This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
preserve code blocks in interpolated qr//s
[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;
b30fcab9 73 REGEXP *src_regex;
3d2bd50a
DM
74};
75
76
882227b7
AB
77/*
78 The regexp/REGEXP struct, see L<perlreapi> for further documentation
79 on the individual fields. The struct is ordered so that the most
80 commonly used fields are placed at the start.
81
82 Any patch that adds items to this struct will need to include
83 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
84 (C<pregfree()>). This involves freeing or cloning items in the
85 regexp's data array based on the data item's type.
86*/
87
08e44740
NC
88#define _REGEXP_COMMON \
89 /* what engine created this regexp? */ \
90 const struct regexp_engine* engine; \
91 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 92 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
93 /* Information about the match that the perl core uses to */ \
94 /* manage things */ \
95 U32 extflags; /* Flags used both externally and internally */ \
96 I32 minlen; /* mininum possible length of string to match */\
97 I32 minlenret; /* mininum possible length of $& */ \
98 U32 gofs; /* chars left of pos that we search from */ \
99 /* substring data about strings that must appear in the */ \
100 /* final match, used for optimisations */ \
101 struct reg_substr_data *substrs; \
102 U32 nparens; /* number of capture buffers */ \
103 /* private engine specific data */ \
104 U32 intflags; /* Engine Specific Internal flags */ \
105 void *pprivate; /* Data private to the regex engine which */ \
106 /* created this object. */ \
107 /* Data about the last/current match. These are modified */ \
108 /* during matching */ \
109 U32 lastparen; /* last open paren matched */ \
110 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 111 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
112 /* Array of offsets for (@-) and (@+) */ \
113 regexp_paren_pair *offs; \
114 /* saved or original string so \digit works forever. */ \
115 char *subbeg; \
116 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
117 I32 sublen; /* Length of string pointed by subbeg */ \
118 /* Information about the match that isn't often used */ \
119 /* offset from wrapped to the start of precomp */ \
654eccd5 120 PERL_BITFIELD32 pre_prefix:4; \
08e44740 121 /* number of eval groups in the pattern - for security checks */\
d63c20f2 122 PERL_BITFIELD32 seen_evals:28; \
3d2bd50a 123 CV *qr_anoncv; /* the anon sub wrapped round qr/(?{..})/ */ \
08e44740 124
bbe252da 125typedef struct regexp {
288b8c02 126 _XPV_HEAD;
08e44740 127 _REGEXP_COMMON;
f8fc2ecf
YO
128} regexp;
129
89f6f287 130#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 131
785a26d5 132/* used for high speed searches */
f9f4320a
YO
133typedef struct re_scream_pos_data_s
134{
135 char **scream_olds; /* match pos */
136 I32 *scream_pos; /* Internal iterator of scream. */
137} re_scream_pos_data;
138
785a26d5
YO
139/* regexp_engine structure. This is the dispatch table for regexes.
140 * Any regex engine implementation must be able to build one of these.
141 */
f9f4320a 142typedef struct regexp_engine {
1593ad57 143 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
144 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
145 char* strbeg, I32 minend, SV* screamer,
146 void* data, U32 flags);
147 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
148 char *strend, const U32 flags,
149 re_scream_pos_data *data);
49d7dfbc
AB
150 SV* (*checkstr) (pTHX_ REGEXP * const rx);
151 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 152 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 153 SV * const sv);
2fdbfb4d
AB
154 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
155 SV const * const value);
156 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
157 const I32 paren);
192b9cd1
AB
158 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
159 SV * const value, const U32 flags);
160 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
161 const U32 flags);
49d7dfbc 162 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 163#ifdef USE_ITHREADS
49d7dfbc 164 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 165#endif
f9f4320a
YO
166} regexp_engine;
167
192b9cd1
AB
168/*
169 These are passed to the numbered capture variable callbacks as the
170 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
171 $2 etc.
172*/
f1b875a0
YO
173#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
174#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
175#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
176
177/*
178 Flags that are passed to the named_buff and named_buff_iter
179 callbacks above. Those routines are called from universal.c via the
180 Tie::Hash::NamedCapture interface for %+ and %- and the re::
181 functions in the same file.
