This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
add the missing =back
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
1129b882 4 * 2005, 2006, 2007, 2008 by Larry Wall and others
d6376244
JH
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
a0d0e21e
LW
9 */
10
378cc40b
LW
11/*
12 * Definitions etc. for regexp(3) routines.
13 *
14 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
15 * not the System V one.
16 */
be8e71aa 17#ifndef PLUGGABLE_RE_EXTENSION
785a26d5
YO
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'*/
378cc40b 20
79a2a0e8
KW
21#include "utf8.h"
22
f55b7b26
KW
23typedef SSize_t regnode_offset;
24
c277df42
IZ
25struct regnode {
26 U8 flags;
27 U8 type;
28 U16 next_off;
29};
30
31typedef struct regnode regnode;
32
cad2e5aa 33struct reg_substr_data;
2779dcf1 34
0ee3c8d0
JH
35struct reg_data;
36
f9f4320a 37struct regexp_engine;
28d8d7f4 38struct regexp;
bbe252da 39
785a26d5 40struct reg_substr_datum {
c4828ca0
DM
41 SSize_t min_offset; /* min pos (in chars) that substr must appear */
42 SSize_t max_offset /* max pos (in chars) that substr must appear */;
785a26d5
YO
43 SV *substr; /* non-utf8 variant */
44 SV *utf8_substr; /* utf8 variant */
c4828ca0 45 SSize_t end_shift; /* how many fixed chars must end the string */
785a26d5
YO
46};
47struct reg_substr_data {
6480a6c4 48 U8 check_ix; /* index into data[] of check substr */
785a26d5
YO
49 struct reg_substr_datum data[3]; /* Actual array */
50};
f9f4320a 51
db2c6cb3 52#ifdef PERL_ANY_COW
bbe252da
YO
53#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
54#else
55#define SV_SAVED_COPY
ed252734 56#endif
28d8d7f4 57
b3fd53f3
DM
58/* offsets within a string of a particular /(.)/ capture */
59
f0ab9afb 60typedef struct regexp_paren_pair {
99a90e59
FC
61 SSize_t start;
62 SSize_t end;
b3fd53f3
DM
63 /* 'start_tmp' records a new opening position before the matching end
64 * has been found, so that the old start and end values are still
65 * valid, e.g.
66 * "abc" =~ /(.(?{print "[$1]"}))+/
67 *outputs [][a][b]
68 * This field is not part of the API. */
ea3daa5d 69 SSize_t start_tmp;
f0ab9afb 70} regexp_paren_pair;
28d8d7f4 71
52ae8f7e 72#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 73#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 74#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
75
76/* Subtracting b from a leaves in a everything that was there that isn't in b,
77 * that is the intersection of a with b's complement */
78#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
79#endif
80
3d2bd50a
DM
81/* record the position of a (?{...}) within a pattern */
82
83struct reg_code_block {
84 STRLEN start;
85 STRLEN end;
86 OP *block;
b30fcab9 87 REGEXP *src_regex;
3d2bd50a
DM
88};
89
1acab4c5
DM
90/* array of reg_code_block's plus header info */
91
92struct reg_code_blocks {
f8def6c7 93 int refcnt; /* we may be pointed to from a regex and from the savestack */
1acab4c5
DM
94 int count; /* how many code blocks */
95 struct reg_code_block *cb; /* array of reg_code_block's */
96};
97
3d2bd50a 98
882227b7
AB
99/*
100 The regexp/REGEXP struct, see L<perlreapi> for further documentation
101 on the individual fields. The struct is ordered so that the most
102 commonly used fields are placed at the start.
103
104 Any patch that adds items to this struct will need to include
105 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
106 (C<pregfree()>). This involves freeing or cloning items in the
107 regexp's data array based on the data item's type.
