This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
in re_op_compile(), keep code_blocks for qr//
[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
52ae8f7e 58#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 59#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 60#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
61
62/* Subtracting b from a leaves in a everything that was there that isn't in b,
63 * that is the intersection of a with b's complement */
64#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
65#endif
66
3d2bd50a
DM
67/* record the position of a (?{...}) within a pattern */
68
69struct reg_code_block {
70 STRLEN start;
71 STRLEN end;
72 OP *block;
73};
74
75
882227b7
AB
76/*
77 The regexp/REGEXP struct, see L<perlreapi> for further documentation
78 on the individual fields. The struct is ordered so that the most
79 commonly used fields are placed at the start.
80
81 Any patch that adds items to this struct will need to include
82 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
83 (C<pregfree()>). This involves freeing or cloning items in the
84 regexp's data array based on the data item's type.
85*/
86
08e44740
NC
87#define _REGEXP_COMMON \
88 /* what engine created this regexp? */ \
89 const struct regexp_engine* engine; \
90 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 91 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
92 /* Information about the match that the perl core uses to */ \
93 /* manage things */ \
94 U32 extflags; /* Flags used both externally and internally */ \
95 I32 minlen; /* mininum possible length of string to match */\
96 I32 minlenret; /* mininum possible length of $& */ \
97 U32 gofs; /* chars left of pos that we search from */ \
98 /* substring data about strings that must appear in the */ \
99 /* final match, used for optimisations */ \
100 struct reg_substr_data *substrs; \
101 U32 nparens; /* number of capture buffers */ \
102 /* private engine specific data */ \
103 U32 intflags; /* Engine Specific Internal flags */ \
104 void *pprivate; /* Data private to the regex engine which */ \
105 /* created this object. */ \
106 /* Data about the last/current match. These are modified */ \
107 /* during matching */ \
108 U32 lastparen; /* last open paren matched */ \
109 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 110 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
111 /* Array of offsets for (@-) and (@+) */ \
112 regexp_paren_pair *offs; \
113 /* saved or original string so \digit works forever. */ \
114 char *subbeg; \
115 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
116 I32 sublen; /* Length of string pointed by subbeg */ \
117 /* Information about the match that isn't often used */ \
118 /* offset from wrapped to the start of precomp */ \
654eccd5 119 PERL_BITFIELD32 pre_prefix:4; \
08e44740 120 /* number of eval groups in the pattern - for security checks */\
d63c20f2 121 PERL_BITFIELD32 seen_evals:28; \
3d2bd50a 122 CV *qr_anoncv; /* the anon sub wrapped round qr/(?{..})/ */ \
08e44740 123
bbe252da 124typedef struct regexp {
288b8c02 125 _XPV_HEAD;
08e44740 126 _REGEXP_COMMON;
f8fc2ecf
YO
127} regexp;
128
89f6f287 129#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 130
785a26d5 131/* used for high speed searches */
f9f4320a
YO
132typedef struct re_scream_pos_data_s
133{
134 char **scream_olds; /* match pos */
135 I32 *scream_pos; /* Internal iterator of scream. */
136} re_scream_pos_data;
137
785a26d5
YO
138/* regexp_engine structure. This is the dispatch table for regexes.
139 * Any regex engine implementation must be able to build one of these.
140 */
f9f4320a 141typedef struct regexp_engine {
1593ad57 142 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
143 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
144 char* strbeg, I32 minend, SV* screamer,
145 void* data, U32 flags);
146 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
147 char *strend, const U32 flags,
148 re_scream_pos_data *data);
49d7dfbc
AB
149 SV* (*checkstr) (pTHX_ REGEXP * const rx);
150 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 151 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 152 SV * const sv);
2fdbfb4d
AB
153 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
154 SV const * const value);
155 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
156 const I32 paren);
192b9cd1
AB
157 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
158 SV * const value, const U32 flags);
159 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
160 const U32 flags);
49d7dfbc 161 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 162#ifdef USE_ITHREADS
49d7dfbc 163 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 164#endif
f9f4320a
YO
165} regexp_engine;
166
192b9cd1
AB
167/*
168 These are passed to the numbered capture variable callbacks as the
169 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
170 $2 etc.
171*/
f1b875a0
YO
172#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
173#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
174#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
175
176/*
177 Flags that are passed to the named_buff and named_buff_iter
178 callbacks above. Those routines are called from universal.c via the
179 Tie::Hash::NamedCapture interface for %+ and %- and the re::
180 functions in the same file.