182*/
183
184/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
185#define RXapif_FETCH 0x0001
186#define RXapif_STORE 0x0002
187#define RXapif_DELETE 0x0004
188#define RXapif_CLEAR 0x0008
189#define RXapif_EXISTS 0x0010
190#define RXapif_SCALAR 0x0020
191#define RXapif_FIRSTKEY 0x0040
192#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
193
194/* Whether %+ or %- is being operated on */
f1b875a0
YO
195#define RXapif_ONE 0x0100 /* %+ */
196#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
197
198/* Whether this is being called from a re:: function */
f1b875a0
YO
199#define RXapif_REGNAME 0x0400
200#define RXapif_REGNAMES 0x0800
7948fc08 201#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 202
f7e71195
AB
203/*
204=head1 REGEXP Functions
205
206=for apidoc Am|REGEXP *|SvRX|SV *sv
207
208Convenience macro to get the REGEXP from a SV. This is approximately
209equivalent to the following snippet:
210
211 if (SvMAGICAL(sv))
212 mg_get(sv);
1af910ea
DL
213 if (SvROK(sv))
214 sv = MUTABLE_SV(SvRV(sv));
215 if (SvTYPE(sv) == SVt_REGEXP)
216 return (REGEXP*) sv;
f7e71195
AB
217
218NULL will be returned if a REGEXP* is not found.
219
220=for apidoc Am|bool|SvRXOK|SV* sv
221
1af910ea
DL
222Returns a boolean indicating whether the SV (or the one it references)
223is a REGEXP.
f7e71195
AB
224
225If you want to do something with the REGEXP* later use SvRX instead
226and check for NULL.
227
228=cut
229*/
230
231#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
232#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
233
234
7948fc08 235/* Flags stored in regexp->extflags
bbe252da 236 * These are used by code external to the regexp engine
e357fc67 237 *
5b126c84
KW
238 * Note that the flags whose names start with RXf_PMf_ are defined in
239 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 240 *
eb2624c9
FC
241 * NOTE: if you modify any RXf flags you should run regen.pl or
242 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 243 *
bbe252da
YO
244 */
245
5b126c84
KW
246#include "op_reg_common.h"
247
cde0cee5 248#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 249
a20207d7
YO
250#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
251 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
252 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
253 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
254 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 255
94b03d7d 256/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 257#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 258 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 259
bcdf7404 260/* chars and strings used as regex pattern modifiers
486ec47a 261 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
262 *
263 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
264 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
265 * for its own uses. So 'k' is out as well.
bcdf7404 266 */
85508812 267#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 268#define EXEC_PAT_MOD 'e'
87e95b7f 269#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
270#define ONCE_PAT_MOD 'o'
271#define GLOBAL_PAT_MOD 'g'
272#define CONTINUE_PAT_MOD 'c'
273#define MULTILINE_PAT_MOD 'm'
274#define SINGLE_PAT_MOD 's'
275#define IGNORE_PAT_MOD 'i'
276#define XTENDED_PAT_MOD 'x'
4f4d7508 277#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
278#define LOCALE_PAT_MOD 'l'
279#define UNICODE_PAT_MOD 'u'
50e91148 280#define DEPENDS_PAT_MOD 'd'
cfaf538b 281#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
282
283#define ONCE_PAT_MODS "o"
87e95b7f 284#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
285#define EXEC_PAT_MODS "e"
286#define LOOP_PAT_MODS "gc"
4f4d7508 287#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
288#define LOCALE_PAT_MODS "l"
289#define UNICODE_PAT_MODS "u"
50e91148 290#define DEPENDS_PAT_MODS "d"
cfaf538b 291#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 292#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 293
78fecfc8 294/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
295 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
296 * character is bit +1, etc. */
bcdf7404
YO
297#define STD_PAT_MODS "msix"
298
94b03d7d
KW
299#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
300
78fecfc8 301/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
302 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
303 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
304#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
305
306#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 307#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 308#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 309#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 310
cb5027f2 311/*
eb2624c9
FC
312 * NOTE: if you modify any RXf flags you should run regen.pl or
313 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
314 *
315 */
bcdf7404 316
e795e964
KW
317/* Leave some space, so future bit allocations can go either in the shared or
318 * unshared area without affecting binary compatibility */
df7a8460 319#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 320
bf28aba1
KW
321/* Manually decorate this function with gcc-style attributes just to
322 * avoid having to restructure the header files and their called order,
323 * as proto.h would have to be included before this file, and isn't */
324
a62b1201
KW
325PERL_STATIC_INLINE const char *
326get_regex_charset_name(const U32 flags, STRLEN* const lenp)
327 __attribute__warn_unused_result__;
328
df7a8460 329#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
330
331PERL_STATIC_INLINE const char *
332get_regex_charset_name(const U32 flags, STRLEN* const lenp)
333{
334 /* Returns a string that corresponds to the name of the regex character set
335 * given by 'flags', and *lenp is set the length of that string, which