108*/
109
bbe252da 110typedef struct regexp {
e1168c11
DM
111 _XPV_HEAD;
112 const struct regexp_engine* engine; /* what engine created this regexp? */
113 REGEXP *mother_re; /* what re is this a lightweight copy of? */
114 HV *paren_names; /* Optional hash of paren names */
115
116 /*----------------------------------------------------------------------
117 * Information about the match that the perl core uses to manage things
118 */
119
120 U32 extflags; /* Flags used both externally and internally */
121 SSize_t minlen; /* minimum possible number of chars in string to match */
122 SSize_t minlenret; /* mininum possible number of chars in $& */
123 STRLEN gofs; /* chars left of pos that we search from */
124 /* substring data about strings that must appear in
125 * the final match, used for optimisations */
126 struct reg_substr_data *substrs;
127 U32 nparens; /* number of capture buffers */
128
129 /* private engine specific data */
130
131 U32 intflags; /* Engine Specific Internal flags */
132 void *pprivate; /* Data private to the regex engine which
133 * created this object. */
134
135 /*----------------------------------------------------------------------
136 * Data about the last/current match. These are modified during matching
137 */
138
0b9dad94
DM
139 U32 lastparen; /* highest close paren matched ($+) */
140 U32 lastcloseparen; /* last close paren matched ($^N) */
e1168c11
DM
141 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
142 char **recurse_locinput; /* used to detect infinite recursion, XXX: move to internal */
143
144 /*---------------------------------------------------------------------- */
145
146 char *subbeg; /* saved or original string so \digit works forever. */
147 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
148 SSize_t sublen; /* Length of string pointed by subbeg */
149 SSize_t suboffset; /* byte offset of subbeg from logical start of str */
150 SSize_t subcoffset; /* suboffset equiv, but in chars (for @-/@+) */
e1168c11
DM
151 SSize_t maxlen; /* minimum possible number of chars in string to match */
152
153 /*---------------------------------------------------------------------- */
154
155 /* offset from wrapped to the start of precomp */
156 PERL_BITFIELD32 pre_prefix:4;
157
158 /* original flags used to compile the pattern, may differ from
159 * extflags in various ways */
160 PERL_BITFIELD32 compflags:9;
161
162 /*---------------------------------------------------------------------- */
163
164 CV *qr_anoncv; /* the anon sub wrapped round qr/(?{..})/ */
f8fc2ecf
YO
165} regexp;
166
e1168c11 167
89f6f287 168#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 169
785a26d5 170/* used for high speed searches */
f9f4320a
YO
171typedef struct re_scream_pos_data_s
172{
173 char **scream_olds; /* match pos */
ea3daa5d 174 SSize_t *scream_pos; /* Internal iterator of scream. */
f9f4320a
YO
175} re_scream_pos_data;
176
785a26d5
YO
177/* regexp_engine structure. This is the dispatch table for regexes.
178 * Any regex engine implementation must be able to build one of these.
179 */
f9f4320a 180typedef struct regexp_engine {
1593ad57 181 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc 182 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
ea3daa5d 183 char* strbeg, SSize_t minend, SV* sv,
49d7dfbc 184 void* data, U32 flags);
52a21eb3
DM
185 char* (*intuit) (pTHX_
186 REGEXP * const rx,
187 SV *sv,
188 const char * const strbeg,
189 char *strpos,
190 char *strend,
191 const U32 flags,
9f61653a 192 re_scream_pos_data *data);
49d7dfbc 193 SV* (*checkstr) (pTHX_ REGEXP * const rx);
fc6bde6f 194 void (*rxfree) (pTHX_ REGEXP * const rx);
2fdbfb4d 195 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 196 SV * const sv);
2fdbfb4d
AB
197 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
198 SV const * const value);
199 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
200 const I32 paren);
192b9cd1
AB
201 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
202 SV * const value, const U32 flags);
203 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
204 const U32 flags);
49d7dfbc 205 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 206#ifdef USE_ITHREADS
49d7dfbc 207 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 208#endif
3c13cae6
DM
209 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
210 OP *expr, const struct regexp_engine* eng,
7984ae89 211 REGEXP *old_re,
76ac488f 212 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
213} regexp_engine;
214
192b9cd1
AB
215/*
216 These are passed to the numbered capture variable callbacks as the
217 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
218 $2 etc.
219*/
2c7b5d76
DM
220#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
221#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
222#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
223#define RX_BUFF_IDX_PREMATCH -2 /* $` */
224#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
225#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
226
227/*
228 Flags that are passed to the named_buff and named_buff_iter
229 callbacks above. Those routines are called from universal.c via the
230 Tie::Hash::NamedCapture interface for %+ and %- and the re::
231 functions in the same file.
232*/
233
234/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
235#define RXapif_FETCH 0x0001
236#define RXapif_STORE 0x0002
237#define RXapif_DELETE 0x0004
238#define RXapif_CLEAR 0x0008
239#define RXapif_EXISTS 0x0010
240#define RXapif_SCALAR 0x0020
241#define RXapif_FIRSTKEY 0x0040
242#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
243
244/* Whether %+ or %- is being operated on */
f1b875a0
YO
245#define RXapif_ONE 0x0100 /* %+ */
246#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
247
248/* Whether this is being called from a re:: function */
f1b875a0
YO
249#define RXapif_REGNAME 0x0400
250#define RXapif_REGNAMES 0x0800
7948fc08 251#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 252
f7e71195
AB
253/*
254=head1 REGEXP Functions
255
256=for apidoc Am|REGEXP *|SvRX|SV *sv
257
72d33970 258Convenience macro to get the REGEXP from a SV. This is approximately
f7e71195
AB
259equivalent to the following snippet:
260
261 if (SvMAGICAL(sv))
262 mg_get(sv);
1af910ea
DL
263 if (SvROK(sv))
264 sv = MUTABLE_SV(SvRV(sv));
265 if (SvTYPE(sv) == SVt_REGEXP)
266 return (REGEXP*) sv;
f7e71195 267
796b6530 268C<NULL> will be returned if a REGEXP* is not found.
f7e71195
AB
269
270=for apidoc Am|bool|SvRXOK|SV* sv
271
1af910ea
DL
272Returns a boolean indicating whether the SV (or the one it references)
273is a REGEXP.
f7e71195
AB
274
275If you want to do something with the REGEXP* later use SvRX instead
276and check for NULL.