181*/
182
183/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
184#define RXapif_FETCH 0x0001
185#define RXapif_STORE 0x0002
186#define RXapif_DELETE 0x0004
187#define RXapif_CLEAR 0x0008
188#define RXapif_EXISTS 0x0010
189#define RXapif_SCALAR 0x0020
190#define RXapif_FIRSTKEY 0x0040
191#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
192
193/* Whether %+ or %- is being operated on */
f1b875a0
YO
194#define RXapif_ONE 0x0100 /* %+ */
195#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
196
197/* Whether this is being called from a re:: function */
f1b875a0
YO
198#define RXapif_REGNAME 0x0400
199#define RXapif_REGNAMES 0x0800
7948fc08 200#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 201
f7e71195
AB
202/*
203=head1 REGEXP Functions
204
205=for apidoc Am|REGEXP *|SvRX|SV *sv
206
207Convenience macro to get the REGEXP from a SV. This is approximately
208equivalent to the following snippet:
209
210 if (SvMAGICAL(sv))
211 mg_get(sv);
1af910ea
DL
212 if (SvROK(sv))
213 sv = MUTABLE_SV(SvRV(sv));
214 if (SvTYPE(sv) == SVt_REGEXP)
215 return (REGEXP*) sv;
f7e71195
AB
216
217NULL will be returned if a REGEXP* is not found.
218
219=for apidoc Am|bool|SvRXOK|SV* sv
220
1af910ea
DL
221Returns a boolean indicating whether the SV (or the one it references)
222is a REGEXP.
f7e71195
AB
223
224If you want to do something with the REGEXP* later use SvRX instead
225and check for NULL.
226
227=cut
228*/
229
230#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
231#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
232
233
7948fc08 234/* Flags stored in regexp->extflags
bbe252da 235 * These are used by code external to the regexp engine
e357fc67 236 *
5b126c84
KW
237 * Note that the flags whose names start with RXf_PMf_ are defined in
238 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 239 *
eb2624c9
FC
240 * NOTE: if you modify any RXf flags you should run regen.pl or
241 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 242 *
bbe252da
YO
243 */
244
5b126c84
KW
245#include "op_reg_common.h"
246
cde0cee5 247#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 248
a20207d7
YO
249#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
250 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
251 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
252 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
253 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 254
94b03d7d 255/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 256#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 257 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 258
bcdf7404 259/* chars and strings used as regex pattern modifiers
486ec47a 260 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
261 *
262 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
263 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
264 * for its own uses. So 'k' is out as well.
bcdf7404 265 */
85508812 266#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 267#define EXEC_PAT_MOD 'e'
87e95b7f 268#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
269#define ONCE_PAT_MOD 'o'
270#define GLOBAL_PAT_MOD 'g'
271#define CONTINUE_PAT_MOD 'c'
272#define MULTILINE_PAT_MOD 'm'
273#define SINGLE_PAT_MOD 's'
274#define IGNORE_PAT_MOD 'i'
275#define XTENDED_PAT_MOD 'x'
4f4d7508 276#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
277#define LOCALE_PAT_MOD 'l'
278#define UNICODE_PAT_MOD 'u'
50e91148 279#define DEPENDS_PAT_MOD 'd'
cfaf538b 280#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
281
282#define ONCE_PAT_MODS "o"
87e95b7f 283#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
284#define EXEC_PAT_MODS "e"
285#define LOOP_PAT_MODS "gc"
4f4d7508 286#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
287#define LOCALE_PAT_MODS "l"
288#define UNICODE_PAT_MODS "u"
50e91148 289#define DEPENDS_PAT_MODS "d"
cfaf538b 290#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 291#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 292
78fecfc8 293/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
294 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
295 * character is bit +1, etc. */
bcdf7404
YO
296#define STD_PAT_MODS "msix"
297
94b03d7d
KW
298#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
299
78fecfc8 300/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
301 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
302 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
303#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
304
305#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 306#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 307#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 308#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 309
cb5027f2 310/*
eb2624c9
FC
311 * NOTE: if you modify any RXf flags you should run regen.pl or
312 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
313 *
314 */
bcdf7404 315
e795e964
KW
316/* Leave some space, so future bit allocations can go either in the shared or
317 * unshared area without affecting binary compatibility */
df7a8460 318#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 319
bf28aba1
KW
320/* Manually decorate this function with gcc-style attributes just to
321 * avoid having to restructure the header files and their called order,
322 * as proto.h would have to be included before this file, and isn't */
323
a62b1201
KW
324PERL_STATIC_INLINE const char *
325get_regex_charset_name(const U32 flags, STRLEN* const lenp)
326 __attribute__warn_unused_result__;
327
df7a8460 328#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
329
330PERL_STATIC_INLINE const char *
331get_regex_charset_name(const U32 flags, STRLEN* const lenp)
332{
333 /* Returns a string that corresponds to the name of the regex character set
334 * given by 'flags', and *lenp is set the length of that string, which
335 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
336
337 *lenp = 1;
338 switch (get_regex_charset(flags)) {
50e91148 339 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
340 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
341 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 342 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 343 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
344 *lenp = 2;
345 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
346 default:
347 return "?"; /* Unknown */
a62b1201 348 }
a62b1201
KW
349}
350
52d81aa8 351/* Anchor and GPOS related stuff */
1ad4ec54
KW
352#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
353#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
354#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
355#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
356#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
357#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
358/* two bits here */
359#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
360#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
361#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
362
bbe252da 363/* What we have seen */
1ad4ec54
KW
364#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
365#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
366#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
367
368/* Special */
1ad4ec54
KW
369#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
370#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
371
372/* UTF8 related */
e795e964 373#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
374
375/* Intuit related */
e795e964
KW
376#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
377#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
378#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
379
380/*
381 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
382 be used by regex engines to check whether they should set
383 RXf_SKIPWHITE
384*/
e795e964 385#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 386
bbe252da
YO
387#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
388
389/* Copy and tainted info */
e795e964 390#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
391
392/* during execution: pattern temporarily tainted by executing locale ops;
393 * post-execution: $1 et al are tainted */
e795e964 394#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
395/* this pattern was tainted during compilation */
396#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
397
398/* Flags indicating special patterns */
e795e964
KW
399#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
400#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
401#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 402#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 403#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
404# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
405#endif
c737faaf 406
cb5027f2 407/*
eb2624c9
FC
408 * NOTE: if you modify any RXf flags you should run regen.pl or
409 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
410 *
411 */
bbe252da 412
bbe252da 413#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
414#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
415#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
416#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
417#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
418#define RX_MATCH_TAINTED_set(prog, t) ((t) \
419 ? RX_MATCH_TAINTED_on(prog) \
420 : RX_MATCH_TAINTED_off(prog))
c277df42 421
07bc277f
NC
422#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
423#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
424#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
425#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
426#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
427#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
428#define RX_MATCH_COPIED_set(prog,t) ((t) \
429 ? RX_MATCH_COPIED_on(prog) \
430 : RX_MATCH_COPIED_off(prog))
f9f4320a 431
07bc277f 432#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 433
07bc277f 434/* For source compatibility. We used to store these explicitly. */
f7c278bf 435#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 436#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
437/* FIXME? Are we hardcoding too much here and constraining plugin extension
438 writers? Specifically, the value 1 assumes that the wrapped version always
439 has exactly one character at the end, a ')'. Will that always be true? */
440#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 441#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 442#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 443#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 444#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 445#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 446#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
447# define RX_EXTFLAGS(prog) \
448 (*({ \
7948fc08
RU
449 const REGEXP *const _rx_extflags = (prog); \
450 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
451 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
452 }))
453# define RX_ENGINE(prog) \
454 (*({ \
7948fc08
RU
455 const REGEXP *const _rx_engine = (prog); \
456 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
457 &SvANY(_rx_engine)->engine; \
4450b630
NC
458 }))
459# define RX_SUBBEG(prog) \
460 (*({ \
7948fc08
RU
461 const REGEXP *const _rx_subbeg = (prog); \
462 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
463 &SvANY(_rx_subbeg)->subbeg; \
4450b630 464 }))
200ff815 465# define RX_OFFS(prog) \
4450b630 466 (*({ \
7948fc08
RU
467 const REGEXP *const _rx_offs = (prog); \
468 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
469 &SvANY(_rx_offs)->offs; \
4450b630
NC
470 }))
471# define RX_NPARENS(prog) \
472 (*({ \
7948fc08
RU
473 const REGEXP *const _rx_nparens = (prog); \
474 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
475 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
476 }))
477#else
478# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
479# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
480# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
481# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
482# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
483#endif
288b8c02 484#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
485#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
486#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
487#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
488#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
489#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
490#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 491#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 492
be8e71aa
YO
493#endif /* PLUGGABLE_RE_EXTENSION */
494
486ec47a 495/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 496
f8c7b90f 497#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 498#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
499 STMT_START {if (RX_SAVED_COPY(rx)) { \
500 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
501 } \
502 if (RX_MATCH_COPIED(rx)) { \
07bc277f 503 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
504 RX_MATCH_COPIED_off(rx); \
505 }} STMT_END
506#else
507#define RX_MATCH_COPY_FREE(rx) \
508 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 509 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
510 RX_MATCH_COPIED_off(rx); \
511 }} STMT_END
512#endif
513
07bc277f
NC
514#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
515#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
516#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
517#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
518#define RX_MATCH_UTF8_set(prog, t) ((t) \
519 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
520 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
521
522/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 523#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 524
cad2e5aa
JH
525#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
526#define REXEC_CHECKED 0x02 /* check_substr already checked. */
527#define REXEC_SCREAM 0x04 /* use scream table. */
528#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 529#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 530
7fe44985 531#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
532# define ReREFCNT_inc(re) \
533 ({ \
534 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
535 REGEXP *const _rerefcnt_inc = (re); \
536 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
537 SvREFCNT_inc(_rerefcnt_inc); \
538 _rerefcnt_inc; \
288b8c02
NC
539 })
540# define ReREFCNT_dec(re) \
541 ({ \
542 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
543 REGEXP *const _rerefcnt_dec = (re); \
544 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
545 })
546#else
547# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 548# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
549#endif
550
551/* FIXME for plugins. */
cf93c79d
IZ
552
553#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
554#define FBMcf_TAIL_DOLLARM 2
555#define FBMcf_TAIL_Z 4
556#define FBMcf_TAIL_z 8
557#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
558
559#define FBMrf_MULTILINE 1
cad2e5aa 560
3b0527fe
DM
561/* some basic information about the current match that is created by
562 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
563
564typedef struct {
288b8c02 565 REGEXP *prog;
3b0527fe
DM
566 char *bol;
567 char *till;
568 SV *sv;
569 char *ganch;
24b23f37 570 char *cutpoint;
3b0527fe
DM
571} regmatch_info;
572
5d9a96ca
DM
573
574/* structures for holding and saving the state maintained by regmatch() */
575
d56b3014 576#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 577#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 578#endif
6bda09f9 579
5d9a96ca
DM
580typedef I32 CHECKPOINT;
581
a0374537 582typedef struct regmatch_state {
40a82448 583 int resume_state; /* where to jump to on return */
24d3c4a9 584 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
585
586 union {
77cb431f
DM
587
588 /* this is a fake union member that matches the first element
589 * of each member that needs to store positive backtrack
590 * information */
591 struct {
592 struct regmatch_state *prev_yes_state;
593 } yes;
594
fae667d5
YO
595 /* branchlike members */
596 /* this is a fake union member that matches the first elements
597 * of each member that needs to behave like a branch */
598 struct {
5d458dd8
YO
599 /* this first element must match u.yes */
600 struct regmatch_state *prev_yes_state;
fae667d5
YO
601 U32 lastparen;
602 CHECKPOINT cp;
603
604 } branchlike;
605
606 struct {
607 /* the first elements must match u.branchlike */
608 struct regmatch_state *prev_yes_state;
609 U32 lastparen;
610 CHECKPOINT cp;
611
612 regnode *next_branch; /* next branch node */
613 } branch;
614
615 struct {
616 /* the first elements must match u.branchlike */
617 struct regmatch_state *prev_yes_state;
618 U32 lastparen;
619 CHECKPOINT cp;
620
2e64971a 621 U32 accepted; /* how many accepting states left */
7f69552c 622 U16 *jump; /* positive offsets from me */
166ba7cd 623 regnode *B; /* node following the trie */
7f69552c 624 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
625 U8 *firstpos;/* pos in string of first trie match */
626 U32 firstchars;/* len in chars of firstpos from start */
627 U16 nextword;/* next word to try */
628 U16 topword; /* longest accepted word */
629 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
630 } trie;
631
fae667d5
YO
632 /* special types - these members are used to store state for special
633 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 634 struct {
77cb431f
DM
635 /* this first element must match u.yes */
636 struct regmatch_state *prev_yes_state;
faec1544
DM
637 struct regmatch_state *prev_eval;
638 struct regmatch_state *prev_curlyx;
288b8c02 639 REGEXP *prev_rex;
faec1544
DM
640 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
641 flip when transitioning between
642 inner and outer rexen */
aa283a38
DM
643 CHECKPOINT cp; /* remember current savestack indexes */
644 CHECKPOINT lastcp;
6bda09f9 645 U32 close_paren; /* which close bracket is our end */
1b5c067c 646 regnode *B; /* the node following us */
e822a8b4
DM
647 } eval;
648
649 struct {
c476f425 650 /* this first element must match u.yes */
