This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Abolish wraplen from struct regexp. We're already storing it in SvCUR.
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
663f364b 4 * 2005, 2006, 2007, 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
bbe252da 69typedef struct regexp {
288b8c02
NC
70 _XPV_HEAD;
71 _XPVMG_HEAD;
f8149455
YO
72 /* what engine created this regexp? */
73 const struct regexp_engine* engine;
288b8c02 74 REGEXP *mother_re; /* what re is this a lightweight copy of? */
f8149455
YO
75
76 /* Information about the match that the perl core uses to manage things */
bbe252da 77 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO
78 I32 minlen; /* mininum possible length of string to match */
79 I32 minlenret; /* mininum possible length of $& */
58e23c8d 80 U32 gofs; /* chars left of pos that we search from */
f8149455 81 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 82 in the final match, used for optimisations */
f8149455 83 U32 nparens; /* number of capture buffers */
bbe252da 84
f8149455
YO
85 /* private engine specific data */
86 U32 intflags; /* Engine Specific Internal flags */
87 void *pprivate; /* Data private to the regex engine which
88 created this object. */
89
90 /* Data about the last/current match. These are modified during matching*/
91 U32 lastparen; /* last open paren matched */
92 U32 lastcloseparen; /* last close paren matched */
f0ab9afb
NC
93 regexp_paren_pair *swap; /* Swap copy of *offs */
94 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
28d8d7f4 95
bbe252da
YO
96 char *subbeg; /* saved or original string
97 so \digit works forever. */
bbe252da 98 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
1b5c067c 99 I32 sublen; /* Length of string pointed by subbeg */
f8149455
YO
100
101
102 /* Information about the match that isn't often used */
2c96c0d6
NC
103 unsigned pre_prefix:4; /* offset from wrapped to the start of precomp */
104 unsigned seen_evals:28; /* number of eval groups in the pattern - for security checks */
f8149455 105 HV *paren_names; /* Optional hash of paren names */
f8fc2ecf
YO
106} regexp;
107
785a26d5 108/* used for high speed searches */
f9f4320a
YO
109typedef struct re_scream_pos_data_s
110{
111 char **scream_olds; /* match pos */
112 I32 *scream_pos; /* Internal iterator of scream. */
113} re_scream_pos_data;
114
785a26d5
YO
115/* regexp_engine structure. This is the dispatch table for regexes.
116 * Any regex engine implementation must be able to build one of these.
117 */
f9f4320a 118typedef struct regexp_engine {
3ab4a224 119 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
49d7dfbc
AB
120 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
121 char* strbeg, I32 minend, SV* screamer,
122 void* data, U32 flags);
123 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
124 char *strend, const U32 flags,
125 re_scream_pos_data *data);
49d7dfbc
AB
126 SV* (*checkstr) (pTHX_ REGEXP * const rx);
127 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 128 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 129 SV * const sv);
2fdbfb4d
AB
130 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
131 SV const * const value);
132 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
133 const I32 paren);
192b9cd1
AB
134 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
135 SV * const value, const U32 flags);
136 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
137 const U32 flags);
49d7dfbc 138 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 139#ifdef USE_ITHREADS
49d7dfbc 140 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 141#endif
f9f4320a
YO
142} regexp_engine;
143
192b9cd1
AB
144/*
145 These are passed to the numbered capture variable callbacks as the
146 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
147 $2 etc.
148*/
f1b875a0
YO
149#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
150#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
151#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
152
153/*
154 Flags that are passed to the named_buff and named_buff_iter
155 callbacks above. Those routines are called from universal.c via the
156 Tie::Hash::NamedCapture interface for %+ and %- and the re::
157 functions in the same file.
158*/
159
160/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
161#define RXapif_FETCH 0x0001
162#define RXapif_STORE 0x0002
163#define RXapif_DELETE 0x0004
164#define RXapif_CLEAR 0x0008
165#define RXapif_EXISTS 0x0010
166#define RXapif_SCALAR 0x0020
167#define RXapif_FIRSTKEY 0x0040
168#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
169
170/* Whether %+ or %- is being operated on */
f1b875a0
YO
171#define RXapif_ONE 0x0100 /* %+ */
172#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
173
174/* Whether this is being called from a re:: function */
f1b875a0
YO
175#define RXapif_REGNAME 0x0400
176#define RXapif_REGNAMES 0x0800
177#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 178
f7e71195
AB
179/*
180=head1 REGEXP Functions
181
182=for apidoc Am|REGEXP *|SvRX|SV *sv
183
184Convenience macro to get the REGEXP from a SV. This is approximately
185equivalent to the following snippet:
186
187 if (SvMAGICAL(sv))
188 mg_get(sv);
189 if (SvROK(sv) &&
190 (tmpsv = (SV*)SvRV(sv)) &&
191 SvTYPE(tmpsv) == SVt_PVMG &&
192 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
193 {
194 return (REGEXP *)tmpmg->mg_obj;
195 }
196
197NULL will be returned if a REGEXP* is not found.
198
199=for apidoc Am|bool|SvRXOK|SV* sv
200
201Returns a boolean indicating whether the SV contains qr magic
202(PERL_MAGIC_qr).
203
204If you want to do something with the REGEXP* later use SvRX instead
205and check for NULL.
206
207=cut
208*/
209
210#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
211#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
212
213
bbe252da
YO
214/* Flags stored in regexp->extflags
215 * These are used by code external to the regexp engine
e357fc67
YO
216 *
217 * Note that flags starting with RXf_PMf_ have exact equivalents
218 * stored in op_pmflags and which are defined in op.h, they are defined
219 * numerically here only for clarity.
cb5027f2
YO
220 *
221 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
222 * so that regnodes.h is updated with the changes.
223 *
bbe252da
YO
224 */
225
52d81aa8 226/* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
a1cf41cd
NC
227 * If you change these you need to change the equivalent flags in op.h, and
228 * vice versa. */
52d81aa8
NC
229#define RXf_PMf_MULTILINE 0x00000001 /* /m */
230#define RXf_PMf_SINGLELINE 0x00000002 /* /s */
231#define RXf_PMf_FOLD 0x00000004 /* /i */
232#define RXf_PMf_EXTENDED 0x00000008 /* /x */
233#define RXf_PMf_KEEPCOPY 0x00000010 /* /p */
234#define RXf_PMf_LOCALE 0x00000020 /* use locale */
e357fc67 235/* these flags are transfered from the PMOP->op_pmflags member during compilation */
52d81aa8 236#define RXf_PMf_STD_PMMOD_SHIFT 0
cde0cee5
YO
237#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
238#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
239
a20207d7
YO
240#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
241 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
242 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
243 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
244 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 245
bcdf7404
YO
246/* chars and strings used as regex pattern modifiers
247 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
248 *
249 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
250 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
251 * for its own uses. So 'k' is out as well.
bcdf7404
YO
252 */
253#define EXEC_PAT_MOD 'e'
87e95b7f 254#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
255#define ONCE_PAT_MOD 'o'
256#define GLOBAL_PAT_MOD 'g'
257#define CONTINUE_PAT_MOD 'c'
258#define MULTILINE_PAT_MOD 'm'
259#define SINGLE_PAT_MOD 's'
260#define IGNORE_PAT_MOD 'i'
261#define XTENDED_PAT_MOD 'x'
262
263#define ONCE_PAT_MODS "o"
87e95b7f 264#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
265#define EXEC_PAT_MODS "e"
266#define LOOP_PAT_MODS "gc"
267
268#define STD_PAT_MODS "msix"
269
270#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
271
272#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
273#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
274#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
275#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
276
cb5027f2
YO
277/*
278 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
279 * so that regnodes.h is updated with the changes.