277
278=cut
279*/
280
281#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
8298454c 282#define SvRXOK(sv) cBOOL(Perl_get_re_arg(aTHX_ sv))
f7e71195
AB
283
284
7948fc08 285/* Flags stored in regexp->extflags
bbe252da 286 * These are used by code external to the regexp engine
e357fc67 287 *
5b126c84
KW
288 * Note that the flags whose names start with RXf_PMf_ are defined in
289 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 290 *
eb2624c9
FC
291 * NOTE: if you modify any RXf flags you should run regen.pl or
292 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 293 *
bbe252da
YO
294 */
295
5b126c84
KW
296#include "op_reg_common.h"
297
77c8f263 298#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_EXTENDED_MORE|RXf_PMf_NOCAPTURE)
cde0cee5 299
cc4d09e1
KW
300#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl, x_count) \
301 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
302 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
303 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
77c8f263
KW
304 case XTENDED_PAT_MOD: if (x_count == 0) { \
305 *(pmfl) |= RXf_PMf_EXTENDED; \
306 *(pmfl) &= ~RXf_PMf_EXTENDED_MORE; \
307 } \
308 else { \
309 *(pmfl) |= RXf_PMf_EXTENDED \
310 |RXf_PMf_EXTENDED_MORE; \
311 } \
312 (x_count)++; break; \
41d7c59e 313 case NOCAPTURE_PAT_MOD: *(pmfl) |= RXf_PMf_NOCAPTURE; break;
cc4d09e1 314
94b03d7d 315/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 316#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
77c8f263 317 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_EXTENDED_MORE|RXf_PMf_CHARSET|RXf_PMf_NOCAPTURE)
fb85c044 318
bcdf7404 319/* chars and strings used as regex pattern modifiers
486ec47a 320 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
321 *
322 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
323 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
324 * for its own uses. So 'k' is out as well.
bcdf7404 325 */
85508812 326#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 327#define EXEC_PAT_MOD 'e'
87e95b7f 328#define KEEPCOPY_PAT_MOD 'p'
41d7c59e 329#define NOCAPTURE_PAT_MOD 'n'
bcdf7404
YO
330#define ONCE_PAT_MOD 'o'
331#define GLOBAL_PAT_MOD 'g'
332#define CONTINUE_PAT_MOD 'c'
333#define MULTILINE_PAT_MOD 'm'
334#define SINGLE_PAT_MOD 's'
335#define IGNORE_PAT_MOD 'i'
336#define XTENDED_PAT_MOD 'x'
4f4d7508 337#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
338#define LOCALE_PAT_MOD 'l'
339#define UNICODE_PAT_MOD 'u'
50e91148 340#define DEPENDS_PAT_MOD 'd'
cfaf538b 341#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
342
343#define ONCE_PAT_MODS "o"
87e95b7f 344#define KEEPCOPY_PAT_MODS "p"
41d7c59e 345#define NOCAPTURE_PAT_MODS "n"
bcdf7404
YO
346#define EXEC_PAT_MODS "e"
347#define LOOP_PAT_MODS "gc"
4f4d7508 348#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
349#define LOCALE_PAT_MODS "l"
350#define UNICODE_PAT_MODS "u"
50e91148 351#define DEPENDS_PAT_MODS "d"
cfaf538b 352#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 353#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 354
78fecfc8 355/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
356 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
357 * character is bit +1, etc. */
41d7c59e 358#define STD_PAT_MODS "msixxn"
bcdf7404 359
94b03d7d
KW
360#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
361
78fecfc8 362/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
363 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
364 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
365#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
366
41d7c59e 367#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS NOCAPTURE_PAT_MODS
94b03d7d 368#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 369#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 370#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 371
cb5027f2 372/*
eb2624c9
FC
373 * NOTE: if you modify any RXf flags you should run regen.pl or
374 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
375 *
376 */
bcdf7404 377
dbc200c5 378/*
5012eebe
DM
379 Set in Perl_pmruntime for a split. Will be used by regex engines to
380 check whether they should set RXf_SKIPWHITE
dbc200c5 381*/
32f62247
KW
382#define RXf_SPLIT RXf_PMf_SPLIT
383
0bd51918
KW
384/* Currently the regex flags occupy a single 32-bit word. Not all bits are
385 * currently used. The lower bits are shared with their corresponding PMf flag
386 * bits, up to but not including _RXf_PMf_SHIFT_NEXT. The unused bits
387 * immediately follow; finally the used RXf-only (unshared) bits, so that the
388 * highest bit in the word is used. This gathers all the unused bits as a pool
389 * in the middle, like so: 11111111111111110000001111111111
390 * where the '1's represent used bits, and the '0's unused. This design allows
391 * us to allocate off one end of the pool if we need to add a shared bit, and
392 * off the other end if we need a non-shared bit, without disturbing the other
393 * bits. This maximizes the likelihood of being able to change things without
394 * breaking binary compatibility.