262b90c4 651 struct regmatch_state *prev_yes_state;
fae667d5
YO
652 I32 wanted;
653 I32 logical; /* saved copy of 'logical' var */
654 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
655 } ifmatch; /* and SUSPEND/UNLESSM */
656
657 struct {
658 /* this first element must match u.yes */
659 struct regmatch_state *prev_yes_state;
660 struct regmatch_state *prev_mark;
661 SV* mark_name;
662 char *mark_loc;
663 } mark;
664
665 struct {
666 int val;
667 } keeper;
668
669 /* quantifiers - these members are used for storing state for
670 for the regops used to implement quantifiers */
671 struct {
672 /* this first element must match u.yes */
673 struct regmatch_state *prev_yes_state;
c476f425 674 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
675 regnode *me; /* the CURLYX node */
676 regnode *B; /* the B node in /A*B/ */
c476f425
DM
677 CHECKPOINT cp; /* remember current savestack index */
678 bool minmod;
a0374537 679 int parenfloor;/* how far back to strip paren data */
c476f425
DM
680
681 /* these two are modified by WHILEM */
682 int count; /* how many instances of A we've matched */
683 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
684 } curlyx;
685
686 struct {
c476f425 687 /* this first element must match u.yes */
262b90c4 688 struct regmatch_state *prev_yes_state;
c476f425
DM
689 struct regmatch_state *save_curlyx;
690 CHECKPOINT cp; /* remember current savestack indexes */
691 CHECKPOINT lastcp;
692 char *save_lastloc; /* previous curlyx.lastloc */
693 I32 cache_offset;
694 I32 cache_mask;
e822a8b4
DM
695 } whilem;
696
697 struct {
5d458dd8
YO
698 /* this first element must match u.yes */
699 struct regmatch_state *prev_yes_state;
e822a8b4 700 I32 c1, c2; /* case fold search */
40a82448
DM
701 CHECKPOINT cp;
702 I32 alen; /* length of first-matched A string */
703 I32 count;
0cadcf80 704 bool minmod;
40a82448
DM
705 regnode *A, *B; /* the nodes corresponding to /A*B/ */
706 regnode *me; /* the curlym node */
e822a8b4
DM
707 } curlym;
708
709 struct {
3b6647e0 710 U32 paren;
c255a977 711 CHECKPOINT cp;
e822a8b4 712 I32 c1, c2; /* case fold search */
c255a977
DM
713 char *maxpos; /* highest possible point in string to match */
714 char *oldloc; /* the previous locinput */
e822a8b4 715 int count;
c255a977
DM
716 int min, max; /* {m,n} */
717 regnode *A, *B; /* the nodes corresponding to /A*B/ */
718 } curly; /* and CURLYN/PLUS/STAR */
dad79028 719
d8319b27 720 } u;
5d9a96ca
DM
721} regmatch_state;
722
723/* how many regmatch_state structs to allocate as a single slab.
724 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
725 * pointers, plus 1 for any mythical malloc overhead. */
726
727#define PERL_REGMATCH_SLAB_SLOTS \
728 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
729
730typedef struct regmatch_slab {
731 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
732 struct regmatch_slab *prev, *next;
733} regmatch_slab;
1ade1aa1 734
46ab3289
NC
735#define PL_reg_flags PL_reg_state.re_state_reg_flags
736#define PL_bostr PL_reg_state.re_state_bostr
737#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 738#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 739#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
740#define PL_reglastparen PL_reg_state.re_state_reglastparen
741#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
742#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
743#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
744#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
745#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
746#define PL_reg_magic PL_reg_state.re_state_reg_magic
747#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
748#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
749#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
750#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
751#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
752#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
753#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
754#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
755#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
756#define PL_regsize PL_reg_state.re_state_regsize
757#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
758#define PL_nrs PL_reg_state.re_state_nrs
759
1ade1aa1
NC
760struct re_save_state {
761 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
762 U32 re_state_reg_start_tmpl; /* from regexec.c */
763 I32 re_state_reg_eval_set; /* from regexec.c */
764 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
765 char *re_state_bostr;
766 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 767 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 768 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
769 U32 *re_state_reglastparen; /* Similarly for lastparen. */
770 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 771 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 772 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
773 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
774 PMOP *re_state_reg_curpm; /* from regexec.c */
775 char *re_state_reg_oldsaved; /* old saved substr during match */
776 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
777 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
778 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
779 I32 re_state_reg_maxiter; /* max wait until caching pos */
780 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 781 U32 re_state_regsize; /* from regexec.c */
1b5c067c 782 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
783 char *re_state_reg_starttry; /* from regexec.c */
784#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 785 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
786#endif
787};
788
789#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
790 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 791
1ade1aa1
NC
792/*
793 * Local variables:
794 * c-indentation-style: bsd
795 * c-basic-offset: 4
14d04a33 796 * indent-tabs-mode: nil
1ade1aa1
NC
797 * End:
798 *
14d04a33 799 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 800 */