280 *
281 */
bcdf7404 282
52d81aa8
NC
283/* Anchor and GPOS related stuff */
284#define RXf_ANCH_BOL 0x00000100
285#define RXf_ANCH_MBOL 0x00000200
286#define RXf_ANCH_SBOL 0x00000400
287#define RXf_ANCH_GPOS 0x00000800
288#define RXf_GPOS_SEEN 0x00001000
289#define RXf_GPOS_FLOAT 0x00002000
290/* two bits here */
291#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
292#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
293#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
294
bbe252da 295/* What we have seen */
52d81aa8
NC
296#define RXf_LOOKBEHIND_SEEN 0x00004000
297#define RXf_EVAL_SEEN 0x00008000
298#define RXf_CANY_SEEN 0x00010000
bbe252da
YO
299
300/* Special */
52d81aa8
NC
301#define RXf_NOSCAN 0x00020000
302#define RXf_CHECK_ALL 0x00040000
bbe252da
YO
303
304/* UTF8 related */
52d81aa8
NC
305#define RXf_UTF8 0x00080000
306#define RXf_MATCH_UTF8 0x00100000
bbe252da
YO
307
308/* Intuit related */
52d81aa8
NC
309#define RXf_USE_INTUIT_NOML 0x00200000
310#define RXf_USE_INTUIT_ML 0x00400000
311#define RXf_INTUIT_TAIL 0x00800000
0ac6acae
AB
312
313/*
314 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
315 be used by regex engines to check whether they should set
316 RXf_SKIPWHITE
317*/
52d81aa8 318#define RXf_SPLIT 0x01000000
0ac6acae 319
bbe252da
YO
320#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
321
322/* Copy and tainted info */
52d81aa8
NC
323#define RXf_COPY_DONE 0x02000000
324#define RXf_TAINTED_SEEN 0x04000000
325#define RXf_TAINTED 0x08000000 /* this pattern is tainted */
326
327/* Flags indicating special patterns */
328#define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */
329#define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */
330#define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */
331#define RXf_NULL 0x80000000 /* Pattern is // */
c737faaf 332
cb5027f2
YO
333/*
334 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
335 * so that regnodes.h is updated with the changes.
336 *
337 */
bbe252da 338
bbe252da 339#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
340#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
341#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
342#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
343#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
344#define RX_MATCH_TAINTED_set(prog, t) ((t) \
345 ? RX_MATCH_TAINTED_on(prog) \
346 : RX_MATCH_TAINTED_off(prog))
c277df42 347
07bc277f
NC
348#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
349#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
350#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
351#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
352#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
353#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
354#define RX_MATCH_COPIED_set(prog,t) ((t) \
355 ? RX_MATCH_COPIED_on(prog) \
356 : RX_MATCH_COPIED_off(prog))
f9f4320a 357
07bc277f 358#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 359
07bc277f 360/* For source compatibility. We used to store these explicitly. */
f7c278bf 361#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
362/* FIXME? Are we hardcoding too much here and constraining plugin extension
363 writers? Specifically, the value 1 assumes that the wrapped version always
364 has exactly one character at the end, a ')'. Will that always be true? */
365#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 366#define RX_WRAPPED(prog) SvPVX(prog)
9d17798d 367#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02
NC
368#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
369#define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
370#define RX_REFCNT(prog) SvREFCNT(prog)
371#define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
372#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
373#define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
374#define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
375#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
376#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
377#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
378#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
379#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
380#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
381#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
382#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 383#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 384
be8e71aa
YO
385#endif /* PLUGGABLE_RE_EXTENSION */
386
387/* Stuff that needs to be included in the plugable extension goes below here */
388
f8c7b90f 389#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 390#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
391 STMT_START {if (RX_SAVED_COPY(rx)) { \
392 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
393 } \
394 if (RX_MATCH_COPIED(rx)) { \
07bc277f 395 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
396 RX_MATCH_COPIED_off(rx); \
397 }} STMT_END
398#else
399#define RX_MATCH_COPY_FREE(rx) \
400 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 401 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
402 RX_MATCH_COPIED_off(rx); \
403 }} STMT_END
404#endif
405
07bc277f
NC
406#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
407#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
408#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
409#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
410#define RX_MATCH_UTF8_set(prog, t) ((t) \
411 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
412 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
413
414/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
415#define RX_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_UTF8)
a30b2f1f 416
cad2e5aa
JH
417#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
418#define REXEC_CHECKED 0x02 /* check_substr already checked. */
419#define REXEC_SCREAM 0x04 /* use scream table. */
420#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 421#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 422
288b8c02
NC
423#if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC)
424# define ReREFCNT_inc(re) \
425 ({ \
426 /* This is here to generate a casting warning if incorrect. */ \
427 REGEXP *const zwapp = (re); \
428 SvREFCNT_inc(zwapp); \
429 })
430# define ReREFCNT_dec(re) \
431 ({ \
432 /* This is here to generate a casting warning if incorrect. */ \
433 REGEXP *const boff = (re); \
434 SvREFCNT_dec(boff); \
435 })
436#else
437# define ReREFCNT_dec(re) SvREFCNT_dec(re)
438# define ReREFCNT_inc(re) SvREFCNT_inc(re)
439#endif
440
441/* FIXME for plugins. */
cf93c79d
IZ
442
443#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
444#define FBMcf_TAIL_DOLLARM 2
445#define FBMcf_TAIL_Z 4
446#define FBMcf_TAIL_z 8
447#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
448
449#define FBMrf_MULTILINE 1
cad2e5aa 450
5d9a96ca
DM
451/* an accepting state/position*/
452struct _reg_trie_accepted {
453 U8 *endpos;
454 U16 wordnum;
455};
456typedef struct _reg_trie_accepted reg_trie_accepted;
457
3b0527fe
DM
458/* some basic information about the current match that is created by
459 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
460
461typedef struct {
288b8c02 462 REGEXP *prog;
3b0527fe
DM
463 char *bol;
464 char *till;
465 SV *sv;
466 char *ganch;
24b23f37 467 char *cutpoint;
3b0527fe
DM
468} regmatch_info;
469
5d9a96ca
DM
470
471/* structures for holding and saving the state maintained by regmatch() */
472
d56b3014 473#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 474#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 475#endif
6bda09f9 476
5d9a96ca
DM
477typedef I32 CHECKPOINT;
478
a0374537 479typedef struct regmatch_state {
40a82448 480 int resume_state; /* where to jump to on return */
24d3c4a9 481 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
482
483 union {
77cb431f
DM
484
485 /* this is a fake union member that matches the first element
486 * of each member that needs to store positive backtrack
487 * information */
488 struct {
489 struct regmatch_state *prev_yes_state;
490 } yes;
491
fae667d5
YO
492 /* branchlike members */
493 /* this is a fake union member that matches the first elements
494 * of each member that needs to behave like a branch */
495 struct {
5d458dd8
YO
496 /* this first element must match u.yes */
497 struct regmatch_state *prev_yes_state;
fae667d5
YO
498 U32 lastparen;
499 CHECKPOINT cp;
500
501 } branchlike;
502
503 struct {
504 /* the first elements must match u.branchlike */
505 struct regmatch_state *prev_yes_state;
506 U32 lastparen;
507 CHECKPOINT cp;
508
509 regnode *next_branch; /* next branch node */
510 } branch;
511
512 struct {
513 /* the first elements must match u.branchlike */
514 struct regmatch_state *prev_yes_state;
515 U32 lastparen;
516 CHECKPOINT cp;
517
518 reg_trie_accepted *accept_buff; /* accepting states we have seen */
166ba7cd 519 U32 accepted; /* how many accepting states we have seen */
7f69552c 520 U16 *jump; /* positive offsets from me */
166ba7cd 521 regnode *B; /* node following the trie */
7f69552c 522 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
523 } trie;
524
fae667d5
YO
525 /* special types - these members are used to store state for special
526 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 527 struct {
77cb431f
DM
528 /* this first element must match u.yes */
529 struct regmatch_state *prev_yes_state;
faec1544
DM
530 struct regmatch_state *prev_eval;
531 struct regmatch_state *prev_curlyx;
288b8c02 532 REGEXP *prev_rex;
faec1544
DM
533 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
534 flip when transitioning between
535 inner and outer rexen */
aa283a38
DM
536 CHECKPOINT cp; /* remember current savestack indexes */
537 CHECKPOINT lastcp;
6bda09f9 538 U32 close_paren; /* which close bracket is our end */
1b5c067c 539 regnode *B; /* the node following us */
e822a8b4
DM
540 } eval;
541
542 struct {
c476f425 543 /* this first element must match u.yes */
262b90c4 544 struct regmatch_state *prev_yes_state;
fae667d5
YO
545 I32 wanted;
546 I32 logical; /* saved copy of 'logical' var */
547 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
548 } ifmatch; /* and SUSPEND/UNLESSM */
549
550 struct {
551 /* this first element must match u.yes */
552 struct regmatch_state *prev_yes_state;
553 struct regmatch_state *prev_mark;
554 SV* mark_name;
555 char *mark_loc;
556 } mark;
557
558 struct {
559 int val;
560 } keeper;
561
562 /* quantifiers - these members are used for storing state for
563 for the regops used to implement quantifiers */
564 struct {
565 /* this first element must match u.yes */
566 struct regmatch_state *prev_yes_state;
c476f425 567 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 568 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425
DM
569 CHECKPOINT cp; /* remember current savestack index */
570 bool minmod;
a0374537 571 int parenfloor;/* how far back to strip paren data */
c476f425
DM
572 int min; /* the minimal number of A's to match */
573 int max; /* the maximal number of A's to match */
c476f425
DM
574
575 /* these two are modified by WHILEM */
576 int count; /* how many instances of A we've matched */
577 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
578 } curlyx;
579
580 struct {
c476f425 581 /* this first element must match u.yes */
262b90c4 582 struct regmatch_state *prev_yes_state;
c476f425
DM
583 struct regmatch_state *save_curlyx;
584 CHECKPOINT cp; /* remember current savestack indexes */
585 CHECKPOINT lastcp;
586 char *save_lastloc; /* previous curlyx.lastloc */
587 I32 cache_offset;
588 I32 cache_mask;
e822a8b4
DM
589 } whilem;
590
591 struct {
5d458dd8
YO
592 /* this first element must match u.yes */
593 struct regmatch_state *prev_yes_state;
e822a8b4 594 I32 c1, c2; /* case fold search */
40a82448
DM
595 CHECKPOINT cp;
596 I32 alen; /* length of first-matched A string */
597 I32 count;
0cadcf80 598 bool minmod;
40a82448
DM
599 regnode *A, *B; /* the nodes corresponding to /A*B/ */
600 regnode *me; /* the curlym node */
e822a8b4
DM
601 } curlym;
602
603 struct {
3b6647e0 604 U32 paren;
c255a977 605 CHECKPOINT cp;
e822a8b4 606 I32 c1, c2; /* case fold search */
c255a977
DM
607 char *maxpos; /* highest possible point in string to match */
608 char *oldloc; /* the previous locinput */
e822a8b4 609 int count;
c255a977
DM
610 int min, max; /* {m,n} */
611 regnode *A, *B; /* the nodes corresponding to /A*B/ */
612 } curly; /* and CURLYN/PLUS/STAR */
dad79028 613
d8319b27 614 } u;
5d9a96ca
DM
615} regmatch_state;
616
617/* how many regmatch_state structs to allocate as a single slab.
618 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
619 * pointers, plus 1 for any mythical malloc overhead. */
620
621#define PERL_REGMATCH_SLAB_SLOTS \
622 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
623
624typedef struct regmatch_slab {
625 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
626 struct regmatch_slab *prev, *next;
627} regmatch_slab;
1ade1aa1 628
46ab3289
NC
629#define PL_reg_flags PL_reg_state.re_state_reg_flags
630#define PL_bostr PL_reg_state.re_state_bostr
631#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 632#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 633#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
634#define PL_reglastparen PL_reg_state.re_state_reglastparen
635#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
636#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
637#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
638#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
639#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
640#define PL_reg_magic PL_reg_state.re_state_reg_magic
641#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
642#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
643#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
644#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
645#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
646#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
647#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
648#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
649#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
650#define PL_regsize PL_reg_state.re_state_regsize
651#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
652#define PL_nrs PL_reg_state.re_state_nrs
653
1ade1aa1
NC
654struct re_save_state {
655 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
656 U32 re_state_reg_start_tmpl; /* from regexec.c */
657 I32 re_state_reg_eval_set; /* from regexec.c */
658 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
659 char *re_state_bostr;
660 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 661 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 662 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
663 U32 *re_state_reglastparen; /* Similarly for lastparen. */
664 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 665 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 666 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
667 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
668 PMOP *re_state_reg_curpm; /* from regexec.c */
669 char *re_state_reg_oldsaved; /* old saved substr during match */
670 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
671 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
672 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
673 I32 re_state_reg_maxiter; /* max wait until caching pos */
674 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 675 U32 re_state_regsize; /* from regexec.c */
1b5c067c 676 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
677 char *re_state_reg_starttry; /* from regexec.c */
678#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 679 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
680#endif
681};
682
683#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
684 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 685
1ade1aa1
NC
686/*
687 * Local variables:
688 * c-indentation-style: bsd
689 * c-basic-offset: 4
690 * indent-tabs-mode: t
691 * End:
692 *
693 * ex: set ts=8 sts=4 sw=4 noet:
694 */