395 *
396 * To add shared bits, do so in op_reg_common.h. This should change
397 * _RXf_PMf_SHIFT_NEXT so that things won't compile. Then come to regexp.h and
398 * op.h and adjust the constant adders in the definitions of RXf_BASE_SHIFT and
399 * Pmf_BASE_SHIFT down by the number of shared bits you added. That's it.
400 * Things should be binary compatible. But if either of these gets to having
401 * to subtract rather than add, leave at 0 and instead adjust all the entries
402 * that are in terms of it. But if the first one of those is already
403 * RXf_BASE_SHIFT+0, there are no bits left, and a redesign is in order.
404 *
405 * To remove unshared bits, just delete its entry. If you're where breaking
406 * binary compatibility is ok to do, you might want to adjust things to move
407 * the newly opened space so that it gets absorbed into the common pool.
408 *
409 * To add unshared bits, first use up any gaps in the middle. Otherwise,
410 * allocate off the low end until you get to RXf_BASE_SHIFT+0. If that isn't
411 * enough, move RXf_BASE_SHIFT down (if possible) and add the new bit at the
412 * other end instead; this preserves binary compatibility.
413 *
414 * For the regexp bits, PL_reg_extflags_name[] in regnodes.h has a comment
415 * giving which bits are used/unused */
416
d262c0c7 417#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT + 2)
52d81aa8 418
bbe252da 419/* What we have seen */
9cba692b
KW
420#define RXf_NO_INPLACE_SUBST (1U<<(RXf_BASE_SHIFT+2))
421#define RXf_EVAL_SEEN (1U<<(RXf_BASE_SHIFT+3))
bbe252da
YO
422
423/* Special */
9cba692b
KW
424#define RXf_UNBOUNDED_QUANTIFIER_SEEN (1U<<(RXf_BASE_SHIFT+4))
425#define RXf_CHECK_ALL (1U<<(RXf_BASE_SHIFT+5))
bbe252da
YO
426
427/* UTF8 related */
9cba692b 428#define RXf_MATCH_UTF8 (1U<<(RXf_BASE_SHIFT+6)) /* $1 etc are utf8 */
bbe252da
YO
429
430/* Intuit related */
9cba692b
KW
431#define RXf_USE_INTUIT_NOML (1U<<(RXf_BASE_SHIFT+7))
432#define RXf_USE_INTUIT_ML (1U<<(RXf_BASE_SHIFT+8))
433#define RXf_INTUIT_TAIL (1U<<(RXf_BASE_SHIFT+9))
bbe252da
YO
434#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
435
a3b51d37 436/* Do we have some sort of anchor? */
9cba692b 437#define RXf_IS_ANCHORED (1U<<(RXf_BASE_SHIFT+10))
a3b51d37 438
bbe252da 439/* Copy and tainted info */
9cba692b 440#define RXf_COPY_DONE (1U<<(RXf_BASE_SHIFT+11))
ef07e810 441
1738e041 442/* post-execution: $1 et al are tainted */
9cba692b 443#define RXf_TAINTED_SEEN (1U<<(RXf_BASE_SHIFT+12))
ef07e810 444/* this pattern was tainted during compilation */
9cba692b 445#define RXf_TAINTED (1U<<(RXf_BASE_SHIFT+13))
52d81aa8
NC
446
447/* Flags indicating special patterns */
9cba692b
KW
448#define RXf_START_ONLY (1U<<(RXf_BASE_SHIFT+14)) /* Pattern is /^/ */
449#define RXf_SKIPWHITE (1U<<(RXf_BASE_SHIFT+15)) /* Pattern is for a */
450 /* split " " */
451#define RXf_WHITE (1U<<(RXf_BASE_SHIFT+16)) /* Pattern is /\s+/ */
3f6e6b29 452#define RXf_NULL (1U<<(RXf_BASE_SHIFT+17)) /* Pattern is // */
0bd51918
KW
453
454/* See comments at the beginning of these defines about adding bits. The
455 * highest bit position should be used, so that if RXf_BASE_SHIFT gets
456 * increased, the #error below will be triggered so that you will be reminded
457 * to adjust things at the other end to keep the bit positions unchanged */
3f6e6b29 458#if RXf_BASE_SHIFT+17 > 31
0bd51918 459# error Too many RXf_PMf bits used. See comments at beginning of these for what to do
1850c8f9 460#endif
c737faaf 461
cb5027f2 462/*
eb2624c9
FC
463 * NOTE: if you modify any RXf flags you should run regen.pl or
464 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
465 *
466 */
bbe252da 467
dc6d7f5c 468#ifdef NO_TAINT_SUPPORT
a885a8e0 469# define RX_ISTAINTED(rx_sv) 0
196a02af 470# define RXp_ISTAINTED(prog) 0
a885a8e0
DM
471# define RX_TAINT_on(rx_sv) NOOP
472# define RXp_MATCH_TAINTED(prog) 0
473# define RX_MATCH_TAINTED(rx_sv) 0
474# define RXp_MATCH_TAINTED_on(prog) NOOP
475# define RX_MATCH_TAINTED_on(rx_sv) NOOP
71e845c3 476# define RXp_MATCH_TAINTED_off(prog) NOOP
a885a8e0 477# define RX_MATCH_TAINTED_off(rx_sv) NOOP
284167a5 478#else
a885a8e0 479# define RX_ISTAINTED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_TAINTED)
196a02af 480# define RXp_ISTAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED)
a885a8e0
DM
481# define RX_TAINT_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_TAINTED)
482# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
