This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexp.h: Fix error check to use correct offset
[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
1850c8f9
KW
11#include "op_reg_common.h"
12
378cc40b
LW
13/*
14 * Definitions etc. for regexp(3) routines.
15 *
16 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
17 * not the System V one.
18 */
be8e71aa 19#ifndef PLUGGABLE_RE_EXTENSION
785a26d5
YO
20/* we don't want to include this stuff if we are inside of
21 an external regex engine based on the core one - like re 'debug'*/
378cc40b 22
c277df42
IZ
23struct regnode {
24 U8 flags;
25 U8 type;
26 U16 next_off;
27};
28
29typedef struct regnode regnode;
30
cad2e5aa 31struct reg_substr_data;
2779dcf1 32
0ee3c8d0
JH
33struct reg_data;
34
f9f4320a 35struct regexp_engine;
28d8d7f4 36struct regexp;
bbe252da 37
785a26d5
YO
38struct reg_substr_datum {
39 I32 min_offset;
40 I32 max_offset;
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
43 I32 end_shift;
44};
45struct reg_substr_data {
46 struct reg_substr_datum data[3]; /* Actual array */
47};
f9f4320a 48
f8c7b90f 49#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
50#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
51#else
52#define SV_SAVED_COPY
ed252734 53#endif
28d8d7f4 54
f0ab9afb
NC
55typedef struct regexp_paren_pair {
56 I32 start;
57 I32 end;
58} regexp_paren_pair;
28d8d7f4 59
882227b7
AB
60/*
61 The regexp/REGEXP struct, see L<perlreapi> for further documentation
62 on the individual fields. The struct is ordered so that the most
63 commonly used fields are placed at the start.
64
65 Any patch that adds items to this struct will need to include
66 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
67 (C<pregfree()>). This involves freeing or cloning items in the
68 regexp's data array based on the data item's type.
69*/
70
08e44740
NC
71#define _REGEXP_COMMON \
72 /* what engine created this regexp? */ \
73 const struct regexp_engine* engine; \
74 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 75 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
76 /* Information about the match that the perl core uses to */ \
77 /* manage things */ \
78 U32 extflags; /* Flags used both externally and internally */ \
79 I32 minlen; /* mininum possible length of string to match */\
80 I32 minlenret; /* mininum possible length of $& */ \
81 U32 gofs; /* chars left of pos that we search from */ \
82 /* substring data about strings that must appear in the */ \
83 /* final match, used for optimisations */ \
84 struct reg_substr_data *substrs; \
85 U32 nparens; /* number of capture buffers */ \
86 /* private engine specific data */ \
87 U32 intflags; /* Engine Specific Internal flags */ \
88 void *pprivate; /* Data private to the regex engine which */ \
89 /* created this object. */ \
90 /* Data about the last/current match. These are modified */ \
91 /* during matching */ \
92 U32 lastparen; /* last open paren matched */ \
93 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 94 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
95 /* Array of offsets for (@-) and (@+) */ \
96 regexp_paren_pair *offs; \
97 /* saved or original string so \digit works forever. */ \
98 char *subbeg; \
99 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
100 I32 sublen; /* Length of string pointed by subbeg */ \
101 /* Information about the match that isn't often used */ \
102 /* offset from wrapped to the start of precomp */ \
654eccd5 103 PERL_BITFIELD32 pre_prefix:4; \
08e44740 104 /* number of eval groups in the pattern - for security checks */\
654eccd5 105 PERL_BITFIELD32 seen_evals:28
08e44740 106
bbe252da 107typedef struct regexp {
288b8c02 108 _XPV_HEAD;
08e44740 109 _REGEXP_COMMON;
f8fc2ecf
YO
110} regexp;
111
89f6f287 112#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 113
785a26d5 114/* used for high speed searches */
f9f4320a
YO
115typedef struct re_scream_pos_data_s
116{
117 char **scream_olds; /* match pos */
118 I32 *scream_pos; /* Internal iterator of scream. */
119} re_scream_pos_data;
120
785a26d5
YO
121/* regexp_engine structure. This is the dispatch table for regexes.
122 * Any regex engine implementation must be able to build one of these.
