This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
rename RX_HAS_CUTGROUP() to RXp_HAS_CUTGROUP()
[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
c277df42
IZ
23struct regnode {
24 U8 flags;
25 U8 type;
26 U16 next_off;
27};
28
29typedef struct regnode regnode;
30
cad2e5aa 31struct reg_substr_data;
2779dcf1 32
0ee3c8d0
JH
33struct reg_data;
34
f9f4320a 35struct regexp_engine;
28d8d7f4 36struct regexp;
bbe252da 37
785a26d5 38struct reg_substr_datum {
c4828ca0
DM
39 SSize_t min_offset; /* min pos (in chars) that substr must appear */
40 SSize_t max_offset /* max pos (in chars) that substr must appear */;
785a26d5
YO
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
c4828ca0 43 SSize_t end_shift; /* how many fixed chars must end the string */
785a26d5
YO
44};
45struct reg_substr_data {
6480a6c4 46 U8 check_ix; /* index into data[] of check substr */
785a26d5
YO
47 struct reg_substr_datum data[3]; /* Actual array */
48};
f9f4320a 49
db2c6cb3 50#ifdef PERL_ANY_COW
bbe252da
YO
51#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
52#else
53#define SV_SAVED_COPY
ed252734 54#endif
28d8d7f4 55
b3fd53f3
DM
56/* offsets within a string of a particular /(.)/ capture */
57
f0ab9afb 58typedef struct regexp_paren_pair {
99a90e59
FC
59 SSize_t start;
60 SSize_t end;
b3fd53f3
DM
61 /* 'start_tmp' records a new opening position before the matching end
62 * has been found, so that the old start and end values are still
63 * valid, e.g.
64 * "abc" =~ /(.(?{print "[$1]"}))+/
65 *outputs [][a][b]
66 * This field is not part of the API. */
ea3daa5d 67 SSize_t start_tmp;
f0ab9afb 68} regexp_paren_pair;
28d8d7f4 69
52ae8f7e 70#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 71#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 72#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
73
74/* Subtracting b from a leaves in a everything that was there that isn't in b,
75 * that is the intersection of a with b's complement */
76#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
77#endif
78
3d2bd50a
DM
79/* record the position of a (?{...}) within a pattern */
80
81struct reg_code_block {
82 STRLEN start;
83 STRLEN end;
84 OP *block;
b30fcab9 85 REGEXP *src_regex;
3d2bd50a
DM
86};
87
1acab4c5
DM
88/* array of reg_code_block's plus header info */
89
90struct reg_code_blocks {
f8def6c7 91 int refcnt; /* we may be pointed to from a regex and from the savestack */
1acab4c5
DM
92 int count; /* how many code blocks */
93 struct reg_code_block *cb; /* array of reg_code_block's */
94};
95
3d2bd50a 96
882227b7
AB
97/*
98 The regexp/REGEXP struct, see L<perlreapi> for further documentation
99 on the individual fields. The struct is ordered so that the most
100 commonly used fields are placed at the start.
101
102 Any patch that adds items to this struct will need to include
103 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
104 (C<pregfree()>). This involves freeing or cloning items in the
105 regexp's data array based on the data item's type.
106*/
107
08e44740
NC
108#define _REGEXP_COMMON \
109 /* what engine created this regexp? */ \
110 const struct regexp_engine* engine; \
111 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 112 HV *paren_names; /* Optional hash of paren names */ \
ba6840fb 113 /*--------------------------------------------------------*/ \
08e44740
NC
114 /* Information about the match that the perl core uses to */ \
115 /* manage things */ \
116 U32 extflags; /* Flags used both externally and internally */ \
ea3daa5d
FC
117 SSize_t minlen; /* mininum possible number of chars in string to match */\
118 SSize_t minlenret; /* mininum possible number of chars in $& */ \
119 STRLEN gofs; /* chars left of pos that we search from */ \
08e44740
NC
120 /* substring data about strings that must appear in the */ \
121 /* final match, used for optimisations */ \
122 struct reg_substr_data *substrs; \
123 U32 nparens; /* number of capture buffers */ \
124 /* private engine specific data */ \
125 U32 intflags; /* Engine Specific Internal flags */ \
126 void *pprivate; /* Data private to the regex engine which */ \
127 /* created this object. */ \
ba6840fb 128 /*--------------------------------------------------------*/ \
08e44740
NC
129 /* Data about the last/current match. These are modified */ \
130 /* during matching */ \
131 U32 lastparen; /* last open paren matched */ \
132 U32 lastcloseparen; /* last close paren matched */ \
08e44740
NC
133 /* Array of offsets for (@-) and (@+) */ \
134 regexp_paren_pair *offs; \
ba6840fb
YO
135 char **recurse_locinput; /* used to detect infinite recursion, XXX: move to internal */ \
136 /*--------------------------------------------------------*/ \
08e44740
NC
137 /* saved or original string so \digit works forever. */ \
138 char *subbeg; \
139 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
19394178 140 SSize_t sublen; /* Length of string pointed by subbeg */ \
ea3daa5d
FC
141 SSize_t suboffset; /* byte offset of subbeg from logical start of str */ \
142 SSize_t subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740 143 /* Information about the match that isn't often used */ \
ee273784 144 SSize_t maxlen; /* mininum possible number of chars in string to match */\
ba6840fb 145 /*--------------------------------------------------------*/ \
08e44740 146 /* offset from wrapped to the start of precomp */ \
654eccd5 147 PERL_BITFIELD32 pre_prefix:4; \
dbc200c5
YO
148 /* original flags used to compile the pattern, may differ */ \
149 /* from extflags in various ways */ \
150 PERL_BITFIELD32 compflags:9; \
ba6840fb 151 /*--------------------------------------------------------*/ \
b82ad30c 152 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 153
bbe252da 154typedef struct regexp {
288b8c02 155 _XPV_HEAD;
08e44740 156 _REGEXP_COMMON;
f8fc2ecf
YO
157} regexp;
158
89f6f287 159#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 160
785a26d5 161/* used for high speed searches */
f9f4320a
YO
162typedef struct re_scream_pos_data_s
163{
164 char **scream_olds; /* match pos */
ea3daa5d 165 SSize_t *scream_pos; /* Internal iterator of scream. */
f9f4320a
YO
166} re_scream_pos_data;
167
785a26d5
YO
168/* regexp_engine structure. This is the dispatch table for regexes.
169 * Any regex engine implementation must be able to build one of these.
170 */
f9f4320a 171typedef struct regexp_engine {
1593ad57 172 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc 173 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
ea3daa5d 174 char* strbeg, SSize_t minend, SV* sv,
49d7dfbc 175 void* data, U32 flags);
52a21eb3
DM
176 char* (*intuit) (pTHX_
177 REGEXP * const rx,
178 SV *sv,
179 const char * const strbeg,
180 char *strpos,
181 char *strend,
182 const U32 flags,
9f61653a 183 re_scream_pos_data *data);
49d7dfbc 184 SV* (*checkstr) (pTHX_ REGEXP * const rx);
fc6bde6f 185 void (*rxfree) (pTHX_ REGEXP * const rx);
2fdbfb4d 186 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 187 SV * const sv);
2fdbfb4d
AB
188 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
189 SV const * const value);
190 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
191 const I32 paren);
192b9cd1
AB
192 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
193 SV * const value, const U32 flags);
194 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
195 const U32 flags);
49d7dfbc 196 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 197#ifdef USE_ITHREADS
49d7dfbc 198 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 199#endif
3c13cae6
DM
200 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
201 OP *expr, const struct regexp_engine* eng,
7984ae89 202 REGEXP *old_re,
76ac488f 203 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
204} regexp_engine;
205
192b9cd1
AB
206/*
207 These are passed to the numbered capture variable callbacks as the
208 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
209 $2 etc.
210*/
2c7b5d76
DM
211#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
212#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
213#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
214#define RX_BUFF_IDX_PREMATCH -2 /* $` */
215#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
216#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
217
218/*
219 Flags that are passed to the named_buff and named_buff_iter
220 callbacks above. Those routines are called from universal.c via the
221 Tie::Hash::NamedCapture interface for %+ and %- and the re::
222 functions in the same file.
223*/
224
225/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
226#define RXapif_FETCH 0x0001
227#define RXapif_STORE 0x0002
228#define RXapif_DELETE 0x0004
229#define RXapif_CLEAR 0x0008
230#define RXapif_EXISTS 0x0010
231#define RXapif_SCALAR 0x0020
232#define RXapif_FIRSTKEY 0x0040
233#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
234
235/* Whether %+ or %- is being operated on */
f1b875a0
YO
236#define RXapif_ONE 0x0100 /* %+ */
237#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
238
239/* Whether this is being called from a re:: function */
f1b875a0
YO
240#define RXapif_REGNAME 0x0400
241#define RXapif_REGNAMES 0x0800
7948fc08 242#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 243
f7e71195
AB
244/*
245=head1 REGEXP Functions
246
247=for apidoc Am|REGEXP *|SvRX|SV *sv
248
72d33970 249Convenience macro to get the REGEXP from a SV. This is approximately
f7e71195
AB
250equivalent to the following snippet:
251
252 if (SvMAGICAL(sv))
253 mg_get(sv);
1af910ea
DL
254 if (SvROK(sv))
255 sv = MUTABLE_SV(SvRV(sv));
256 if (SvTYPE(sv) == SVt_REGEXP)
257 return (REGEXP*) sv;
f7e71195 258
796b6530 259C<NULL> will be returned if a REGEXP* is not found.
f7e71195
AB
260
261=for apidoc Am|bool|SvRXOK|SV* sv
262
1af910ea
DL
263Returns a boolean indicating whether the SV (or the one it references)
264is a REGEXP.
f7e71195
AB
265
266If you want to do something with the REGEXP* later use SvRX instead
267and check for NULL.
268
269=cut
270*/
271
272#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
8298454c 273#define SvRXOK(sv) cBOOL(Perl_get_re_arg(aTHX_ sv))
f7e71195
AB
274
275
7948fc08 276/* Flags stored in regexp->extflags
bbe252da 277 * These are used by code external to the regexp engine
e357fc67 278 *
5b126c84
KW
279 * Note that the flags whose names start with RXf_PMf_ are defined in
280 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 281 *
eb2624c9
FC
282 * NOTE: if you modify any RXf flags you should run regen.pl or
283 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 284 *
bbe252da
YO
285 */
286
5b126c84
KW
287#include "op_reg_common.h"
288
77c8f263 289#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 290
cc4d09e1
KW
291#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl, x_count) \
292 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
293 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
294 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
77c8f263
KW
295 case XTENDED_PAT_MOD: if (x_count == 0) { \
296 *(pmfl) |= RXf_PMf_EXTENDED; \
297 *(pmfl) &= ~RXf_PMf_EXTENDED_MORE; \
298 } \
299 else { \
300 *(pmfl) |= RXf_PMf_EXTENDED \
301 |RXf_PMf_EXTENDED_MORE; \
302 } \
303 (x_count)++; break; \
41d7c59e 304 case NOCAPTURE_PAT_MOD: *(pmfl) |= RXf_PMf_NOCAPTURE; break;
cc4d09e1 305
94b03d7d 306/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 307#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
77c8f263 308 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_EXTENDED_MORE|RXf_PMf_CHARSET|RXf_PMf_NOCAPTURE)
fb85c044 309
bcdf7404 310/* chars and strings used as regex pattern modifiers
486ec47a 311 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
312 *
313 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
314 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
315 * for its own uses. So 'k' is out as well.
bcdf7404 316 */
85508812 317#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 318#define EXEC_PAT_MOD 'e'
87e95b7f 319#define KEEPCOPY_PAT_MOD 'p'
41d7c59e 320#define NOCAPTURE_PAT_MOD 'n'
bcdf7404
YO
321#define ONCE_PAT_MOD 'o'
322#define GLOBAL_PAT_MOD 'g'
323#define CONTINUE_PAT_MOD 'c'
324#define MULTILINE_PAT_MOD 'm'
325#define SINGLE_PAT_MOD 's'
326#define IGNORE_PAT_MOD 'i'
327#define XTENDED_PAT_MOD 'x'
4f4d7508 328#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
329#define LOCALE_PAT_MOD 'l'
330#define UNICODE_PAT_MOD 'u'
50e91148 331#define DEPENDS_PAT_MOD 'd'
cfaf538b 332#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
333
334#define ONCE_PAT_MODS "o"
87e95b7f 335#define KEEPCOPY_PAT_MODS "p"
41d7c59e 336#define NOCAPTURE_PAT_MODS "n"
bcdf7404
YO
337#define EXEC_PAT_MODS "e"
338#define LOOP_PAT_MODS "gc"
4f4d7508 339#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
340#define LOCALE_PAT_MODS "l"
341#define UNICODE_PAT_MODS "u"
50e91148 342#define DEPENDS_PAT_MODS "d"
cfaf538b 343#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 344#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 345
78fecfc8 346/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
347 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
348 * character is bit +1, etc. */
41d7c59e 349#define STD_PAT_MODS "msixxn"
bcdf7404 350
94b03d7d
KW
351#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
352
78fecfc8 353/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
354 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
355 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
356#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
357
41d7c59e 358#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS NOCAPTURE_PAT_MODS
94b03d7d 359#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 360#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 361#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 362
cb5027f2 363/*
eb2624c9
FC
364 * NOTE: if you modify any RXf flags you should run regen.pl or
365 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
366 *
367 */
bcdf7404 368
dbc200c5 369/*
5012eebe
DM
370 Set in Perl_pmruntime for a split. Will be used by regex engines to
371 check whether they should set RXf_SKIPWHITE
dbc200c5 372*/
32f62247
KW
373#define RXf_SPLIT RXf_PMf_SPLIT
374
0bd51918
KW
375/* Currently the regex flags occupy a single 32-bit word. Not all bits are
376 * currently used. The lower bits are shared with their corresponding PMf flag
377 * bits, up to but not including _RXf_PMf_SHIFT_NEXT. The unused bits
378 * immediately follow; finally the used RXf-only (unshared) bits, so that the
379 * highest bit in the word is used. This gathers all the unused bits as a pool
380 * in the middle, like so: 11111111111111110000001111111111
381 * where the '1's represent used bits, and the '0's unused. This design allows
382 * us to allocate off one end of the pool if we need to add a shared bit, and
383 * off the other end if we need a non-shared bit, without disturbing the other
384 * bits. This maximizes the likelihood of being able to change things without
385 * breaking binary compatibility.
386 *
387 * To add shared bits, do so in op_reg_common.h. This should change
388 * _RXf_PMf_SHIFT_NEXT so that things won't compile. Then come to regexp.h and
389 * op.h and adjust the constant adders in the definitions of RXf_BASE_SHIFT and
390 * Pmf_BASE_SHIFT down by the number of shared bits you added. That's it.
391 * Things should be binary compatible. But if either of these gets to having
392 * to subtract rather than add, leave at 0 and instead adjust all the entries
393 * that are in terms of it. But if the first one of those is already
394 * RXf_BASE_SHIFT+0, there are no bits left, and a redesign is in order.
395 *
396 * To remove unshared bits, just delete its entry. If you're where breaking
397 * binary compatibility is ok to do, you might want to adjust things to move
398 * the newly opened space so that it gets absorbed into the common pool.
399 *
400 * To add unshared bits, first use up any gaps in the middle. Otherwise,
401 * allocate off the low end until you get to RXf_BASE_SHIFT+0. If that isn't
402 * enough, move RXf_BASE_SHIFT down (if possible) and add the new bit at the
403 * other end instead; this preserves binary compatibility.
404 *
405 * For the regexp bits, PL_reg_extflags_name[] in regnodes.h has a comment
406 * giving which bits are used/unused */
407
d262c0c7 408#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT + 2)
52d81aa8 409
bbe252da 410/* What we have seen */
9cba692b
KW
411#define RXf_NO_INPLACE_SUBST (1U<<(RXf_BASE_SHIFT+2))
412#define RXf_EVAL_SEEN (1U<<(RXf_BASE_SHIFT+3))
bbe252da
YO
413
414/* Special */
9cba692b
KW
415#define RXf_UNBOUNDED_QUANTIFIER_SEEN (1U<<(RXf_BASE_SHIFT+4))
416#define RXf_CHECK_ALL (1U<<(RXf_BASE_SHIFT+5))
bbe252da
YO
417
418/* UTF8 related */
9cba692b 419#define RXf_MATCH_UTF8 (1U<<(RXf_BASE_SHIFT+6)) /* $1 etc are utf8 */
bbe252da
YO
420
421/* Intuit related */
9cba692b
KW
422#define RXf_USE_INTUIT_NOML (1U<<(RXf_BASE_SHIFT+7))
423#define RXf_USE_INTUIT_ML (1U<<(RXf_BASE_SHIFT+8))
424#define RXf_INTUIT_TAIL (1U<<(RXf_BASE_SHIFT+9))
bbe252da
YO
425#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
426
a3b51d37 427/* Do we have some sort of anchor? */
9cba692b 428#define RXf_IS_ANCHORED (1U<<(RXf_BASE_SHIFT+10))
a3b51d37 429
bbe252da 430/* Copy and tainted info */
9cba692b 431#define RXf_COPY_DONE (1U<<(RXf_BASE_SHIFT+11))
ef07e810 432
1738e041 433/* post-execution: $1 et al are tainted */
9cba692b 434#define RXf_TAINTED_SEEN (1U<<(RXf_BASE_SHIFT+12))
ef07e810 435/* this pattern was tainted during compilation */
9cba692b 436#define RXf_TAINTED (1U<<(RXf_BASE_SHIFT+13))
52d81aa8
NC
437
438/* Flags indicating special patterns */
9cba692b
KW
439#define RXf_START_ONLY (1U<<(RXf_BASE_SHIFT+14)) /* Pattern is /^/ */
440#define RXf_SKIPWHITE (1U<<(RXf_BASE_SHIFT+15)) /* Pattern is for a */
441 /* split " " */
442#define RXf_WHITE (1U<<(RXf_BASE_SHIFT+16)) /* Pattern is /\s+/ */
3f6e6b29 443#define RXf_NULL (1U<<(RXf_BASE_SHIFT+17)) /* Pattern is // */
0bd51918
KW
444
445/* See comments at the beginning of these defines about adding bits. The
446 * highest bit position should be used, so that if RXf_BASE_SHIFT gets
447 * increased, the #error below will be triggered so that you will be reminded
448 * to adjust things at the other end to keep the bit positions unchanged */
3f6e6b29 449#if RXf_BASE_SHIFT+17 > 31
0bd51918 450# error Too many RXf_PMf bits used. See comments at beginning of these for what to do
1850c8f9 451#endif
c737faaf 452
cb5027f2 453/*
eb2624c9
FC
454 * NOTE: if you modify any RXf flags you should run regen.pl or
455 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
456 *
457 */
bbe252da 458
dc6d7f5c 459#ifdef NO_TAINT_SUPPORT
a885a8e0
DM
460# define RX_ISTAINTED(rx_sv) 0
461# define RX_TAINT_on(rx_sv) NOOP
462# define RXp_MATCH_TAINTED(prog) 0
463# define RX_MATCH_TAINTED(rx_sv) 0
464# define RXp_MATCH_TAINTED_on(prog) NOOP
465# define RX_MATCH_TAINTED_on(rx_sv) NOOP
466# define RX_MATCH_TAINTED_off(rx_sv) NOOP
284167a5 467#else
a885a8e0
DM
468# define RX_ISTAINTED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_TAINTED)
469# define RX_TAINT_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_TAINTED)
470# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
471# define RX_MATCH_TAINTED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_TAINTED_SEEN)
472# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
473# define RX_MATCH_TAINTED_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_TAINTED_SEEN)
474# define RX_MATCH_TAINTED_off(rx_sv) (RX_EXTFLAGS(rx_sv) &= ~RXf_TAINTED_SEEN)
284167a5
SM
475#endif
476
ea0673ee 477#define RXp_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
a885a8e0
DM
478
479#define RX_MATCH_TAINTED_set(rx_sv, t) ((t) \
480 ? RX_MATCH_TAINTED_on(rx_sv) \
481 : RX_MATCH_TAINTED_off(rx_sv))
482
483#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
484#define RX_MATCH_COPIED(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_COPY_DONE)
485#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
486#define RX_MATCH_COPIED_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_COPY_DONE)
487#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
488#define RX_MATCH_COPIED_off(rx_sv) (RX_EXTFLAGS(rx_sv) &= ~RXf_COPY_DONE)
489#define RX_MATCH_COPIED_set(rx_sv,t) ((t) \
490 ? RX_MATCH_COPIED_on(rx_sv) \
491 : RX_MATCH_COPIED_off(rx_sv))
492
493#define RXp_EXTFLAGS(rx) ((rx)->extflags)
494#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 495
07bc277f 496/* For source compatibility. We used to store these explicitly. */
a885a8e0
DM
497#define RX_PRECOMP(rx_sv) (RX_WRAPPED(rx_sv) \
498 + ReANY(rx_sv)->pre_prefix)
499#define RX_PRECOMP_const(rx_sv) (RX_WRAPPED_const(rx_sv) \
500 + ReANY(rx_sv)->pre_prefix)
5509d87a
NC
501/* FIXME? Are we hardcoding too much here and constraining plugin extension
502 writers? Specifically, the value 1 assumes that the wrapped version always
503 has exactly one character at the end, a ')'. Will that always be true? */
a885a8e0
DM
504#define RX_PRELEN(rx_sv) (RX_WRAPLEN(rx_sv) \
505 - ReANY(rx_sv)->pre_prefix - 1)
506
507#define RX_WRAPPED(rx_sv) SvPVX(rx_sv)
508#define RX_WRAPPED_const(rx_sv) SvPVX_const(rx_sv)
509#define RX_WRAPLEN(rx_sv) SvCUR(rx_sv)
510#define RX_CHECK_SUBSTR(rx_sv) (ReANY(rx_sv)->check_substr)
511#define RX_REFCNT(rx_sv) SvREFCNT(rx_sv)
512#define RX_EXTFLAGS(rx_sv) RXp_EXTFLAGS(ReANY(rx_sv))
513#define RX_COMPFLAGS(rx_sv) RXp_COMPFLAGS(ReANY(rx_sv))
514#define RX_ENGINE(rx_sv) (ReANY(rx_sv)->engine)
515#define RX_SUBBEG(rx_sv) (ReANY(rx_sv)->subbeg)
516#define RX_SUBOFFSET(rx_sv) (ReANY(rx_sv)->suboffset)
517#define RX_SUBCOFFSET(rx_sv) (ReANY(rx_sv)->subcoffset)
518#define RX_OFFS(rx_sv) (ReANY(rx_sv)->offs)
519#define RX_NPARENS(rx_sv) (ReANY(rx_sv)->nparens)
520#define RX_SUBLEN(rx_sv) (ReANY(rx_sv)->sublen)
521#define RX_MINLEN(rx_sv) (ReANY(rx_sv)->minlen)
522#define RX_MINLENRET(rx_sv) (ReANY(rx_sv)->minlenret)
523#define RX_GOFS(rx_sv) (ReANY(rx_sv)->gofs)
524#define RX_LASTPAREN(rx_sv) (ReANY(rx_sv)->lastparen)
525#define RX_LASTCLOSEPAREN(rx_sv) (ReANY(rx_sv)->lastcloseparen)
526#define RX_SAVED_COPY(rx_sv) (ReANY(rx_sv)->saved_copy)
03c83e26 527/* last match was zero-length */
55922dbe
DM
528#define RX_ZERO_LEN(rx_sv) \
529 (RX_OFFS(rx_sv)[0].start + (SSize_t)RX_GOFS(rx_sv) \
530 == RX_OFFS(rx_sv)[0].end)
220fc49f 531
be8e71aa
YO
532#endif /* PLUGGABLE_RE_EXTENSION */
533
486ec47a 534/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 535
db2c6cb3 536#ifdef PERL_ANY_COW
ed252734 537#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
538 STMT_START {if (RX_SAVED_COPY(rx)) { \
539 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
540 } \
541 if (RX_MATCH_COPIED(rx)) { \
07bc277f 542 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
543 RX_MATCH_COPIED_off(rx); \
544 }} STMT_END
545#else
546#define RX_MATCH_COPY_FREE(rx) \
547 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 548 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
549 RX_MATCH_COPIED_off(rx); \
550 }} STMT_END
551#endif
552
a885a8e0
DM
553#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
554#define RX_MATCH_UTF8(rx_sv) (RX_EXTFLAGS(rx_sv) & RXf_MATCH_UTF8)
555#define RX_MATCH_UTF8_on(rx_sv) (RX_EXTFLAGS(rx_sv) |= RXf_MATCH_UTF8)
556#define RX_MATCH_UTF8_off(rx_sv) (RX_EXTFLAGS(rx_sv) &= ~RXf_MATCH_UTF8)
557#define RX_MATCH_UTF8_set(rx_sv, t) ((t) \
558 ? RX_MATCH_UTF8_on(rx_sv) \
559 : RX_MATCH_UTF8_off(rx_sv))
efd26800
NC
560
561/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
a885a8e0 562#define RX_UTF8(rx_sv) SvUTF8(rx_sv)
7948fc08 563
a340edde
DM
564
565/* bits in flags arg of Perl_regexec_flags() */
566
567#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
568#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
569#define REXEC_SCREAM 0x04 /* currently unused. */
570#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
571#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
572 no need to copy string again */
573
574 /* under REXEC_COPY_STR, it's ok for the
575 engine (modulo PL_sawamperand etc)
576 to skip copying: ... */
577#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
578#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
579#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
580 the start pos (so s/.\G// would fail
581 on second iteration */
c277df42 582
7fe44985 583#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
584# define ReREFCNT_inc(re) \
585 ({ \
586 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
587 REGEXP *const _rerefcnt_inc = (re); \
588 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
589 SvREFCNT_inc(_rerefcnt_inc); \
590 _rerefcnt_inc; \
288b8c02
NC
591 })
592# define ReREFCNT_dec(re) \
593 ({ \
594 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
595 REGEXP *const _rerefcnt_dec = (re); \
596 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
597 })
598#else
599# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 600# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 601#endif
8d919b0a 602#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
603
604/* FIXME for plugins. */
cf93c79d
IZ
605
606#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
607#define FBMcf_TAIL_DOLLARM 2
608#define FBMcf_TAIL_Z 4
609#define FBMcf_TAIL_z 8
610#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
611
612#define FBMrf_MULTILINE 1
cad2e5aa 613
331b2dcc
DM
614struct regmatch_state;
615struct regmatch_slab;
8adc0f72 616
bf2039a9
DM
617/* like regmatch_info_aux, but contains extra fields only needed if the
618 * pattern contains (?{}). If used, is snuck into the second slot in the
619 * regmatch_state stack at the start of execution */
8adc0f72
DM
620
621typedef struct {
622 regexp *rex;
623 PMOP *curpm; /* saved PL_curpm */
624#ifdef PERL_ANY_COW
625 SV *saved_copy; /* saved saved_copy field from rex */
626#endif
627 char *subbeg; /* saved subbeg field from rex */
628 STRLEN sublen; /* saved sublen field from rex */
629 STRLEN suboffset; /* saved suboffset field from rex */
630 STRLEN subcoffset; /* saved subcoffset field from rex */
631 MAGIC *pos_magic; /* pos() magic attached to $_ */
ea3daa5d 632 SSize_t pos; /* the original value of pos() in pos_magic */
25fdce4a 633 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
634} regmatch_info_aux_eval;
635
636
637/* fields that logically live in regmatch_info, but which need cleaning
638 * up on croak(), and so are instead are snuck into the first slot in
639 * the regmatch_state stack at the start of execution */
640
641typedef struct {
642 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
643 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
644 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 645 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
646} regmatch_info_aux;
647
8adc0f72 648
3b0527fe 649/* some basic information about the current match that is created by
bf2039a9
DM
650 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
651 * It is allocated as a local var on the stack, so nothing should be
652 * stored in it that needs preserving or clearing up on croak().
653 * For that, see the aux_info and aux_info_eval members of the
654 * regmatch_state union. */
3b0527fe
DM
655
656typedef struct {
28d03b21 657 REGEXP *prog; /* the regex being executed */
9d9163fb 658 const char * strbeg; /* real start of string */
28d03b21
DM
659 char *strend; /* one byte beyond last char of match string */
660 char *till; /* matches shorter than this fail (see minlen arg) */
661 SV *sv; /* the SV string currently being matched */
662 char *ganch; /* position of \G anchor */
663 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
664 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
665 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
666 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
667 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 668 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 669 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
670 bool is_utf8_pat; /* regex is utf8 */
671 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 672 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
673} regmatch_info;
674
5d9a96ca
DM
675
676/* structures for holding and saving the state maintained by regmatch() */
677
d56b3014 678#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
ba6840fb 679#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 10
d56b3014 680#endif
6bda09f9 681
5d9a96ca
DM
682typedef I32 CHECKPOINT;
683
a0374537 684typedef struct regmatch_state {
40a82448 685 int resume_state; /* where to jump to on return */
24d3c4a9 686 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
687
688 union {
77cb431f 689
bf2039a9
DM
690 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
691 * the nest. They aren't saved backtrack state; rather they
692 * represent one or two extra chunks of data that need allocating
693 * at the start of a match. These fields would logically live in
694 * the regmatch_info struct, except that is allocated on the
695 * C stack, and these fields are all things that require cleanup
696 * after a croak(), when the stack is lost.
697 * As a convenience, we just use the first 1 or 2 regmatch_state
698 * slots to store this info, as we will be allocating a slab of
699 * these anyway. Otherwise we'd have to malloc and then free them,
700 * or allocate them on the save stack (where they will get
701 * realloced if the save stack grows).
702 * info_aux contains the extra fields that are always needed;
703 * info_aux_eval contains extra fields that only needed if
704 * the pattern contains code blocks
705 * We split them into two separate structs to avoid increasing
706 * the size of the union.
707 */
708
709 regmatch_info_aux info_aux;
710
711 regmatch_info_aux_eval info_aux_eval;
712
77cb431f
DM
713 /* this is a fake union member that matches the first element
714 * of each member that needs to store positive backtrack
715 * information */
716 struct {
717 struct regmatch_state *prev_yes_state;
718 } yes;
719
fae667d5
YO
720 /* branchlike members */
721 /* this is a fake union member that matches the first elements
722 * of each member that needs to behave like a branch */
723 struct {
5d458dd8
YO
724 /* this first element must match u.yes */
725 struct regmatch_state *prev_yes_state;
fae667d5 726 U32 lastparen;
f6033a9d 727 U32 lastcloseparen;
fae667d5
YO
728 CHECKPOINT cp;
729
730 } branchlike;
731
732 struct {
733 /* the first elements must match u.branchlike */
734 struct regmatch_state *prev_yes_state;
735 U32 lastparen;
f6033a9d 736 U32 lastcloseparen;
fae667d5
YO
737 CHECKPOINT cp;
738
739 regnode *next_branch; /* next branch node */
740 } branch;
741
742 struct {
743 /* the first elements must match u.branchlike */
744 struct regmatch_state *prev_yes_state;
745 U32 lastparen;
f6033a9d 746 U32 lastcloseparen;
fae667d5
YO
747 CHECKPOINT cp;
748
2e64971a 749 U32 accepted; /* how many accepting states left */
6603fe3e 750 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 751 U16 *jump; /* positive offsets from me */
7f69552c 752 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
753 U8 *firstpos;/* pos in string of first trie match */
754 U32 firstchars;/* len in chars of firstpos from start */
755 U16 nextword;/* next word to try */
756 U16 topword; /* longest accepted word */
e822a8b4
DM
757 } trie;
758
fae667d5
YO
759 /* special types - these members are used to store state for special
760 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 761 struct {
77cb431f
DM
762 /* this first element must match u.yes */
763 struct regmatch_state *prev_yes_state;
faec1544 764 struct regmatch_state *prev_curlyx;
d1c49ad5 765 struct regmatch_state *prev_eval;
288b8c02 766 REGEXP *prev_rex;
aa283a38
DM
767 CHECKPOINT cp; /* remember current savestack indexes */
768 CHECKPOINT lastcp;
24be3102 769 U32 close_paren; /* which close bracket is our end (+1) */
1b5c067c 770 regnode *B; /* the node following us */
ba6840fb 771 char *prev_recurse_locinput;
e822a8b4
DM
772 } eval;
773
774 struct {
c476f425 775 /* this first element must match u.yes */
262b90c4 776 struct regmatch_state *prev_yes_state;
fae667d5
YO
777 I32 wanted;
778 I32 logical; /* saved copy of 'logical' var */
779 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
780 } ifmatch; /* and SUSPEND/UNLESSM */
781
782 struct {
783 /* this first element must match u.yes */
784 struct regmatch_state *prev_yes_state;
785 struct regmatch_state *prev_mark;
786 SV* mark_name;
787 char *mark_loc;
788 } mark;
789
790 struct {
791 int val;
792 } keeper;
793
794 /* quantifiers - these members are used for storing state for
795 for the regops used to implement quantifiers */
796 struct {
797 /* this first element must match u.yes */
798 struct regmatch_state *prev_yes_state;
c476f425 799 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
800 regnode *me; /* the CURLYX node */
801 regnode *B; /* the B node in /A*B/ */
c476f425
DM
802 CHECKPOINT cp; /* remember current savestack index */
803 bool minmod;
a0374537 804 int parenfloor;/* how far back to strip paren data */
c476f425
DM
805
806 /* these two are modified by WHILEM */
807 int count; /* how many instances of A we've matched */
808 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
809 } curlyx;
810
811 struct {
c476f425 812 /* this first element must match u.yes */
262b90c4 813 struct regmatch_state *prev_yes_state;
c476f425
DM
814 struct regmatch_state *save_curlyx;
815 CHECKPOINT cp; /* remember current savestack indexes */
816 CHECKPOINT lastcp;
817 char *save_lastloc; /* previous curlyx.lastloc */
818 I32 cache_offset;
819 I32 cache_mask;
e822a8b4
DM
820 } whilem;
821
822 struct {
5d458dd8
YO
823 /* this first element must match u.yes */
824 struct regmatch_state *prev_yes_state;
79a2a0e8 825 int c1, c2; /* case fold search */
40a82448 826 CHECKPOINT cp;
f6033a9d
DM
827 U32 lastparen;
828 U32 lastcloseparen;
40a82448
DM
829 I32 alen; /* length of first-matched A string */
830 I32 count;
0cadcf80 831 bool minmod;
40a82448
DM
832 regnode *A, *B; /* the nodes corresponding to /A*B/ */
833 regnode *me; /* the curlym node */
79a2a0e8
KW
834 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
835 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
836 } curlym;
837
838 struct {
3b6647e0 839 U32 paren;
c255a977 840 CHECKPOINT cp;
f6033a9d
DM
841 U32 lastparen;
842 U32 lastcloseparen;
79a2a0e8 843 int c1, c2; /* case fold search */
c255a977
DM
844 char *maxpos; /* highest possible point in string to match */
845 char *oldloc; /* the previous locinput */
e822a8b4 846 int count;
c255a977
DM
847 int min, max; /* {m,n} */
848 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
849 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
850 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 851 } curly; /* and CURLYN/PLUS/STAR */
dad79028 852
d8319b27 853 } u;
5d9a96ca
DM
854} regmatch_state;
855
ce12e254 856
d5a00e4a 857
5d9a96ca
DM
858/* how many regmatch_state structs to allocate as a single slab.
859 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
860 * pointers, plus 1 for any mythical malloc overhead. */
861
862#define PERL_REGMATCH_SLAB_SLOTS \
863 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
864
865typedef struct regmatch_slab {
866 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
867 struct regmatch_slab *prev, *next;
868} regmatch_slab;
1ade1aa1 869
46ab3289 870
cde0cee5 871
1ade1aa1 872/*
14d04a33 873 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 874 */