483# define RX_MATCH_TAINTED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_TAINTED_SEEN)
484# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
485# define RX_MATCH_TAINTED_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_TAINTED_SEEN)
196a02af 486# define RXp_MATCH_TAINTED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
a885a8e0 487# define RX_MATCH_TAINTED_off(rx_sv) (RX_EXTFLAGS(rx_sv) &= ~RXf_TAINTED_SEEN)
284167a5
SM
488#endif
489
ea0673ee 490#define RXp_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
a885a8e0
DM
491
492#define RX_MATCH_TAINTED_set(rx_sv, t) ((t) \
493 ? RX_MATCH_TAINTED_on(rx_sv) \
494 : RX_MATCH_TAINTED_off(rx_sv))
495
496#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
497#define RX_MATCH_COPIED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_COPY_DONE)
498#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
499#define RX_MATCH_COPIED_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_COPY_DONE)
500#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
501#define RX_MATCH_COPIED_off(rx_sv) (RX_EXTFLAGS(rx_sv) &= ~RXf_COPY_DONE)
502#define RX_MATCH_COPIED_set(rx_sv,t) ((t) \
503 ? RX_MATCH_COPIED_on(rx_sv) \
504 : RX_MATCH_COPIED_off(rx_sv))
505
506#define RXp_EXTFLAGS(rx) ((rx)->extflags)
507#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 508
07bc277f 509/* For source compatibility. We used to store these explicitly. */
a885a8e0
DM
510#define RX_PRECOMP(rx_sv) (RX_WRAPPED(rx_sv) \
511 + ReANY(rx_sv)->pre_prefix)
512#define RX_PRECOMP_const(rx_sv) (RX_WRAPPED_const(rx_sv) \
513 + ReANY(rx_sv)->pre_prefix)
5509d87a
NC
514/* FIXME? Are we hardcoding too much here and constraining plugin extension
515 writers? Specifically, the value 1 assumes that the wrapped version always
516 has exactly one character at the end, a ')'. Will that always be true? */
a885a8e0
DM
517#define RX_PRELEN(rx_sv) (RX_WRAPLEN(rx_sv) \
518 - ReANY(rx_sv)->pre_prefix - 1)
519
520#define RX_WRAPPED(rx_sv) SvPVX(rx_sv)
521#define RX_WRAPPED_const(rx_sv) SvPVX_const(rx_sv)
522#define RX_WRAPLEN(rx_sv) SvCUR(rx_sv)
523#define RX_CHECK_SUBSTR(rx_sv) (ReANY(rx_sv)->check_substr)
524#define RX_REFCNT(rx_sv) SvREFCNT(rx_sv)
525#define RX_EXTFLAGS(rx_sv) RXp_EXTFLAGS(ReANY(rx_sv))
526#define RX_COMPFLAGS(rx_sv) RXp_COMPFLAGS(ReANY(rx_sv))
196a02af
DM
527#define RXp_ENGINE(prog) ((prog)->engine)
528#define RX_ENGINE(rx_sv) (RXp_ENGINE(ReANY(rx_sv)))
529#define RXp_SUBBEG(prog) (prog->subbeg)
530#define RX_SUBBEG(rx_sv) (RXp_SUBBEG(ReANY(rx_sv)))
531#define RXp_SUBOFFSET(prog) (prog->suboffset)
532#define RX_SUBOFFSET(rx_sv) (RXp_SUBOFFSET(ReANY(rx_sv)))
a885a8e0 533#define RX_SUBCOFFSET(rx_sv) (ReANY(rx_sv)->subcoffset)
196a02af
DM
534#define RXp_OFFS(prog) (prog->offs)
535#define RX_OFFS(rx_sv) (RXp_OFFS(ReANY(rx_sv)))
536#define RXp_NPARENS(prog) (prog->nparens)
537#define RX_NPARENS(rx_sv) (RXp_NPARENS(ReANY(rx_sv)))
a885a8e0 538#define RX_SUBLEN(rx_sv) (ReANY(rx_sv)->sublen)
196a02af
DM
539#define RXp_MINLEN(prog) (prog->minlen)
540#define RX_MINLEN(rx_sv) (RXp_MINLEN(ReANY(rx_sv)))
541#define RXp_MINLENRET(prog) (prog->minlenret)
542#define RX_MINLENRET(rx_sv) (RXp_MINLENRET(ReANY(rx_sv)))
543#define RXp_GOFS(prog) (prog->gofs)
544#define RX_GOFS(rx_sv) (RXp_GOFS(ReANY(rx_sv)))
a885a8e0
DM
545#define RX_LASTPAREN(rx_sv) (ReANY(rx_sv)->lastparen)
546#define RX_LASTCLOSEPAREN(rx_sv) (ReANY(rx_sv)->lastcloseparen)
196a02af
DM
547#define RXp_SAVED_COPY(prog) (prog->saved_copy)
548#define RX_SAVED_COPY(rx_sv) (RXp_SAVED_COPY(ReANY(rx_sv)))
03c83e26 549/* last match was zero-length */
196a02af
DM
550#define RXp_ZERO_LEN(prog) \
551 (RXp_OFFS(prog)[0].start + (SSize_t)RXp_GOFS(prog) \
552 == RXp_OFFS(prog)[0].end)
553#define RX_ZERO_LEN(rx_sv) (RXp_ZERO_LEN(ReANY(rx_sv)))
220fc49f 554
be8e71aa
YO
555#endif /* PLUGGABLE_RE_EXTENSION */
556
486ec47a 557/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 558
db2c6cb3 559#ifdef PERL_ANY_COW
196a02af
DM
560#define RXp_MATCH_COPY_FREE(prog) \
561 STMT_START {if (RXp_SAVED_COPY(prog)) { \
562 SV_CHECK_THINKFIRST_COW_DROP(RXp_SAVED_COPY(prog)); \
ed252734 563 } \
196a02af
DM
564 if (RXp_MATCH_COPIED(prog)) { \
565 Safefree(RXp_SUBBEG(prog)); \
566 RXp_MATCH_COPIED_off(prog); \
ed252734
NC
567 }} STMT_END
568#else
196a02af
DM
569#define RXp_MATCH_COPY_FREE(prog) \
570 STMT_START {if (RXp_MATCH_COPIED(prog)) { \
571 Safefree(RXp_SUBBEG(prog)); \
572 RXp_MATCH_COPIED_off(prog); \
ed252734
NC
573 }} STMT_END
574#endif
