This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Convert all accesses of the member paren_names of struct regexp to
[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
5daac39c
NC
108#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
109
785a26d5 110/* used for high speed searches */
f9f4320a
YO
111typedef struct re_scream_pos_data_s
112{
113 char **scream_olds; /* match pos */
114 I32 *scream_pos; /* Internal iterator of scream. */
115} re_scream_pos_data;
116
785a26d5
YO
117/* regexp_engine structure. This is the dispatch table for regexes.
118 * Any regex engine implementation must be able to build one of these.
119 */
f9f4320a 120typedef struct regexp_engine {
3ab4a224 121 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
49d7dfbc
AB
122 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
123 char* strbeg, I32 minend, SV* screamer,
124 void* data, U32 flags);
125 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
126 char *strend, const U32 flags,
127 re_scream_pos_data *data);
49d7dfbc
AB
128 SV* (*checkstr) (pTHX_ REGEXP * const rx);
129 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 130 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 131 SV * const sv);
2fdbfb4d
AB
132 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
133 SV const * const value);
134 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
135 const I32 paren);
192b9cd1
AB
136 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
137 SV * const value, const U32 flags);
138 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
139 const U32 flags);
49d7dfbc 140 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 141#ifdef USE_ITHREADS
49d7dfbc 142 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 143#endif
f9f4320a
YO
144} regexp_engine;
145
192b9cd1
AB
146/*
147 These are passed to the numbered capture variable callbacks as the
148 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
149 $2 etc.
150*/
f1b875a0
YO
151#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
152#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
153#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
154
155/*
156 Flags that are passed to the named_buff and named_buff_iter
157 callbacks above. Those routines are called from universal.c via the
158 Tie::Hash::NamedCapture interface for %+ and %- and the re::
159 functions in the same file.
160*/
161
162/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
163#define RXapif_FETCH 0x0001
164#define RXapif_STORE 0x0002
165#define RXapif_DELETE 0x0004
166#define RXapif_CLEAR 0x0008
167#define RXapif_EXISTS 0x0010
168#define RXapif_SCALAR 0x0020
169#define RXapif_FIRSTKEY 0x0040
170#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
171
172/* Whether %+ or %- is being operated on */
f1b875a0
YO
173#define RXapif_ONE 0x0100 /* %+ */
174#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
175
176/* Whether this is being called from a re:: function */
f1b875a0
YO
177#define RXapif_REGNAME 0x0400
178#define RXapif_REGNAMES 0x0800
179#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 180
f7e71195
AB
181/*
182=head1 REGEXP Functions
183
184=for apidoc Am|REGEXP *|SvRX|SV *sv
185
186Convenience macro to get the REGEXP from a SV. This is approximately
187equivalent to the following snippet:
188
189 if (SvMAGICAL(sv))
190 mg_get(sv);
191 if (SvROK(sv) &&
192 (tmpsv = (SV*)SvRV(sv)) &&
193 SvTYPE(tmpsv) == SVt_PVMG &&
194 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
195 {
196 return (REGEXP *)tmpmg->mg_obj;
197 }
198
199NULL will be returned if a REGEXP* is not found.
200
201=for apidoc Am|bool|SvRXOK|SV* sv
202
203Returns a boolean indicating whether the SV contains qr magic
204(PERL_MAGIC_qr).
205
206If you want to do something with the REGEXP* later use SvRX instead
207and check for NULL.
208
209=cut
210*/
211
212#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
213#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
214
215
bbe252da
YO
216/* Flags stored in regexp->extflags
217 * These are used by code external to the regexp engine
e357fc67
YO
218 *
219 * Note that flags starting with RXf_PMf_ have exact equivalents
220 * stored in op_pmflags and which are defined in op.h, they are defined
221 * numerically here only for clarity.
cb5027f2
YO
222 *
223 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
224 * so that regnodes.h is updated with the changes.
225 *
bbe252da
YO
226 */
227
52d81aa8 228/* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
a1cf41cd
NC
229 * If you change these you need to change the equivalent flags in op.h, and
230 * vice versa. */
52d81aa8
NC
231#define RXf_PMf_MULTILINE 0x00000001 /* /m */
232#define RXf_PMf_SINGLELINE 0x00000002 /* /s */
233#define RXf_PMf_FOLD 0x00000004 /* /i */
234#define RXf_PMf_EXTENDED 0x00000008 /* /x */
235#define RXf_PMf_KEEPCOPY 0x00000010 /* /p */
236#define RXf_PMf_LOCALE 0x00000020 /* use locale */
e357fc67 237/* these flags are transfered from the PMOP->op_pmflags member during compilation */
52d81aa8 238#define RXf_PMf_STD_PMMOD_SHIFT 0
cde0cee5
YO
239#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
240#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
241
a20207d7
YO
242#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
243 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
244 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
245 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
246 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 247
bcdf7404
YO
248/* chars and strings used as regex pattern modifiers
249 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
250 *
251 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
252 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
253 * for its own uses. So 'k' is out as well.
bcdf7404
YO
254 */
255#define EXEC_PAT_MOD 'e'
87e95b7f 256#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
257#define ONCE_PAT_MOD 'o'
258#define GLOBAL_PAT_MOD 'g'
259#define CONTINUE_PAT_MOD 'c'
260#define MULTILINE_PAT_MOD 'm'
261#define SINGLE_PAT_MOD 's'
262#define IGNORE_PAT_MOD 'i'
263#define XTENDED_PAT_MOD 'x'
264
265#define ONCE_PAT_MODS "o"
87e95b7f 266#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
267#define EXEC_PAT_MODS "e"
268#define LOOP_PAT_MODS "gc"
269
270#define STD_PAT_MODS "msix"
271
272#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
273
274#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
275#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
276#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
277#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
278
cb5027f2
YO
279/*
280 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
281 * so that regnodes.h is updated with the changes.
282 *
283 */
bcdf7404 284
52d81aa8
NC
285/* Anchor and GPOS related stuff */
286#define RXf_ANCH_BOL 0x00000100
287#define RXf_ANCH_MBOL 0x00000200
288#define RXf_ANCH_SBOL 0x00000400
289#define RXf_ANCH_GPOS 0x00000800
290#define RXf_GPOS_SEEN 0x00001000
291#define RXf_GPOS_FLOAT 0x00002000
292/* two bits here */
293#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
294#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
295#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
296
bbe252da 297/* What we have seen */
52d81aa8
NC
298#define RXf_LOOKBEHIND_SEEN 0x00004000
299#define RXf_EVAL_SEEN 0x00008000
300#define RXf_CANY_SEEN 0x00010000
bbe252da
YO
301
302/* Special */
52d81aa8
NC
303#define RXf_NOSCAN 0x00020000
304#define RXf_CHECK_ALL 0x00040000
bbe252da
YO
305
306/* UTF8 related */
52d81aa8 307#define RXf_MATCH_UTF8 0x00100000
bbe252da
YO
308
309/* Intuit related */
52d81aa8
NC
310#define RXf_USE_INTUIT_NOML 0x00200000
311#define RXf_USE_INTUIT_ML 0x00400000
312#define RXf_INTUIT_TAIL 0x00800000
0ac6acae
AB
313
314/*
315 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
316 be used by regex engines to check whether they should set
317 RXf_SKIPWHITE
318*/
52d81aa8 319#define RXf_SPLIT 0x01000000
0ac6acae 320
bbe252da
YO
321#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
322
323/* Copy and tainted info */
52d81aa8
NC
324#define RXf_COPY_DONE 0x02000000
325#define RXf_TAINTED_SEEN 0x04000000
326#define RXf_TAINTED 0x08000000 /* this pattern is tainted */
327
328/* Flags indicating special patterns */
329#define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */
330#define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */
331#define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */
332#define RXf_NULL 0x80000000 /* Pattern is // */
c737faaf 333
cb5027f2
YO
334/*
335 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
336 * so that regnodes.h is updated with the changes.
337 *
338 */
bbe252da 339
bbe252da 340#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
341#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
342#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
343#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
344#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
345#define RX_MATCH_TAINTED_set(prog, t) ((t) \
346 ? RX_MATCH_TAINTED_on(prog) \
347 : RX_MATCH_TAINTED_off(prog))
c277df42 348
07bc277f
NC
349#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
350#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
351#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
352#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
353#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
354#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
355#define RX_MATCH_COPIED_set(prog,t) ((t) \
356 ? RX_MATCH_COPIED_on(prog) \
357 : RX_MATCH_COPIED_off(prog))
f9f4320a 358
07bc277f 359#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 360
07bc277f 361/* For source compatibility. We used to store these explicitly. */
f7c278bf 362#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
363/* FIXME? Are we hardcoding too much here and constraining plugin extension
364 writers? Specifically, the value 1 assumes that the wrapped version always
365 has exactly one character at the end, a ')'. Will that always be true? */
366#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 367#define RX_WRAPPED(prog) SvPVX(prog)
9d17798d 368#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02
NC
369#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
370#define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
371#define RX_REFCNT(prog) SvREFCNT(prog)
372#define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
373#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
374#define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
375#define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
376#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
377#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
378#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
379#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
380#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
381#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
382#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
383#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 384#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 385
be8e71aa
YO
386#endif /* PLUGGABLE_RE_EXTENSION */
387
388/* Stuff that needs to be included in the plugable extension goes below here */
389
f8c7b90f 390#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 391#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
392 STMT_START {if (RX_SAVED_COPY(rx)) { \
393 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
394 } \
395 if (RX_MATCH_COPIED(rx)) { \
07bc277f 396 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
397 RX_MATCH_COPIED_off(rx); \
398 }} STMT_END
399#else
400#define RX_MATCH_COPY_FREE(rx) \
401 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 402 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
403 RX_MATCH_COPIED_off(rx); \
404 }} STMT_END
405#endif
406
07bc277f
NC
407#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
408#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
409#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
410#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
411#define RX_MATCH_UTF8_set(prog, t) ((t) \
412 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
413 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
414
415/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 416#define RX_UTF8(prog) SvUTF8(prog)
a30b2f1f 417
cad2e5aa
JH
418#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
419#define REXEC_CHECKED 0x02 /* check_substr already checked. */
420#define REXEC_SCREAM 0x04 /* use scream table. */
421#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 422#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 423
288b8c02
NC
424#if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC)
425# define ReREFCNT_inc(re) \
426 ({ \
427 /* This is here to generate a casting warning if incorrect. */ \
428 REGEXP *const zwapp = (re); \
429 SvREFCNT_inc(zwapp); \
430 })
431# define ReREFCNT_dec(re) \
432 ({ \
433 /* This is here to generate a casting warning if incorrect. */ \
434 REGEXP *const boff = (re); \
435 SvREFCNT_dec(boff); \
436 })
437#else
438# define ReREFCNT_dec(re) SvREFCNT_dec(re)
439# define ReREFCNT_inc(re) SvREFCNT_inc(re)
440#endif
441
442/* FIXME for plugins. */
cf93c79d
IZ
443
444#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
445#define FBMcf_TAIL_DOLLARM 2
446#define FBMcf_TAIL_Z 4
447#define FBMcf_TAIL_z 8
448#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
449
450#define FBMrf_MULTILINE 1
cad2e5aa 451
5d9a96ca
DM
452/* an accepting state/position*/
453struct _reg_trie_accepted {
454 U8 *endpos;
455 U16 wordnum;
456};
457typedef struct _reg_trie_accepted reg_trie_accepted;
458
3b0527fe
DM
459/* some basic information about the current match that is created by
460 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
461
462typedef struct {
288b8c02 463 REGEXP *prog;
3b0527fe
DM
464 char *bol;
465 char *till;
466 SV *sv;
467 char *ganch;
24b23f37 468 char *cutpoint;
3b0527fe
DM
469} regmatch_info;
470
5d9a96ca
DM
471
472/* structures for holding and saving the state maintained by regmatch() */
473
d56b3014 474#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 475#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 476#endif
6bda09f9 477
5d9a96ca
DM
478typedef I32 CHECKPOINT;
479
a0374537 480typedef struct regmatch_state {
40a82448 481 int resume_state; /* where to jump to on return */
24d3c4a9 482 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
483
484 union {
77cb431f
DM
485
486 /* this is a fake union member that matches the first element
487 * of each member that needs to store positive backtrack
488 * information */
489 struct {
490 struct regmatch_state *prev_yes_state;
491 } yes;
492
fae667d5
YO
493 /* branchlike members */
494 /* this is a fake union member that matches the first elements
495 * of each member that needs to behave like a branch */
496 struct {
5d458dd8
YO
497 /* this first element must match u.yes */
498 struct regmatch_state *prev_yes_state;
fae667d5
YO
499 U32 lastparen;
500 CHECKPOINT cp;
501
502 } branchlike;
503
504 struct {
505 /* the first elements must match u.branchlike */
506 struct regmatch_state *prev_yes_state;
507 U32 lastparen;
508 CHECKPOINT cp;
509
510 regnode *next_branch; /* next branch node */
511 } branch;
512
513 struct {
514 /* the first elements must match u.branchlike */
515 struct regmatch_state *prev_yes_state;
516 U32 lastparen;
517 CHECKPOINT cp;
518
519 reg_trie_accepted *accept_buff; /* accepting states we have seen */
166ba7cd 520 U32 accepted; /* how many accepting states we have seen */
7f69552c 521 U16 *jump; /* positive offsets from me */
166ba7cd 522 regnode *B; /* node following the trie */
7f69552c 523 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
524 } trie;
525
fae667d5
YO
526 /* special types - these members are used to store state for special
527 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 528 struct {
77cb431f
DM
529 /* this first element must match u.yes */
530 struct regmatch_state *prev_yes_state;
faec1544
DM
531 struct regmatch_state *prev_eval;
532 struct regmatch_state *prev_curlyx;
288b8c02 533 REGEXP *prev_rex;
faec1544
DM
534 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
535 flip when transitioning between
536 inner and outer rexen */
aa283a38
DM
537 CHECKPOINT cp; /* remember current savestack indexes */
538 CHECKPOINT lastcp;
6bda09f9 539 U32 close_paren; /* which close bracket is our end */
1b5c067c 540 regnode *B; /* the node following us */
e822a8b4
DM
541 } eval;
542
543 struct {
c476f425 544 /* this first element must match u.yes */
262b90c4 545 struct regmatch_state *prev_yes_state;
fae667d5
YO
546 I32 wanted;
547 I32 logical; /* saved copy of 'logical' var */
548 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
549 } ifmatch; /* and SUSPEND/UNLESSM */
550
551 struct {
552 /* this first element must match u.yes */
553 struct regmatch_state *prev_yes_state;
554 struct regmatch_state *prev_mark;
555 SV* mark_name;
556 char *mark_loc;
557 } mark;
558
559 struct {
560 int val;
561 } keeper;
562
563 /* quantifiers - these members are used for storing state for
564 for the regops used to implement quantifiers */
565 struct {
566 /* this first element must match u.yes */
567 struct regmatch_state *prev_yes_state;
c476f425 568 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 569 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425
DM
570 CHECKPOINT cp; /* remember current savestack index */
571 bool minmod;
a0374537 572 int parenfloor;/* how far back to strip paren data */
c476f425
DM
573 int min; /* the minimal number of A's to match */
574 int max; /* the maximal number of A's to match */
c476f425
DM
575
576 /* these two are modified by WHILEM */
577 int count; /* how many instances of A we've matched */
578 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
579 } curlyx;
580
581 struct {
c476f425 582 /* this first element must match u.yes */
262b90c4 583 struct regmatch_state *prev_yes_state;
c476f425
DM
584 struct regmatch_state *save_curlyx;
585 CHECKPOINT cp; /* remember current savestack indexes */
586 CHECKPOINT lastcp;
587 char *save_lastloc; /* previous curlyx.lastloc */
588 I32 cache_offset;
589 I32 cache_mask;
e822a8b4
DM
590 } whilem;
591
592 struct {
5d458dd8
YO
593 /* this first element must match u.yes */
594 struct regmatch_state *prev_yes_state;
e822a8b4 595 I32 c1, c2; /* case fold search */
40a82448
DM
596 CHECKPOINT cp;
597 I32 alen; /* length of first-matched A string */
598 I32 count;
0cadcf80 599 bool minmod;
40a82448
DM
600 regnode *A, *B; /* the nodes corresponding to /A*B/ */
601 regnode *me; /* the curlym node */
e822a8b4
DM
602 } curlym;
603
604 struct {
3b6647e0 605 U32 paren;
c255a977 606 CHECKPOINT cp;
e822a8b4 607 I32 c1, c2; /* case fold search */
c255a977
DM
608 char *maxpos; /* highest possible point in string to match */
609 char *oldloc; /* the previous locinput */
e822a8b4 610 int count;
c255a977
DM
611 int min, max; /* {m,n} */
612 regnode *A, *B; /* the nodes corresponding to /A*B/ */
613 } curly; /* and CURLYN/PLUS/STAR */
dad79028 614
d8319b27 615 } u;
5d9a96ca
DM
616} regmatch_state;
617
618/* how many regmatch_state structs to allocate as a single slab.
619 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
620 * pointers, plus 1 for any mythical malloc overhead. */
621
622#define PERL_REGMATCH_SLAB_SLOTS \
623 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
624
625typedef struct regmatch_slab {
626 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
627 struct regmatch_slab *prev, *next;
628} regmatch_slab;
1ade1aa1 629
46ab3289
NC
630#define PL_reg_flags PL_reg_state.re_state_reg_flags
631#define PL_bostr PL_reg_state.re_state_bostr
632#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 633#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 634#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
635#define PL_reglastparen PL_reg_state.re_state_reglastparen
636#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
637#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
638#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
639#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
640#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
641#define PL_reg_magic PL_reg_state.re_state_reg_magic
642#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
643#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
644#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
645#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
646#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
647#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
648#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
649#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
650#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
651#define PL_regsize PL_reg_state.re_state_regsize
652#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
653#define PL_nrs PL_reg_state.re_state_nrs
654
1ade1aa1
NC
655struct re_save_state {
656 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
657 U32 re_state_reg_start_tmpl; /* from regexec.c */
658 I32 re_state_reg_eval_set; /* from regexec.c */
659 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
660 char *re_state_bostr;
661 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 662 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 663 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
664 U32 *re_state_reglastparen; /* Similarly for lastparen. */
665 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 666 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 667 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
668 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
669 PMOP *re_state_reg_curpm; /* from regexec.c */
670 char *re_state_reg_oldsaved; /* old saved substr during match */
671 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
672 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
673 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
674 I32 re_state_reg_maxiter; /* max wait until caching pos */
675 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 676 U32 re_state_regsize; /* from regexec.c */
1b5c067c 677 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
678 char *re_state_reg_starttry; /* from regexec.c */
679#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 680 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
681#endif
682};
683
684#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
685 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 686
1ade1aa1
NC
687/*
688 * Local variables:
689 * c-indentation-style: bsd
690 * c-basic-offset: 4
691 * indent-tabs-mode: t
692 * End:
693 *
694 * ex: set ts=8 sts=4 sw=4 noet:
695 */