123 */
f9f4320a 124typedef struct regexp_engine {
1593ad57 125 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
126 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
127 char* strbeg, I32 minend, SV* screamer,
128 void* data, U32 flags);
129 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
130 char *strend, const U32 flags,
131 re_scream_pos_data *data);
49d7dfbc
AB
132 SV* (*checkstr) (pTHX_ REGEXP * const rx);
133 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 134 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 135 SV * const sv);
2fdbfb4d
AB
136 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
137 SV const * const value);
138 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
139 const I32 paren);
192b9cd1
AB
140 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
141 SV * const value, const U32 flags);
142 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
143 const U32 flags);
49d7dfbc 144 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 145#ifdef USE_ITHREADS
49d7dfbc 146 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 147#endif
f9f4320a
YO
148} regexp_engine;
149
192b9cd1
AB
150/*
151 These are passed to the numbered capture variable callbacks as the
152 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
153 $2 etc.
154*/
f1b875a0
YO
155#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
156#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
157#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
158
159/*
160 Flags that are passed to the named_buff and named_buff_iter
161 callbacks above. Those routines are called from universal.c via the
162 Tie::Hash::NamedCapture interface for %+ and %- and the re::
163 functions in the same file.
164*/
165
166/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
167#define RXapif_FETCH 0x0001
168#define RXapif_STORE 0x0002
169#define RXapif_DELETE 0x0004
170#define RXapif_CLEAR 0x0008
171#define RXapif_EXISTS 0x0010
172#define RXapif_SCALAR 0x0020
173#define RXapif_FIRSTKEY 0x0040
174#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
175
176/* Whether %+ or %- is being operated on */
f1b875a0
YO
177#define RXapif_ONE 0x0100 /* %+ */
178#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
179
180/* Whether this is being called from a re:: function */
f1b875a0
YO
181#define RXapif_REGNAME 0x0400
182#define RXapif_REGNAMES 0x0800
7948fc08 183#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 184
f7e71195
AB
185/*
186=head1 REGEXP Functions
187
188=for apidoc Am|REGEXP *|SvRX|SV *sv
189
190Convenience macro to get the REGEXP from a SV. This is approximately
191equivalent to the following snippet:
192
193 if (SvMAGICAL(sv))
194 mg_get(sv);
195 if (SvROK(sv) &&
196 (tmpsv = (SV*)SvRV(sv)) &&
197 SvTYPE(tmpsv) == SVt_PVMG &&
198 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
199 {
200 return (REGEXP *)tmpmg->mg_obj;
201 }
202
203NULL will be returned if a REGEXP* is not found.
204
205=for apidoc Am|bool|SvRXOK|SV* sv
206
207Returns a boolean indicating whether the SV contains qr magic
208(PERL_MAGIC_qr).
209
210If you want to do something with the REGEXP* later use SvRX instead
211and check for NULL.
212
213=cut
214*/
215
216#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
217#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
218
219
7948fc08 220/* Flags stored in regexp->extflags
bbe252da 221 * These are used by code external to the regexp engine
e357fc67
YO
222 *
223 * Note that flags starting with RXf_PMf_ have exact equivalents
224 * stored in op_pmflags and which are defined in op.h, they are defined
225 * numerically here only for clarity.
cb5027f2
YO
226 *
227 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 228 * so that regnodes.h is updated with the changes.
cb5027f2 229 *
bbe252da
YO
230 */
231
52d81aa8 232/* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
a1cf41cd 233 * If you change these you need to change the equivalent flags in op.h, and
1850c8f9 234 * vice versa. These need to be ordered so that the msix are contiguous
78fecfc8
KW
235 * starting at bit 0, followed by the p; bit 0 is because of the shift below
236 * being 0; see STD_PAT_MODS and INT_PAT_MODS below for the contiguity cause */
237/* the flags above are transfered from the PMOP->op_pmflags member during
238 * compilation */
52d81aa8 239#define RXf_PMf_STD_PMMOD_SHIFT 0
cde0cee5
YO
240#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
241#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
242
a20207d7
YO
243#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
244 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
245 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
246 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
247 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 248
bcdf7404
YO
249/* chars and strings used as regex pattern modifiers
250 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
251 *
252 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
253 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
254 * for its own uses. So 'k' is out as well.
bcdf7404
YO
255 */
256#define EXEC_PAT_MOD 'e'
87e95b7f 257#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
258#define ONCE_PAT_MOD 'o'
259#define GLOBAL_PAT_MOD 'g'
260#define CONTINUE_PAT_MOD 'c'
261#define MULTILINE_PAT_MOD 'm'
262#define SINGLE_PAT_MOD 's'
263#define IGNORE_PAT_MOD 'i'
264#define XTENDED_PAT_MOD 'x'
4f4d7508 265#define NONDESTRUCT_PAT_MOD 'r'
bcdf7404
YO
266
267#define ONCE_PAT_MODS "o"
87e95b7f 268#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
269#define EXEC_PAT_MODS "e"
270#define LOOP_PAT_MODS "gc"
4f4d7508 271#define NONDESTRUCT_PAT_MODS "r"
bcdf7404 272
78fecfc8
KW
273/* This string is expected by regcomp.c to be ordered so that the first
274 * character is the flag in bit 0 of extflags; the next character is bit 1,
275 * etc. */
bcdf7404
YO
276#define STD_PAT_MODS "msix"
277
78fecfc8
KW
278/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
279 * so that the first character is the flag in bit 0 of extflags; the next
280 * character is bit 1, etc. */
bcdf7404
YO
281#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
282
283#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
284#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
285#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 286#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 287
cb5027f2
YO
288/*
289 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 290 * so that regnodes.h is updated with the changes.
cb5027f2
YO
291 *
292 */
bcdf7404 293
52d81aa8 294/* Anchor and GPOS related stuff */
1850c8f9
KW
295#define RXf_ANCH_BOL (1<<(_RXf_PMf_SHIFT+3))
296#define RXf_ANCH_MBOL (1<<(_RXf_PMf_SHIFT+4))
297#define RXf_ANCH_SBOL (1<<(_RXf_PMf_SHIFT+5))
298#define RXf_ANCH_GPOS (1<<(_RXf_PMf_SHIFT+6))
299#define RXf_GPOS_SEEN (1<<(_RXf_PMf_SHIFT+7))
300#define RXf_GPOS_FLOAT (1<<(_RXf_PMf_SHIFT+8))
52d81aa8
NC
301/* two bits here */
302#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
303#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
304#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
305
bbe252da 306/* What we have seen */
1850c8f9
KW
307#define RXf_LOOKBEHIND_SEEN (1<<(_RXf_PMf_SHIFT+9))
308#define RXf_EVAL_SEEN (1<<(_RXf_PMf_SHIFT+10))
309#define RXf_CANY_SEEN (1<<(_RXf_PMf_SHIFT+11))
bbe252da
YO
310
311/* Special */
1850c8f9
KW
312#define RXf_NOSCAN (1<<(_RXf_PMf_SHIFT+12))
313#define RXf_CHECK_ALL (1<<(_RXf_PMf_SHIFT+13))
bbe252da
YO
314
315/* UTF8 related */
1850c8f9 316#define RXf_MATCH_UTF8 (1<<(_RXf_PMf_SHIFT+15))
bbe252da
YO
317
318/* Intuit related */
1850c8f9
KW
319#define RXf_USE_INTUIT_NOML (1<<(_RXf_PMf_SHIFT+16))
320#define RXf_USE_INTUIT_ML (1<<(_RXf_PMf_SHIFT+17))
321#define RXf_INTUIT_TAIL (1<<(_RXf_PMf_SHIFT+18))
0ac6acae
AB
322
323/*
324 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
325 be used by regex engines to check whether they should set
326 RXf_SKIPWHITE
327*/
1850c8f9 328#define RXf_SPLIT (1<<(_RXf_PMf_SHIFT+19))
0ac6acae 329
bbe252da
YO
330#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
331
332/* Copy and tainted info */
1850c8f9
KW
333#define RXf_COPY_DONE (1<<(_RXf_PMf_SHIFT+20))
334#define RXf_TAINTED_SEEN (1<<(_RXf_PMf_SHIFT+21))
335#define RXf_TAINTED (1<<(_RXf_PMf_SHIFT+22)) /* this pattern is tainted */
52d81aa8
NC
336
337/* Flags indicating special patterns */
1850c8f9
KW
338#define RXf_START_ONLY (1<<(_RXf_PMf_SHIFT+23)) /* Pattern is /^/ */
339#define RXf_SKIPWHITE (1<<(_RXf_PMf_SHIFT+24)) /* Pattern is for a split / / */
340#define RXf_WHITE (1<<(_RXf_PMf_SHIFT+25)) /* Pattern is /\s+/ */
341#define RXf_NULL (1<<(_RXf_PMf_SHIFT+26)) /* Pattern is // */
930c12f8 342#if _RXf_PMf_SHIFT+26 > 31
1850c8f9
KW
343# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
344#endif
c737faaf 345
cb5027f2
YO
346/*
347 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 348 * so that regnodes.h is updated with the changes.
cb5027f2
YO
349 *
350 */
bbe252da 351
bbe252da 352#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
353#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
354#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
355#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
356#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
357#define RX_MATCH_TAINTED_set(prog, t) ((t) \
358 ? RX_MATCH_TAINTED_on(prog) \
359 : RX_MATCH_TAINTED_off(prog))
c277df42 360
07bc277f
NC
361#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
362#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
363#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
364#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
365#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
366#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
367#define RX_MATCH_COPIED_set(prog,t) ((t) \
368 ? RX_MATCH_COPIED_on(prog) \
369 : RX_MATCH_COPIED_off(prog))
f9f4320a 370
07bc277f 371#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 372
07bc277f 373/* For source compatibility. We used to store these explicitly. */
f7c278bf 374#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 375#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
376/* FIXME? Are we hardcoding too much here and constraining plugin extension
377 writers? Specifically, the value 1 assumes that the wrapped version always
378 has exactly one character at the end, a ')'. Will that always be true? */
379#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 380#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 381#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 382#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 383#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 384#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 385#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
386# define RX_EXTFLAGS(prog) \
387 (*({ \
7948fc08
RU
388 const REGEXP *const _rx_extflags = (prog); \
389 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
390 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
391 }))
392# define RX_ENGINE(prog) \
393 (*({ \
7948fc08
RU
394 const REGEXP *const _rx_engine = (prog); \
395 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
396 &SvANY(_rx_engine)->engine; \
4450b630
NC
397 }))
398# define RX_SUBBEG(prog) \
399 (*({ \
7948fc08
RU
400 const REGEXP *const _rx_subbeg = (prog); \
401 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
402 &SvANY(_rx_subbeg)->subbeg; \
4450b630 403 }))
200ff815 404# define RX_OFFS(prog) \
4450b630 405 (*({ \
7948fc08
RU
406 const REGEXP *const _rx_offs = (prog); \
407 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
408 &SvANY(_rx_offs)->offs; \
4450b630
NC
409 }))
410# define RX_NPARENS(prog) \
411 (*({ \
7948fc08
RU
412 const REGEXP *const _rx_nparens = (prog); \
413 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
414 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
415 }))
416#else
417# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
418# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
419# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
420# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
421# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
422#endif
288b8c02 423#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
424#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
425#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
426#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
427#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
428#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
429#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 430#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 431
be8e71aa
YO
432#endif /* PLUGGABLE_RE_EXTENSION */
433
434/* Stuff that needs to be included in the plugable extension goes below here */
435
f8c7b90f 436#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 437#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
438 STMT_START {if (RX_SAVED_COPY(rx)) { \
439 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
440 } \
441 if (RX_MATCH_COPIED(rx)) { \
07bc277f 442 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
443 RX_MATCH_COPIED_off(rx); \
444 }} STMT_END
445#else
446#define RX_MATCH_COPY_FREE(rx) \
447 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 448 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
449 RX_MATCH_COPIED_off(rx); \
450 }} STMT_END
451#endif
452
07bc277f
NC
453#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
454#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
455#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
456#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
457#define RX_MATCH_UTF8_set(prog, t) ((t) \
458 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
459 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
460
461/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 462#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 463
cad2e5aa
JH
464#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
465#define REXEC_CHECKED 0x02 /* check_substr already checked. */
466#define REXEC_SCREAM 0x04 /* use scream table. */
467#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 468#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 469
7fe44985 470#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
471# define ReREFCNT_inc(re) \
472 ({ \
473 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
474 REGEXP *const _rerefcnt_inc = (re); \
475 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
476 SvREFCNT_inc(_rerefcnt_inc); \
477 _rerefcnt_inc; \
288b8c02
NC
478 })
479# define ReREFCNT_dec(re) \
480 ({ \
481 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
482 REGEXP *const _rerefcnt_dec = (re); \
483 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
484 })
485#else
486# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 487# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
488#endif
489
490/* FIXME for plugins. */
cf93c79d
IZ
491
492#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
493#define FBMcf_TAIL_DOLLARM 2
494#define FBMcf_TAIL_Z 4
495#define FBMcf_TAIL_z 8
496#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
497
498#define FBMrf_MULTILINE 1
cad2e5aa 499
3b0527fe
DM
500/* some basic information about the current match that is created by
501 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
502
503typedef struct {
288b8c02 504 REGEXP *prog;
3b0527fe
DM
505 char *bol;
506 char *till;
507 SV *sv;
508 char *ganch;
24b23f37 509 char *cutpoint;
3b0527fe
DM
510} regmatch_info;
511
5d9a96ca
DM
512
513/* structures for holding and saving the state maintained by regmatch() */
514
d56b3014 515#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 516#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 517#endif
6bda09f9 518
5d9a96ca
DM
519typedef I32 CHECKPOINT;
520
a0374537 521typedef struct regmatch_state {
40a82448 522 int resume_state; /* where to jump to on return */
24d3c4a9 523 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
524
525 union {
77cb431f
DM
526
527 /* this is a fake union member that matches the first element
528 * of each member that needs to store positive backtrack
529 * information */
530 struct {
531 struct regmatch_state *prev_yes_state;
532 } yes;
533
fae667d5
YO
534 /* branchlike members */
535 /* this is a fake union member that matches the first elements
536 * of each member that needs to behave like a branch */
537 struct {
5d458dd8
YO
538 /* this first element must match u.yes */
539 struct regmatch_state *prev_yes_state;
fae667d5
YO
540 U32 lastparen;
541 CHECKPOINT cp;
542
543 } branchlike;
544
545 struct {
546 /* the first elements must match u.branchlike */
547 struct regmatch_state *prev_yes_state;
548 U32 lastparen;
549 CHECKPOINT cp;
550
551 regnode *next_branch; /* next branch node */
552 } branch;
553
554 struct {
555 /* the first elements must match u.branchlike */
556 struct regmatch_state *prev_yes_state;
557 U32 lastparen;
558 CHECKPOINT cp;
559
2e64971a 560 U32 accepted; /* how many accepting states left */
7f69552c 561 U16 *jump; /* positive offsets from me */
166ba7cd 562 regnode *B; /* node following the trie */
7f69552c 563 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
564 U8 *firstpos;/* pos in string of first trie match */
565 U32 firstchars;/* len in chars of firstpos from start */
566 U16 nextword;/* next word to try */
567 U16 topword; /* longest accepted word */
568 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
569 } trie;
570
fae667d5
YO
571 /* special types - these members are used to store state for special
572 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 573 struct {
77cb431f
DM
574 /* this first element must match u.yes */
575 struct regmatch_state *prev_yes_state;
faec1544
DM
576 struct regmatch_state *prev_eval;
577 struct regmatch_state *prev_curlyx;
288b8c02 578 REGEXP *prev_rex;
faec1544
DM
579 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
580 flip when transitioning between
581 inner and outer rexen */
aa283a38
DM
582 CHECKPOINT cp; /* remember current savestack indexes */
583 CHECKPOINT lastcp;
6bda09f9 584 U32 close_paren; /* which close bracket is our end */
1b5c067c 585 regnode *B; /* the node following us */
e822a8b4
DM
586 } eval;
587
588 struct {
c476f425 589 /* this first element must match u.yes */
262b90c4 590 struct regmatch_state *prev_yes_state;
fae667d5
YO
591 I32 wanted;
592 I32 logical; /* saved copy of 'logical' var */
593 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
594 } ifmatch; /* and SUSPEND/UNLESSM */
595
596 struct {
597 /* this first element must match u.yes */
598 struct regmatch_state *prev_yes_state;
599 struct regmatch_state *prev_mark;
600 SV* mark_name;
601 char *mark_loc;
602 } mark;
603
604 struct {
605 int val;
606 } keeper;
607
608 /* quantifiers - these members are used for storing state for
609 for the regops used to implement quantifiers */
610 struct {
611 /* this first element must match u.yes */
612 struct regmatch_state *prev_yes_state;
c476f425 613 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
614 regnode *me; /* the CURLYX node */
615 regnode *B; /* the B node in /A*B/ */
c476f425
DM
616 CHECKPOINT cp; /* remember current savestack index */
