This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Remove union _xivu from struct xpvhv - replace it with a non-union xav_keys.
[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? */ \
73 /* Information about the match that the perl core uses to */ \
74 /* manage things */ \
75 U32 extflags; /* Flags used both externally and internally */ \
76 I32 minlen; /* mininum possible length of string to match */\
77 I32 minlenret; /* mininum possible length of $& */ \
78 U32 gofs; /* chars left of pos that we search from */ \
79 /* substring data about strings that must appear in the */ \
80 /* final match, used for optimisations */ \
81 struct reg_substr_data *substrs; \
82 U32 nparens; /* number of capture buffers */ \
83 /* private engine specific data */ \
84 U32 intflags; /* Engine Specific Internal flags */ \
85 void *pprivate; /* Data private to the regex engine which */ \
86 /* created this object. */ \
87 /* Data about the last/current match. These are modified */ \
88 /* during matching */ \
89 U32 lastparen; /* last open paren matched */ \
90 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 91 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
92 /* Array of offsets for (@-) and (@+) */ \
93 regexp_paren_pair *offs; \
94 /* saved or original string so \digit works forever. */ \
95 char *subbeg; \
96 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
97 I32 sublen; /* Length of string pointed by subbeg */ \
98 /* Information about the match that isn't often used */ \
99 /* offset from wrapped to the start of precomp */ \
654eccd5 100 PERL_BITFIELD32 pre_prefix:4; \
08e44740 101 /* number of eval groups in the pattern - for security checks */\
654eccd5 102 PERL_BITFIELD32 seen_evals:28
08e44740 103
bbe252da 104typedef struct regexp {
288b8c02 105 _XPV_HEAD;
6e128786 106 union _xivu xiv_u;
08e44740 107 _REGEXP_COMMON;
f8fc2ecf
YO
108} regexp;
109
1af6200d
NC
110/* HV *paren_names; Optional hash of paren names
111 now stored in the IV union */
112
113#define RXp_PAREN_NAMES(rx) ((rx)->xiv_u.xivu_hv)
5daac39c 114
785a26d5 115/* used for high speed searches */
f9f4320a
YO
116typedef struct re_scream_pos_data_s
117{
118 char **scream_olds; /* match pos */
119 I32 *scream_pos; /* Internal iterator of scream. */
120} re_scream_pos_data;
121
785a26d5
YO
122/* regexp_engine structure. This is the dispatch table for regexes.
123 * Any regex engine implementation must be able to build one of these.
124 */
f9f4320a 125typedef struct regexp_engine {
1593ad57 126 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
127 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
128 char* strbeg, I32 minend, SV* screamer,
129 void* data, U32 flags);
130 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
131 char *strend, const U32 flags,
132 re_scream_pos_data *data);
49d7dfbc
AB
133 SV* (*checkstr) (pTHX_ REGEXP * const rx);
134 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 135 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 136 SV * const sv);
2fdbfb4d
AB
137 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
138 SV const * const value);
139 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
140 const I32 paren);
192b9cd1
AB
141 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
142 SV * const value, const U32 flags);
143 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
144 const U32 flags);
49d7dfbc 145 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 146#ifdef USE_ITHREADS
49d7dfbc 147 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 148#endif
f9f4320a
YO
149} regexp_engine;
150
192b9cd1
AB
151/*
152 These are passed to the numbered capture variable callbacks as the
153 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
154 $2 etc.
155*/
f1b875a0
YO
156#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
157#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
158#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
159
160/*
161 Flags that are passed to the named_buff and named_buff_iter
162 callbacks above. Those routines are called from universal.c via the
163 Tie::Hash::NamedCapture interface for %+ and %- and the re::
164 functions in the same file.
165*/
166
167/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
168#define RXapif_FETCH 0x0001
169#define RXapif_STORE 0x0002
170#define RXapif_DELETE 0x0004
171#define RXapif_CLEAR 0x0008
172#define RXapif_EXISTS 0x0010
173#define RXapif_SCALAR 0x0020
174#define RXapif_FIRSTKEY 0x0040
175#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
176
177/* Whether %+ or %- is being operated on */
f1b875a0
YO
178#define RXapif_ONE 0x0100 /* %+ */
179#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
180
181/* Whether this is being called from a re:: function */
f1b875a0
YO
182#define RXapif_REGNAME 0x0400
183#define RXapif_REGNAMES 0x0800
7948fc08 184#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 185
f7e71195
AB
186/*
187=head1 REGEXP Functions
188
189=for apidoc Am|REGEXP *|SvRX|SV *sv
190
191Convenience macro to get the REGEXP from a SV. This is approximately
192equivalent to the following snippet:
193
194 if (SvMAGICAL(sv))
195 mg_get(sv);
196 if (SvROK(sv) &&
197 (tmpsv = (SV*)SvRV(sv)) &&
198 SvTYPE(tmpsv) == SVt_PVMG &&
199 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
200 {
201 return (REGEXP *)tmpmg->mg_obj;
202 }
203
204NULL will be returned if a REGEXP* is not found.
205
206=for apidoc Am|bool|SvRXOK|SV* sv
207
208Returns a boolean indicating whether the SV contains qr magic
209(PERL_MAGIC_qr).
210
211If you want to do something with the REGEXP* later use SvRX instead
212and check for NULL.
213
214=cut
215*/
216
217#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
218#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
219
220
7948fc08 221/* Flags stored in regexp->extflags
bbe252da 222 * These are used by code external to the regexp engine
e357fc67
YO
223 *
224 * Note that flags starting with RXf_PMf_ have exact equivalents
225 * stored in op_pmflags and which are defined in op.h, they are defined
226 * numerically here only for clarity.
cb5027f2
YO
227 *
228 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 229 * so that regnodes.h is updated with the changes.
cb5027f2 230 *
bbe252da
YO
231 */
232
52d81aa8 233/* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
a1cf41cd
NC
234 * If you change these you need to change the equivalent flags in op.h, and
235 * vice versa. */
52d81aa8
NC
236#define RXf_PMf_MULTILINE 0x00000001 /* /m */
237#define RXf_PMf_SINGLELINE 0x00000002 /* /s */
238#define RXf_PMf_FOLD 0x00000004 /* /i */
239#define RXf_PMf_EXTENDED 0x00000008 /* /x */
240#define RXf_PMf_KEEPCOPY 0x00000010 /* /p */
241#define RXf_PMf_LOCALE 0x00000020 /* use locale */
e357fc67 242/* these flags are transfered from the PMOP->op_pmflags member during compilation */
52d81aa8 243#define RXf_PMf_STD_PMMOD_SHIFT 0
cde0cee5
YO
244#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
245#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
246
a20207d7
YO
247#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
248 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
249 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
250 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
251 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 252
bcdf7404
YO
253/* chars and strings used as regex pattern modifiers
254 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
255 *
256 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
257 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
258 * for its own uses. So 'k' is out as well.
bcdf7404
YO
259 */
260#define EXEC_PAT_MOD 'e'
87e95b7f 261#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
262#define ONCE_PAT_MOD 'o'
263#define GLOBAL_PAT_MOD 'g'
264#define CONTINUE_PAT_MOD 'c'
265#define MULTILINE_PAT_MOD 'm'
266#define SINGLE_PAT_MOD 's'
267#define IGNORE_PAT_MOD 'i'
268#define XTENDED_PAT_MOD 'x'
269
270#define ONCE_PAT_MODS "o"
87e95b7f 271#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
272#define EXEC_PAT_MODS "e"
273#define LOOP_PAT_MODS "gc"
274
275#define STD_PAT_MODS "msix"
276
277#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
278
279#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
280#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
281#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
282#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
283
cb5027f2
YO
284/*
285 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 286 * so that regnodes.h is updated with the changes.
cb5027f2
YO
287 *
288 */
bcdf7404 289
52d81aa8
NC
290/* Anchor and GPOS related stuff */
291#define RXf_ANCH_BOL 0x00000100
292#define RXf_ANCH_MBOL 0x00000200
293#define RXf_ANCH_SBOL 0x00000400
294#define RXf_ANCH_GPOS 0x00000800
295#define RXf_GPOS_SEEN 0x00001000
296#define RXf_GPOS_FLOAT 0x00002000
297/* two bits here */
298#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
299#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
300#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
301
bbe252da 302/* What we have seen */
52d81aa8
NC
303#define RXf_LOOKBEHIND_SEEN 0x00004000
304#define RXf_EVAL_SEEN 0x00008000
305#define RXf_CANY_SEEN 0x00010000
bbe252da
YO
306
307/* Special */
52d81aa8
NC
308#define RXf_NOSCAN 0x00020000
309#define RXf_CHECK_ALL 0x00040000
bbe252da
YO
310
311/* UTF8 related */
52d81aa8 312#define RXf_MATCH_UTF8 0x00100000
bbe252da
YO
313
314/* Intuit related */
52d81aa8
NC
315#define RXf_USE_INTUIT_NOML 0x00200000
316#define RXf_USE_INTUIT_ML 0x00400000
317#define RXf_INTUIT_TAIL 0x00800000
0ac6acae
AB
318
319/*
320 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
321 be used by regex engines to check whether they should set
322 RXf_SKIPWHITE
323*/
52d81aa8 324#define RXf_SPLIT 0x01000000
0ac6acae 325
bbe252da
YO
326#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
327
328/* Copy and tainted info */
52d81aa8
NC
329#define RXf_COPY_DONE 0x02000000
330#define RXf_TAINTED_SEEN 0x04000000
331#define RXf_TAINTED 0x08000000 /* this pattern is tainted */
332
333/* Flags indicating special patterns */
334#define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */
335#define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */
336#define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */
337#define RXf_NULL 0x80000000 /* Pattern is // */
c737faaf 338
cb5027f2
YO
339/*
340 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 341 * so that regnodes.h is updated with the changes.
cb5027f2
YO
342 *
343 */
bbe252da 344
bbe252da 345#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
346#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
347#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
348#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
349#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
350#define RX_MATCH_TAINTED_set(prog, t) ((t) \
351 ? RX_MATCH_TAINTED_on(prog) \
352 : RX_MATCH_TAINTED_off(prog))
c277df42 353
07bc277f
NC
354#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
355#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
356#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
357#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
358#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
359#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
360#define RX_MATCH_COPIED_set(prog,t) ((t) \
361 ? RX_MATCH_COPIED_on(prog) \
362 : RX_MATCH_COPIED_off(prog))
f9f4320a 363
07bc277f 364#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 365
07bc277f 366/* For source compatibility. We used to store these explicitly. */
f7c278bf 367#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 368#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
369/* FIXME? Are we hardcoding too much here and constraining plugin extension
370 writers? Specifically, the value 1 assumes that the wrapped version always
371 has exactly one character at the end, a ')'. Will that always be true? */
372#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 373#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 374#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 375#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 376#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 377#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 378#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
379# define RX_EXTFLAGS(prog) \
380 (*({ \
7948fc08
RU
381 const REGEXP *const _rx_extflags = (prog); \
382 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
383 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
384 }))
385# define RX_ENGINE(prog) \
386 (*({ \
7948fc08
RU
387 const REGEXP *const _rx_engine = (prog); \
388 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
389 &SvANY(_rx_engine)->engine; \
4450b630
NC
390 }))
391# define RX_SUBBEG(prog) \
392 (*({ \
7948fc08
RU
393 const REGEXP *const _rx_subbeg = (prog); \
394 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
395 &SvANY(_rx_subbeg)->subbeg; \
4450b630 396 }))
200ff815 397# define RX_OFFS(prog) \
4450b630 398 (*({ \
7948fc08
RU
399 const REGEXP *const _rx_offs = (prog); \
400 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
401 &SvANY(_rx_offs)->offs; \
4450b630
NC
402 }))
403# define RX_NPARENS(prog) \
404 (*({ \
7948fc08
RU
405 const REGEXP *const _rx_nparens = (prog); \
406 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
407 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
408 }))
409#else
410# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
411# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
412# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
413# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
414# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
415#endif
288b8c02 416#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
417#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
418#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
419#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
420#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
421#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
422#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 423#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 424
be8e71aa
YO
425#endif /* PLUGGABLE_RE_EXTENSION */
426
427/* Stuff that needs to be included in the plugable extension goes below here */
428
f8c7b90f 429#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 430#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
431 STMT_START {if (RX_SAVED_COPY(rx)) { \
432 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
433 } \
434 if (RX_MATCH_COPIED(rx)) { \
07bc277f 435 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
436 RX_MATCH_COPIED_off(rx); \
437 }} STMT_END
438#else
439#define RX_MATCH_COPY_FREE(rx) \
440 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 441 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
442 RX_MATCH_COPIED_off(rx); \
443 }} STMT_END
444#endif
445
07bc277f
NC
446#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
447#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
448#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
449#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
450#define RX_MATCH_UTF8_set(prog, t) ((t) \
451 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
452 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
453
454/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 455#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 456
cad2e5aa
JH
457#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
458#define REXEC_CHECKED 0x02 /* check_substr already checked. */
459#define REXEC_SCREAM 0x04 /* use scream table. */
460#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 461#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 462
7fe44985 463#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
464# define ReREFCNT_inc(re) \
465 ({ \
466 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
467 REGEXP *const _rerefcnt_inc = (re); \
468 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
469 SvREFCNT_inc(_rerefcnt_inc); \
470 _rerefcnt_inc; \
288b8c02
NC
471 })
472# define ReREFCNT_dec(re) \
473 ({ \
474 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
475 REGEXP *const _rerefcnt_dec = (re); \
476 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
477 })
478#else
479# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 480# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
481#endif
482
483/* FIXME for plugins. */
cf93c79d
IZ
484
485#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
486#define FBMcf_TAIL_DOLLARM 2
487#define FBMcf_TAIL_Z 4
488#define FBMcf_TAIL_z 8
489#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
490
491#define FBMrf_MULTILINE 1
cad2e5aa 492
3b0527fe
DM
493/* some basic information about the current match that is created by
494 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
495
496typedef struct {
288b8c02 497 REGEXP *prog;
3b0527fe
DM
498 char *bol;
499 char *till;
500 SV *sv;
501 char *ganch;
24b23f37 502 char *cutpoint;
3b0527fe
DM
503} regmatch_info;
504
5d9a96ca
DM
505
506/* structures for holding and saving the state maintained by regmatch() */
507
d56b3014 508#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 509#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 510#endif
6bda09f9 511
5d9a96ca
DM
512typedef I32 CHECKPOINT;
513
a0374537 514typedef struct regmatch_state {
40a82448 515 int resume_state; /* where to jump to on return */
24d3c4a9 516 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
517
518 union {
77cb431f
DM
519
520 /* this is a fake union member that matches the first element
521 * of each member that needs to store positive backtrack
522 * information */
523 struct {
524 struct regmatch_state *prev_yes_state;
525 } yes;
526
fae667d5
YO
527 /* branchlike members */
528 /* this is a fake union member that matches the first elements
529 * of each member that needs to behave like a branch */
530 struct {
5d458dd8
YO
531 /* this first element must match u.yes */
532 struct regmatch_state *prev_yes_state;
fae667d5
YO
533 U32 lastparen;
534 CHECKPOINT cp;
535
536 } branchlike;
537
538 struct {
539 /* the first elements must match u.branchlike */
540 struct regmatch_state *prev_yes_state;
541 U32 lastparen;
542 CHECKPOINT cp;
543
544 regnode *next_branch; /* next branch node */
545 } branch;
546
547 struct {
548 /* the first elements must match u.branchlike */
549 struct regmatch_state *prev_yes_state;
550 U32 lastparen;
551 CHECKPOINT cp;
552
2e64971a 553 U32 accepted; /* how many accepting states left */
7f69552c 554 U16 *jump; /* positive offsets from me */
166ba7cd 555 regnode *B; /* node following the trie */
7f69552c 556 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
557 U8 *firstpos;/* pos in string of first trie match */
558 U32 firstchars;/* len in chars of firstpos from start */
559 U16 nextword;/* next word to try */
560 U16 topword; /* longest accepted word */
561 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
562 } trie;
563
fae667d5
YO
564 /* special types - these members are used to store state for special
565 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 566 struct {
77cb431f
DM
567 /* this first element must match u.yes */
568 struct regmatch_state *prev_yes_state;
faec1544
DM
569 struct regmatch_state *prev_eval;
570 struct regmatch_state *prev_curlyx;
288b8c02 571 REGEXP *prev_rex;
faec1544
DM
572 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
573 flip when transitioning between
574 inner and outer rexen */
aa283a38
DM
575 CHECKPOINT cp; /* remember current savestack indexes */
576 CHECKPOINT lastcp;
6bda09f9 577 U32 close_paren; /* which close bracket is our end */
1b5c067c 578 regnode *B; /* the node following us */
e822a8b4
DM
579 } eval;
580
581 struct {
c476f425 582 /* this first element must match u.yes */
262b90c4 583 struct regmatch_state *prev_yes_state;
fae667d5
YO
584 I32 wanted;
585 I32 logical; /* saved copy of 'logical' var */
586 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
587 } ifmatch; /* and SUSPEND/UNLESSM */
588
589 struct {
590 /* this first element must match u.yes */
591 struct regmatch_state *prev_yes_state;
592 struct regmatch_state *prev_mark;
593 SV* mark_name;
594 char *mark_loc;
595 } mark;
596
597 struct {
598 int val;
599 } keeper;
600
601 /* quantifiers - these members are used for storing state for
602 for the regops used to implement quantifiers */
603 struct {
604 /* this first element must match u.yes */
605 struct regmatch_state *prev_yes_state;
c476f425 606 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 607 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425
DM
608 CHECKPOINT cp; /* remember current savestack index */
609 bool minmod;
a0374537 610 int parenfloor;/* how far back to strip paren data */
c476f425
DM
611 int min; /* the minimal number of A's to match */
612 int max; /* the maximal number of A's to match */
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 */