a8cb1947 575#define RX_MATCH_COPY_FREE(rx_sv) RXp_MATCH_COPY_FREE(ReANY(rx_sv))
ed252734 576
a885a8e0
DM
577#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
578#define RX_MATCH_UTF8(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_MATCH_UTF8)
196a02af 579#define RXp_MATCH_UTF8_on(prog) (RXp_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
97f6857b 580#define RX_MATCH_UTF8_on(rx_sv) (RXp_MATCH_UTF8_on(ReANY(rx_sv)))
196a02af
DM
581#define RXp_MATCH_UTF8_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
582#define RX_MATCH_UTF8_off(rx_sv) (RXp_MATCH_UTF8_off(ReANY(rx_sv))
583#define RXp_MATCH_UTF8_set(prog, t) ((t) \
584 ? RXp_MATCH_UTF8_on(prog) \
585 : RXp_MATCH_UTF8_off(prog))
586#define RX_MATCH_UTF8_set(rx_sv, t) (RXp_MATCH_UTF8_set(ReANY(rx_sv), t))
efd26800
NC
587
588/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
a885a8e0 589#define RX_UTF8(rx_sv) SvUTF8(rx_sv)
7948fc08 590
a340edde
DM
591
592/* bits in flags arg of Perl_regexec_flags() */
593
594#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
595#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
596#define REXEC_SCREAM 0x04 /* currently unused. */
597#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
598#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
599 no need to copy string again */
600
601 /* under REXEC_COPY_STR, it's ok for the
602 engine (modulo PL_sawamperand etc)
603 to skip copying: ... */
604#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
605#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
606#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
607 the start pos (so s/.\G// would fail
608 on second iteration */
c277df42 609
7fe44985 610#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
611# define ReREFCNT_inc(re) \
612 ({ \
613 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
614 REGEXP *const _rerefcnt_inc = (re); \
615 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
616 SvREFCNT_inc(_rerefcnt_inc); \
617 _rerefcnt_inc; \
288b8c02
NC
618 })
619# define ReREFCNT_dec(re) \
620 ({ \
621 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
622 REGEXP *const _rerefcnt_dec = (re); \
623 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
624 })
625#else
626# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 627# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 628#endif
c9182d9c 629#define ReANY(re) Perl_ReANY((const REGEXP *)(re))
288b8c02
NC
630
631/* FIXME for plugins. */
cf93c79d
IZ
632
633#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
634#define FBMcf_TAIL_DOLLARM 2
635#define FBMcf_TAIL_Z 4
636#define FBMcf_TAIL_z 8
637#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
638
639#define FBMrf_MULTILINE 1
cad2e5aa 640
331b2dcc
DM
641struct regmatch_state;
642struct regmatch_slab;
8adc0f72 643
bf2039a9
DM
644/* like regmatch_info_aux, but contains extra fields only needed if the
645 * pattern contains (?{}). If used, is snuck into the second slot in the
646 * regmatch_state stack at the start of execution */
8adc0f72
DM
647
648typedef struct {
649 regexp *rex;
650 PMOP *curpm; /* saved PL_curpm */
651#ifdef PERL_ANY_COW
652 SV *saved_copy; /* saved saved_copy field from rex */
653#endif
654 char *subbeg; /* saved subbeg field from rex */
655 STRLEN sublen; /* saved sublen field from rex */
656 STRLEN suboffset; /* saved suboffset field from rex */
657 STRLEN subcoffset; /* saved subcoffset field from rex */
1d48e83d 658 SV *sv; /* $_ during (?{}) */
8adc0f72 659 MAGIC *pos_magic; /* pos() magic attached to $_ */
ea3daa5d 660 SSize_t pos; /* the original value of pos() in pos_magic */
25fdce4a 661 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
662} regmatch_info_aux_eval;
663
664
665/* fields that logically live in regmatch_info, but which need cleaning
666 * up on croak(), and so are instead are snuck into the first slot in
667 * the regmatch_state stack at the start of execution */
668
669typedef struct {
670 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
671 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
672 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 673 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
674} regmatch_info_aux;
675
8adc0f72 676
3b0527fe 677/* some basic information about the current match that is created by
bf2039a9
DM
678 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
679 * It is allocated as a local var on the stack, so nothing should be
680 * stored in it that needs preserving or clearing up on croak().
681 * For that, see the aux_info and aux_info_eval members of the
682 * regmatch_state union. */
3b0527fe
DM
683
684typedef struct {
28d03b21 685 REGEXP *prog; /* the regex being executed */
9d9163fb 686 const char * strbeg; /* real start of string */
28d03b21
DM
687 char *strend; /* one byte beyond last char of match string */
688 char *till; /* matches shorter than this fail (see minlen arg) */
689 SV *sv; /* the SV string currently being matched */
690 char *ganch; /* position of \G anchor */
691 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
692 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
693 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
694 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
695 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 696 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 697 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
698 bool is_utf8_pat; /* regex is utf8 */
699 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 700 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
701} regmatch_info;
702
5d9a96ca
DM
703
704/* structures for holding and saving the state maintained by regmatch() */
705
d56b3014 706#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
ba6840fb 707#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 10
d56b3014 708#endif
6bda09f9 709
5d9a96ca
DM
710typedef I32 CHECKPOINT;
711
a0374537 712typedef struct regmatch_state {
40a82448 713 int resume_state; /* where to jump to on return */
24d3c4a9 714 char *locinput; /* where to backtrack in string on failure */
fd1dd2eb 715 char *loceol;
8e9f3eef 716 U8 *sr0; /* position of start of script run, or NULL */
e822a8b4
DM
717
718 union {
77cb431f 719
bf2039a9
DM
720 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
721 * the nest. They aren't saved backtrack state; rather they
722 * represent one or two extra chunks of data that need allocating
723 * at the start of a match. These fields would logically live in
724 * the regmatch_info struct, except that is allocated on the
725 * C stack, and these fields are all things that require cleanup
726 * after a croak(), when the stack is lost.
727 * As a convenience, we just use the first 1 or 2 regmatch_state
728 * slots to store this info, as we will be allocating a slab of
729 * these anyway. Otherwise we'd have to malloc and then free them,
730 * or allocate them on the save stack (where they will get
731 * realloced if the save stack grows).
732 * info_aux contains the extra fields that are always needed;
733 * info_aux_eval contains extra fields that only needed if
734 * the pattern contains code blocks
735 * We split them into two separate structs to avoid increasing
736 * the size of the union.
