;
570 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
571 regnode *A, *B; /* the nodes corresponding to /A*B/ */
572 CHECKPOINT cp; /* remember current savestack index */
574 int parenfloor;/* how far back to strip paren data */
575 int min; /* the minimal number of A's to match */
576 int max; /* the maximal number of A's to match */
578 /* these two are modified by WHILEM */
579 int count; /* how many instances of A we've matched */
580 char *lastloc;/* where previous A matched (0-len detect) */
584 /* this first element must match u.yes */
585 struct regmatch_state *prev_yes_state;
586 struct regmatch_state *save_curlyx;
587 CHECKPOINT cp; /* remember current savestack indexes */
589 char *save_lastloc; /* previous curlyx.lastloc */
595 /* this first element must match u.yes */
596 struct regmatch_state *prev_yes_state;
597 I32 c1, c2; /* case fold search */
599 I32 alen; /* length of first-matched A string */
602 regnode *A, *B; /* the nodes corresponding to /A*B/ */
603 regnode *me; /* the curlym node */
609 I32 c1, c2; /* case fold search */
610 char *maxpos; /* highest possible point in string to match */
611 char *oldloc; /* the previous locinput */
613 int min, max; /* {m,n} */
614 regnode *A, *B; /* the nodes corresponding to /A*B/ */
615 } curly; /* and CURLYN/PLUS/STAR */
620 /* how many regmatch_state structs to allocate as a single slab.
621 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
622 * pointers, plus 1 for any mythical malloc overhead. */
624 #define PERL_REGMATCH_SLAB_SLOTS \
625 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
627 typedef struct regmatch_slab {
628 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
629 struct regmatch_slab *prev, *next;
632 #define PL_reg_flags PL_reg_state.re_state_reg_flags
633 #define PL_bostr PL_reg_state.re_state_bostr
634 #define PL_reginput PL_reg_state.re_state_reginput
635 #define PL_regeol PL_reg_state.re_state_regeol
636 #define PL_regoffs PL_reg_state.re_state_regoffs
637 #define PL_reglastparen PL_reg_state.re_state_reglastparen
638 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
639 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
640 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
641 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
642 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
643 #define PL_reg_magic PL_reg_state.re_state_reg_magic
644 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
645 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
646 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
647 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
648 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
649 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
650 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
651 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
652 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
653 #define PL_regsize PL_reg_state.re_state_regsize
654 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
655 #define PL_nrs PL_reg_state.re_state_nrs
657 struct re_save_state {
658 U32 re_state_reg_flags; /* from regexec.c */
659 U32 re_state_reg_start_tmpl; /* from regexec.c */
660 I32 re_state_reg_eval_set; /* from regexec.c */
661 bool re_state_reg_match_utf8; /* from regexec.c */
662 char *re_state_bostr;
663 char *re_state_reginput; /* String-input pointer. */
664 char *re_state_regeol; /* End of input, for $ check. */
665 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
666 U32 *re_state_reglastparen; /* Similarly for lastparen. */
667 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
668 char **re_state_reg_start_tmp; /* from regexec.c */
669 MAGIC *re_state_reg_magic; /* from regexec.c */
670 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
671 PMOP *re_state_reg_curpm; /* from regexec.c */
672 char *re_state_reg_oldsaved; /* old saved substr during match */
673 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
674 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
675 I32 re_state_reg_oldpos; /* from regexec.c */
676 I32 re_state_reg_maxiter; /* max wait until caching pos */
677 I32 re_state_reg_leftiter; /* wait until caching pos */
678 U32 re_state_regsize; /* from regexec.c */
679 char *re_state_reg_poscache; /* cache of pos of WHILEM */
680 char *re_state_reg_starttry; /* from regexec.c */
681 #ifdef PERL_OLD_COPY_ON_WRITE
682 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 */
686 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
687 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
691 * c-indentation-style: bsd
693 * indent-tabs-mode: t
696 * ex: set ts=8 sts=4 sw=4 noet: