This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: fix an improper indent
[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
c277df42
IZ
21struct regnode {
22 U8 flags;
23 U8 type;
24 U16 next_off;
25};
26
27typedef struct regnode regnode;
28
cad2e5aa 29struct reg_substr_data;
2779dcf1 30
0ee3c8d0
JH
31struct reg_data;
32
f9f4320a 33struct regexp_engine;
28d8d7f4 34struct regexp;
bbe252da 35
785a26d5
YO
36struct reg_substr_datum {
37 I32 min_offset;
38 I32 max_offset;
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
41 I32 end_shift;
42};
43struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
45};
f9f4320a 46
f8c7b90f 47#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
48#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
49#else
50#define SV_SAVED_COPY
ed252734 51#endif
28d8d7f4 52
f0ab9afb
NC
53typedef struct regexp_paren_pair {
54 I32 start;
55 I32 end;
56} regexp_paren_pair;
28d8d7f4 57
882227b7
AB
58/*
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.
62
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.
67*/
68
08e44740
NC
69#define _REGEXP_COMMON \
70 /* what engine created this regexp? */ \
71 const struct regexp_engine* engine; \
72 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 73 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
74 /* Information about the match that the perl core uses to */ \
75 /* manage things */ \
76 U32 extflags; /* Flags used both externally and internally */ \
77 I32 minlen; /* mininum possible length of string to match */\
78 I32 minlenret; /* mininum possible length of $& */ \
79 U32 gofs; /* chars left of pos that we search from */ \
80 /* substring data about strings that must appear in the */ \
81 /* final match, used for optimisations */ \
82 struct reg_substr_data *substrs; \
83 U32 nparens; /* number of capture buffers */ \
84 /* private engine specific data */ \
85 U32 intflags; /* Engine Specific Internal flags */ \
86 void *pprivate; /* Data private to the regex engine which */ \
87 /* created this object. */ \
88 /* Data about the last/current match. These are modified */ \
89 /* during matching */ \
90 U32 lastparen; /* last open paren matched */ \
91 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 92 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
93 /* Array of offsets for (@-) and (@+) */ \
94 regexp_paren_pair *offs; \
95 /* saved or original string so \digit works forever. */ \
96 char *subbeg; \
97 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
98 I32 sublen; /* Length of string pointed by subbeg */ \
99 /* Information about the match that isn't often used */ \
100 /* offset from wrapped to the start of precomp */ \
654eccd5 101 PERL_BITFIELD32 pre_prefix:4; \
08e44740 102 /* number of eval groups in the pattern - for security checks */\
654eccd5 103 PERL_BITFIELD32 seen_evals:28
08e44740 104
bbe252da 105typedef struct regexp {
288b8c02 106 _XPV_HEAD;
08e44740 107 _REGEXP_COMMON;
f8fc2ecf
YO
108} regexp;
109
89f6f287 110#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 111
785a26d5 112/* used for high speed searches */
f9f4320a
YO
113typedef struct re_scream_pos_data_s
114{
115 char **scream_olds; /* match pos */
116 I32 *scream_pos; /* Internal iterator of scream. */
117} re_scream_pos_data;
118
785a26d5
YO
119/* regexp_engine structure. This is the dispatch table for regexes.
120 * Any regex engine implementation must be able to build one of these.
121 */
f9f4320a 122typedef struct regexp_engine {
1593ad57 123 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
124 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
125 char* strbeg, I32 minend, SV* screamer,
126 void* data, U32 flags);
127 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
128 char *strend, const U32 flags,
129 re_scream_pos_data *data);
49d7dfbc
AB
130 SV* (*checkstr) (pTHX_ REGEXP * const rx);
131 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 132 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 133 SV * const sv);
2fdbfb4d
AB
134 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
135 SV const * const value);
136 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
137 const I32 paren);
192b9cd1
AB
138 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
139 SV * const value, const U32 flags);
140 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
141 const U32 flags);
49d7dfbc 142 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 143#ifdef USE_ITHREADS
49d7dfbc 144 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 145#endif
f9f4320a
YO
146} regexp_engine;
147
192b9cd1
AB
148/*
149 These are passed to the numbered capture variable callbacks as the
150 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
151 $2 etc.
152*/
f1b875a0
YO
153#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
154#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
155#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
156
157/*
158 Flags that are passed to the named_buff and named_buff_iter
159 callbacks above. Those routines are called from universal.c via the
160 Tie::Hash::NamedCapture interface for %+ and %- and the re::
161 functions in the same file.
162*/
163
164/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
165#define RXapif_FETCH 0x0001
166#define RXapif_STORE 0x0002
167#define RXapif_DELETE 0x0004
168#define RXapif_CLEAR 0x0008
169#define RXapif_EXISTS 0x0010
170#define RXapif_SCALAR 0x0020
171#define RXapif_FIRSTKEY 0x0040
172#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
173
174/* Whether %+ or %- is being operated on */
f1b875a0
YO
175#define RXapif_ONE 0x0100 /* %+ */
176#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
177
178/* Whether this is being called from a re:: function */
f1b875a0
YO
179#define RXapif_REGNAME 0x0400
180#define RXapif_REGNAMES 0x0800
7948fc08 181#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 182
f7e71195
AB
183/*
184=head1 REGEXP Functions
185
186=for apidoc Am|REGEXP *|SvRX|SV *sv
187
188Convenience macro to get the REGEXP from a SV. This is approximately
189equivalent to the following snippet:
190
191 if (SvMAGICAL(sv))
192 mg_get(sv);
193 if (SvROK(sv) &&
194 (tmpsv = (SV*)SvRV(sv)) &&
195 SvTYPE(tmpsv) == SVt_PVMG &&
196 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
197 {
198 return (REGEXP *)tmpmg->mg_obj;
199 }
200
201NULL will be returned if a REGEXP* is not found.
202
203=for apidoc Am|bool|SvRXOK|SV* sv
204
205Returns a boolean indicating whether the SV contains qr magic
206(PERL_MAGIC_qr).
207
208If you want to do something with the REGEXP* later use SvRX instead
209and check for NULL.
210
211=cut
212*/
213
214#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
215#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
216
217
7948fc08 218/* Flags stored in regexp->extflags
bbe252da 219 * These are used by code external to the regexp engine
e357fc67 220 *
5b126c84
KW
221 * Note that the flags whose names start with RXf_PMf_ are defined in
222 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2
YO
223 *
224 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 225 * so that regnodes.h is updated with the changes.
cb5027f2 226 *
bbe252da
YO
227 */
228
5b126c84
KW
229#include "op_reg_common.h"
230
cde0cee5 231#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 232
a20207d7
YO
233#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
234 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
235 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
236 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
237 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 238
bcdf7404
YO
239/* chars and strings used as regex pattern modifiers
240 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
241 *
242 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
243 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
244 * for its own uses. So 'k' is out as well.
bcdf7404
YO
245 */
246#define EXEC_PAT_MOD 'e'
87e95b7f 247#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
248#define ONCE_PAT_MOD 'o'
249#define GLOBAL_PAT_MOD 'g'
250#define CONTINUE_PAT_MOD 'c'
251#define MULTILINE_PAT_MOD 'm'
252#define SINGLE_PAT_MOD 's'
253#define IGNORE_PAT_MOD 'i'
254#define XTENDED_PAT_MOD 'x'
4f4d7508 255#define NONDESTRUCT_PAT_MOD 'r'
bcdf7404
YO
256
257#define ONCE_PAT_MODS "o"
87e95b7f 258#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
259#define EXEC_PAT_MODS "e"
260#define LOOP_PAT_MODS "gc"
4f4d7508 261#define NONDESTRUCT_PAT_MODS "r"
bcdf7404 262
78fecfc8 263/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
264 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
265 * character is bit +1, etc. */
bcdf7404
YO
266#define STD_PAT_MODS "msix"
267
78fecfc8 268/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
269 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
270 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
271#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
272
273#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
274#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
275#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 276#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 277
cb5027f2
YO
278/*
279 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 280 * so that regnodes.h is updated with the changes.
cb5027f2
YO
281 *
282 */
bcdf7404 283
e795e964
KW
284/* Leave some space, so future bit allocations can go either in the shared or
285 * unshared area without affecting binary compatibility */
286#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+3)
f4a4cecb 287
52d81aa8 288/* Anchor and GPOS related stuff */
1ad4ec54
KW
289#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
290#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
291#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
292#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
293#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
294#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
295/* two bits here */
296#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
297#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
298#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
299
bbe252da 300/* What we have seen */
1ad4ec54
KW
301#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
302#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
303#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
304
305/* Special */
1ad4ec54
KW
306#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
307#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
308
309/* UTF8 related */
e795e964 310#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
311
312/* Intuit related */
e795e964
KW
313#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
314#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
315#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
316
317/*
318 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
319 be used by regex engines to check whether they should set
320 RXf_SKIPWHITE
321*/
e795e964 322#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 323
bbe252da
YO
324#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
325
326/* Copy and tainted info */
e795e964
KW
327#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
328#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
329#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18)) /* this pattern is tainted */
52d81aa8
NC
330
331/* Flags indicating special patterns */
e795e964
KW
332#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
333#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
334#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
335#define RXf_NULL (1<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
336#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
337# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
338#endif
c737faaf 339
cb5027f2
YO
340/*
341 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 342 * so that regnodes.h is updated with the changes.
cb5027f2
YO
343 *
344 */
bbe252da 345
bbe252da 346#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
347#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
348#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
349#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
350#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
351#define RX_MATCH_TAINTED_set(prog, t) ((t) \
352 ? RX_MATCH_TAINTED_on(prog) \
353 : RX_MATCH_TAINTED_off(prog))
c277df42 354
07bc277f
NC
355#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
356#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
357#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
358#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
359#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
360#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
361#define RX_MATCH_COPIED_set(prog,t) ((t) \
362 ? RX_MATCH_COPIED_on(prog) \
363 : RX_MATCH_COPIED_off(prog))
f9f4320a 364
07bc277f 365#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 366
07bc277f 367/* For source compatibility. We used to store these explicitly. */
f7c278bf 368#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 369#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
370/* FIXME? Are we hardcoding too much here and constraining plugin extension
371 writers? Specifically, the value 1 assumes that the wrapped version always
372 has exactly one character at the end, a ')'. Will that always be true? */
373#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 374#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 375#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 376#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 377#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 378#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 379#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
380# define RX_EXTFLAGS(prog) \
381 (*({ \
7948fc08
RU
382 const REGEXP *const _rx_extflags = (prog); \
383 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
384 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
385 }))
386# define RX_ENGINE(prog) \
387 (*({ \
7948fc08
RU
388 const REGEXP *const _rx_engine = (prog); \
389 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
390 &SvANY(_rx_engine)->engine; \
4450b630
NC
391 }))
392# define RX_SUBBEG(prog) \
393 (*({ \
7948fc08
RU
394 const REGEXP *const _rx_subbeg = (prog); \
395 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
396 &SvANY(_rx_subbeg)->subbeg; \
4450b630 397 }))
200ff815 398# define RX_OFFS(prog) \
4450b630 399 (*({ \
7948fc08
RU
400 const REGEXP *const _rx_offs = (prog); \
401 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
402 &SvANY(_rx_offs)->offs; \
4450b630
NC
403 }))
404# define RX_NPARENS(prog) \
405 (*({ \
7948fc08
RU
406 const REGEXP *const _rx_nparens = (prog); \
407 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
408 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
409 }))
410#else
411# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
412# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
413# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
414# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
415# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
416#endif
288b8c02 417#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
418#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
419#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
420#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
421#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
422#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
423#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 424#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 425
be8e71aa
YO
426#endif /* PLUGGABLE_RE_EXTENSION */
427
428/* Stuff that needs to be included in the plugable extension goes below here */
429
f8c7b90f 430#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 431#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
432 STMT_START {if (RX_SAVED_COPY(rx)) { \
433 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
434 } \
435 if (RX_MATCH_COPIED(rx)) { \
07bc277f 436 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
437 RX_MATCH_COPIED_off(rx); \
438 }} STMT_END
439#else
440#define RX_MATCH_COPY_FREE(rx) \
441 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 442 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
443 RX_MATCH_COPIED_off(rx); \
444 }} STMT_END
445#endif
446
07bc277f
NC
447#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
448#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
449#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
450#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
451#define RX_MATCH_UTF8_set(prog, t) ((t) \
452 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
453 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
454
455/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 456#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 457
cad2e5aa
JH
458#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
459#define REXEC_CHECKED 0x02 /* check_substr already checked. */
460#define REXEC_SCREAM 0x04 /* use scream table. */
461#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 462#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 463
7fe44985 464#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
465# define ReREFCNT_inc(re) \
466 ({ \
467 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
468 REGEXP *const _rerefcnt_inc = (re); \
469 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
470 SvREFCNT_inc(_rerefcnt_inc); \
471 _rerefcnt_inc; \
288b8c02
NC
472 })
473# define ReREFCNT_dec(re) \
474 ({ \
475 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
476 REGEXP *const _rerefcnt_dec = (re); \
477 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
478 })
479#else
480# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 481# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
482#endif
483
484/* FIXME for plugins. */
cf93c79d
IZ
485
486#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
487#define FBMcf_TAIL_DOLLARM 2
488#define FBMcf_TAIL_Z 4
489#define FBMcf_TAIL_z 8
490#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
491
492#define FBMrf_MULTILINE 1
cad2e5aa 493
3b0527fe
DM
494/* some basic information about the current match that is created by
495 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
496
497typedef struct {
288b8c02 498 REGEXP *prog;
3b0527fe
DM
499 char *bol;
500 char *till;
501 SV *sv;
502 char *ganch;
24b23f37 503 char *cutpoint;
3b0527fe
DM
504} regmatch_info;
505
5d9a96ca
DM
506
507/* structures for holding and saving the state maintained by regmatch() */
508
d56b3014 509#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 510#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 511#endif
6bda09f9 512
5d9a96ca
DM
513typedef I32 CHECKPOINT;
514
a0374537 515typedef struct regmatch_state {
40a82448 516 int resume_state; /* where to jump to on return */
24d3c4a9 517 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
518
519 union {
77cb431f
DM
520
521 /* this is a fake union member that matches the first element
522 * of each member that needs to store positive backtrack
523 * information */
524 struct {
525 struct regmatch_state *prev_yes_state;
526 } yes;
527
fae667d5
YO
528 /* branchlike members */
529 /* this is a fake union member that matches the first elements
530 * of each member that needs to behave like a branch */
531 struct {
5d458dd8
YO
532 /* this first element must match u.yes */
533 struct regmatch_state *prev_yes_state;
fae667d5
YO
534 U32 lastparen;
535 CHECKPOINT cp;
536
537 } branchlike;
538
539 struct {
540 /* the first elements must match u.branchlike */
541 struct regmatch_state *prev_yes_state;
542 U32 lastparen;
543 CHECKPOINT cp;
544
545 regnode *next_branch; /* next branch node */
546 } branch;
547
548 struct {
549 /* the first elements must match u.branchlike */
550 struct regmatch_state *prev_yes_state;
551 U32 lastparen;
552 CHECKPOINT cp;
553
2e64971a 554 U32 accepted; /* how many accepting states left */
7f69552c 555 U16 *jump; /* positive offsets from me */
166ba7cd 556 regnode *B; /* node following the trie */
7f69552c 557 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
558 U8 *firstpos;/* pos in string of first trie match */
559 U32 firstchars;/* len in chars of firstpos from start */
560 U16 nextword;/* next word to try */
561 U16 topword; /* longest accepted word */
562 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
563 } trie;
564
fae667d5
YO
565 /* special types - these members are used to store state for special
566 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 567 struct {
77cb431f
DM
568 /* this first element must match u.yes */
569 struct regmatch_state *prev_yes_state;
faec1544
DM
570 struct regmatch_state *prev_eval;
571 struct regmatch_state *prev_curlyx;
288b8c02 572 REGEXP *prev_rex;
faec1544
DM
573 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
574 flip when transitioning between
575 inner and outer rexen */
aa283a38
DM
576 CHECKPOINT cp; /* remember current savestack indexes */
577 CHECKPOINT lastcp;
6bda09f9 578 U32 close_paren; /* which close bracket is our end */
1b5c067c 579 regnode *B; /* the node following us */
e822a8b4
DM
580 } eval;
581
582 struct {
c476f425 583 /* this first element must match u.yes */