737 */
738
739 regmatch_info_aux info_aux;
740
741 regmatch_info_aux_eval info_aux_eval;
742
77cb431f
DM
743 /* this is a fake union member that matches the first element
744 * of each member that needs to store positive backtrack
745 * information */
746 struct {
747 struct regmatch_state *prev_yes_state;
748 } yes;
749
fae667d5
YO
750 /* branchlike members */
751 /* this is a fake union member that matches the first elements
752 * of each member that needs to behave like a branch */
753 struct {
5d458dd8
YO
754 /* this first element must match u.yes */
755 struct regmatch_state *prev_yes_state;
fae667d5 756 U32 lastparen;
f6033a9d 757 U32 lastcloseparen;
fae667d5
YO
758 CHECKPOINT cp;
759
760 } branchlike;
761
762 struct {
763 /* the first elements must match u.branchlike */
764 struct regmatch_state *prev_yes_state;
765 U32 lastparen;
f6033a9d 766 U32 lastcloseparen;
fae667d5
YO
767 CHECKPOINT cp;
768
769 regnode *next_branch; /* next branch node */
770 } branch;
771
772 struct {
773 /* the first elements must match u.branchlike */
774 struct regmatch_state *prev_yes_state;
775 U32 lastparen;
f6033a9d 776 U32 lastcloseparen;
fae667d5
YO
777 CHECKPOINT cp;
778
2e64971a 779 U32 accepted; /* how many accepting states left */
6603fe3e 780 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 781 U16 *jump; /* positive offsets from me */
7f69552c 782 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
783 U8 *firstpos;/* pos in string of first trie match */
784 U32 firstchars;/* len in chars of firstpos from start */
785 U16 nextword;/* next word to try */
786 U16 topword; /* longest accepted word */
e822a8b4
DM
787 } trie;
788
fae667d5
YO
789 /* special types - these members are used to store state for special
790 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 791 struct {
77cb431f
DM
792 /* this first element must match u.yes */
793 struct regmatch_state *prev_yes_state;
faec1544 794 struct regmatch_state *prev_curlyx;
d1c49ad5 795 struct regmatch_state *prev_eval;
288b8c02 796 REGEXP *prev_rex;
aa283a38
DM
797 CHECKPOINT cp; /* remember current savestack indexes */
798 CHECKPOINT lastcp;
24be3102 799 U32 close_paren; /* which close bracket is our end (+1) */
1b5c067c 800 regnode *B; /* the node following us */
ba6840fb 801 char *prev_recurse_locinput;
e822a8b4
DM
802 } eval;
803
804 struct {
c476f425 805 /* this first element must match u.yes */
262b90c4 806 struct regmatch_state *prev_yes_state;
fae667d5
YO
807 I32 wanted;
808 I32 logical; /* saved copy of 'logical' var */
2abbd513
KW
809 U8 count; /* number of beginning positions */
810 char *start;
811 char *end;
fae667d5
YO
812 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
813 } ifmatch; /* and SUSPEND/UNLESSM */
814
815 struct {
816 /* this first element must match u.yes */
817 struct regmatch_state *prev_yes_state;
818 struct regmatch_state *prev_mark;
819 SV* mark_name;
820 char *mark_loc;
821 } mark;
822
823 struct {
824 int val;
825 } keeper;
826
827 /* quantifiers - these members are used for storing state for
828 for the regops used to implement quantifiers */
829 struct {
830 /* this first element must match u.yes */
831 struct regmatch_state *prev_yes_state;
c476f425 832 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
833 regnode *me; /* the CURLYX node */
834 regnode *B; /* the B node in /A*B/ */
c476f425
DM
835 CHECKPOINT cp; /* remember current savestack index */
836 bool minmod;
a0374537 837 int parenfloor;/* how far back to strip paren data */
c476f425
DM
838
839 /* these two are modified by WHILEM */
840 int count; /* how many instances of A we've matched */
841 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
842 } curlyx;
843
844 struct {
c476f425 845 /* this first element must match u.yes */
262b90c4 846 struct regmatch_state *prev_yes_state;
c476f425
DM
847 struct regmatch_state *save_curlyx;
848 CHECKPOINT cp; /* remember current savestack indexes */
849 CHECKPOINT lastcp;
850 char *save_lastloc; /* previous curlyx.lastloc */
851 I32 cache_offset;
852 I32 cache_mask;
e822a8b4
DM
853 } whilem;
854
855 struct {
5d458dd8
YO
856 /* this first element must match u.yes */
857 struct regmatch_state *prev_yes_state;
79a2a0e8 858 int c1, c2; /* case fold search */
40a82448 859 CHECKPOINT cp;
f6033a9d
DM
860 U32 lastparen;
861 U32 lastcloseparen;
40a82448
DM
862 I32 alen; /* length of first-matched A string */
863 I32 count;
0cadcf80 864 bool minmod;
40a82448
DM
865 regnode *A, *B; /* the nodes corresponding to /A*B/ */
866 regnode *me; /* the curlym node */
79a2a0e8
KW
867 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
868 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
869 } curlym;
870
871 struct {
3b6647e0 872 U32 paren;
c255a977 873 CHECKPOINT cp;
f6033a9d
DM
874 U32 lastparen;
875 U32 lastcloseparen;
79a2a0e8 876 int c1, c2; /* case fold search */
c255a977
DM
877 char *maxpos; /* highest possible point in string to match */
878 char *oldloc; /* the previous locinput */
e822a8b4 879 int count;
c255a977
DM
880 int min, max; /* {m,n} */
881 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
882 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
883 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 884 } curly; /* and CURLYN/PLUS/STAR */
dad79028 885
d8319b27 886 } u;
5d9a96ca
DM
887} regmatch_state;
888
ce12e254 889
d5a00e4a 890
5d9a96ca
DM
891/* how many regmatch_state structs to allocate as a single slab.
892 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
893 * pointers, plus 1 for any mythical malloc overhead. */
894
895#define PERL_REGMATCH_SLAB_SLOTS \
896 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
897
898typedef struct regmatch_slab {
899 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
900 struct regmatch_slab *prev, *next;
901} regmatch_slab;
1ade1aa1 902
46ab3289 903
cde0cee5 904
1ade1aa1 905/*
14d04a33 906 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 907 */