617 bool minmod;
a0374537 618 int parenfloor;/* how far back to strip paren data */
c476f425
DM
619
620 /* these two are modified by WHILEM */
621 int count; /* how many instances of A we've matched */
622 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
623 } curlyx;
624
625 struct {
c476f425 626 /* this first element must match u.yes */
262b90c4 627 struct regmatch_state *prev_yes_state;
c476f425
DM
628 struct regmatch_state *save_curlyx;
629 CHECKPOINT cp; /* remember current savestack indexes */
630 CHECKPOINT lastcp;
631 char *save_lastloc; /* previous curlyx.lastloc */
632 I32 cache_offset;
633 I32 cache_mask;
e822a8b4
DM
634 } whilem;
635
636 struct {
5d458dd8
YO
637 /* this first element must match u.yes */
638 struct regmatch_state *prev_yes_state;
e822a8b4 639 I32 c1, c2; /* case fold search */
40a82448
DM
640 CHECKPOINT cp;
641 I32 alen; /* length of first-matched A string */
642 I32 count;
0cadcf80 643 bool minmod;
40a82448
DM
644 regnode *A, *B; /* the nodes corresponding to /A*B/ */
645 regnode *me; /* the curlym node */
e822a8b4
DM
646 } curlym;
647
648 struct {
3b6647e0 649 U32 paren;
c255a977 650 CHECKPOINT cp;
e822a8b4 651 I32 c1, c2; /* case fold search */
c255a977
DM
652 char *maxpos; /* highest possible point in string to match */
653 char *oldloc; /* the previous locinput */
e822a8b4 654 int count;
c255a977
DM
655 int min, max; /* {m,n} */
656 regnode *A, *B; /* the nodes corresponding to /A*B/ */
657 } curly; /* and CURLYN/PLUS/STAR */
dad79028 658
d8319b27 659 } u;
5d9a96ca
DM
660} regmatch_state;
661
662/* how many regmatch_state structs to allocate as a single slab.
663 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
664 * pointers, plus 1 for any mythical malloc overhead. */
665
666#define PERL_REGMATCH_SLAB_SLOTS \
667 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
668
669typedef struct regmatch_slab {
670 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
671 struct regmatch_slab *prev, *next;
672} regmatch_slab;
1ade1aa1 673
46ab3289
NC
674#define PL_reg_flags PL_reg_state.re_state_reg_flags
675#define PL_bostr PL_reg_state.re_state_bostr
676#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 677#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 678#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
679#define PL_reglastparen PL_reg_state.re_state_reglastparen
680#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
681#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
682#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
683#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
684#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
685#define PL_reg_magic PL_reg_state.re_state_reg_magic
686#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
687#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
688#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
689#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
690#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
691#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
692#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
693#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
694#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
695#define PL_regsize PL_reg_state.re_state_regsize
696#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
697#define PL_nrs PL_reg_state.re_state_nrs
698
1ade1aa1
NC
699struct re_save_state {
700 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
701 U32 re_state_reg_start_tmpl; /* from regexec.c */
702 I32 re_state_reg_eval_set; /* from regexec.c */
703 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
704 char *re_state_bostr;
705 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 706 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 707 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
708 U32 *re_state_reglastparen; /* Similarly for lastparen. */
709 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 710 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 711 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
712 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
713 PMOP *re_state_reg_curpm; /* from regexec.c */
714 char *re_state_reg_oldsaved; /* old saved substr during match */
715 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
716 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
717 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
718 I32 re_state_reg_maxiter; /* max wait until caching pos */
719 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 720 U32 re_state_regsize; /* from regexec.c */
1b5c067c 721 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
722 char *re_state_reg_starttry; /* from regexec.c */
723#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 724 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
725#endif
726};
727
728#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
729 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 730
1ade1aa1
NC
731/*
732 * Local variables:
733 * c-indentation-style: bsd
734 * c-basic-offset: 4
735 * indent-tabs-mode: t
736 * End:
737 *
738 * ex: set ts=8 sts=4 sw=4 noet:
739 */