262b90c4 584 struct regmatch_state *prev_yes_state;
fae667d5
YO
585 I32 wanted;
586 I32 logical; /* saved copy of 'logical' var */
587 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
588 } ifmatch; /* and SUSPEND/UNLESSM */
589
590 struct {
591 /* this first element must match u.yes */
592 struct regmatch_state *prev_yes_state;
593 struct regmatch_state *prev_mark;
594 SV* mark_name;
595 char *mark_loc;
596 } mark;
597
598 struct {
599 int val;
600 } keeper;
601
602 /* quantifiers - these members are used for storing state for
603 for the regops used to implement quantifiers */
604 struct {
605 /* this first element must match u.yes */
606 struct regmatch_state *prev_yes_state;
c476f425 607 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
608 regnode *me; /* the CURLYX node */
609 regnode *B; /* the B node in /A*B/ */
c476f425
DM
610 CHECKPOINT cp; /* remember current savestack index */
611 bool minmod;
a0374537 612 int parenfloor;/* how far back to strip paren data */
c476f425
DM
613
614 /* these two are modified by WHILEM */
615 int count; /* how many instances of A we've matched */
616 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
617 } curlyx;
618
619 struct {
c476f425 620 /* this first element must match u.yes */
262b90c4 621 struct regmatch_state *prev_yes_state;
c476f425
DM
622 struct regmatch_state *save_curlyx;
623 CHECKPOINT cp; /* remember current savestack indexes */
624 CHECKPOINT lastcp;
625 char *save_lastloc; /* previous curlyx.lastloc */
626 I32 cache_offset;
627 I32 cache_mask;
e822a8b4
DM
628 } whilem;
629
630 struct {
5d458dd8
YO
631 /* this first element must match u.yes */
632 struct regmatch_state *prev_yes_state;
e822a8b4 633 I32 c1, c2; /* case fold search */
40a82448
DM
634 CHECKPOINT cp;
635 I32 alen; /* length of first-matched A string */
636 I32 count;
0cadcf80 637 bool minmod;
40a82448
DM
638 regnode *A, *B; /* the nodes corresponding to /A*B/ */
639 regnode *me; /* the curlym node */
e822a8b4
DM
640 } curlym;
641
642 struct {
3b6647e0 643 U32 paren;
c255a977 644 CHECKPOINT cp;
e822a8b4 645 I32 c1, c2; /* case fold search */
c255a977
DM
646 char *maxpos; /* highest possible point in string to match */
647 char *oldloc; /* the previous locinput */
e822a8b4 648 int count;
c255a977
DM
649 int min, max; /* {m,n} */
650 regnode *A, *B; /* the nodes corresponding to /A*B/ */
651 } curly; /* and CURLYN/PLUS/STAR */
dad79028 652
d8319b27 653 } u;
5d9a96ca
DM
654} regmatch_state;
655
656/* how many regmatch_state structs to allocate as a single slab.
657 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
658 * pointers, plus 1 for any mythical malloc overhead. */
659
660#define PERL_REGMATCH_SLAB_SLOTS \
661 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
662
663typedef struct regmatch_slab {
664 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
665 struct regmatch_slab *prev, *next;
666} regmatch_slab;
1ade1aa1 667
46ab3289
NC
668#define PL_reg_flags PL_reg_state.re_state_reg_flags
669#define PL_bostr PL_reg_state.re_state_bostr
670#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 671#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 672#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
673#define PL_reglastparen PL_reg_state.re_state_reglastparen
674#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
675#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
676#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
677#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
678#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
679#define PL_reg_magic PL_reg_state.re_state_reg_magic
680#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
681#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
682#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
683#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
684#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
685#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
686#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
687#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
688#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
689#define PL_regsize PL_reg_state.re_state_regsize
690#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
691#define PL_nrs PL_reg_state.re_state_nrs
692
1ade1aa1
NC
693struct re_save_state {
694 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
695 U32 re_state_reg_start_tmpl; /* from regexec.c */
696 I32 re_state_reg_eval_set; /* from regexec.c */
697 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
698 char *re_state_bostr;
699 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 700 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 701 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
702 U32 *re_state_reglastparen; /* Similarly for lastparen. */
703 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 704 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 705 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
706 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
707 PMOP *re_state_reg_curpm; /* from regexec.c */
708 char *re_state_reg_oldsaved; /* old saved substr during match */
709 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
710 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
711 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
712 I32 re_state_reg_maxiter; /* max wait until caching pos */
713 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 714 U32 re_state_regsize; /* from regexec.c */
1b5c067c 715 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
716 char *re_state_reg_starttry; /* from regexec.c */
717#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 718 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 */
1ade1aa1
NC
719#endif
720};
721
722#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
723 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 724
1ade1aa1
NC
725/*
726 * Local variables:
727 * c-indentation-style: bsd
728 * c-basic-offset: 4
729 * indent-tabs-mode: t
730 * End:
731 *
732 * ex: set ts=8 sts=4 sw=4 noet:
733 */