336 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
337
338 *lenp = 1;
339 switch (get_regex_charset(flags)) {
50e91148 340 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
341 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
342 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 343 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 344 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
345 *lenp = 2;
346 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
347 default:
348 return "?"; /* Unknown */
a62b1201 349 }
a62b1201
KW
350}
351
52d81aa8 352/* Anchor and GPOS related stuff */
1ad4ec54
KW
353#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
354#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
355#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
356#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
357#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
358#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
359/* two bits here */
360#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
361#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
362#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
363
bbe252da 364/* What we have seen */
1ad4ec54
KW
365#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
366#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
367#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
368
369/* Special */
1ad4ec54
KW
370#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
371#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
372
373/* UTF8 related */
e795e964 374#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
375
376/* Intuit related */
e795e964
KW
377#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
378#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
379#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
380
381/*
382 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
383 be used by regex engines to check whether they should set
384 RXf_SKIPWHITE
385*/
e795e964 386#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 387
bbe252da
YO
388#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
389
390/* Copy and tainted info */
e795e964 391#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
392
393/* during execution: pattern temporarily tainted by executing locale ops;
394 * post-execution: $1 et al are tainted */
e795e964 395#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
396/* this pattern was tainted during compilation */
397#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
398
399/* Flags indicating special patterns */
e795e964
KW
400#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
401#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
402#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 403#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 404#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
405# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
406#endif
c737faaf 407
cb5027f2 408/*
eb2624c9
FC
409 * NOTE: if you modify any RXf flags you should run regen.pl or
410 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
411 *
412 */
bbe252da 413
bbe252da 414#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
415#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
416#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
417#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
418#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
419#define RX_MATCH_TAINTED_set(prog, t) ((t) \
420 ? RX_MATCH_TAINTED_on(prog) \
421 : RX_MATCH_TAINTED_off(prog))
c277df42 422
07bc277f
NC
423#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
424#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
425#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
426#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
427#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
428#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
429#define RX_MATCH_COPIED_set(prog,t) ((t) \
430 ? RX_MATCH_COPIED_on(prog) \
431 : RX_MATCH_COPIED_off(prog))
f9f4320a 432
07bc277f 433#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 434
07bc277f 435/* For source compatibility. We used to store these explicitly. */
f7c278bf 436#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 437#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
438/* FIXME? Are we hardcoding too much here and constraining plugin extension
439 writers? Specifically, the value 1 assumes that the wrapped version always
440 has exactly one character at the end, a ')'. Will that always be true? */
441#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 442#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 443#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 444#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 445#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 446#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 447#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
448# define RX_EXTFLAGS(prog) \
449 (*({ \
7948fc08
RU
450 const REGEXP *const _rx_extflags = (prog); \
451 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
452 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
453 }))
454# define RX_ENGINE(prog) \
455 (*({ \
7948fc08
RU
456 const REGEXP *const _rx_engine = (prog); \
457 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
458 &SvANY(_rx_engine)->engine; \
4450b630
NC
459 }))
460# define RX_SUBBEG(prog) \
461 (*({ \
7948fc08
RU
462 const REGEXP *const _rx_subbeg = (prog); \
463 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
464 &SvANY(_rx_subbeg)->subbeg; \
4450b630 465 }))
200ff815 466# define RX_OFFS(prog) \
4450b630 467 (*({ \
7948fc08
RU
468 const REGEXP *const _rx_offs = (prog); \
469 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
470 &SvANY(_rx_offs)->offs; \
4450b630
NC
471 }))
472# define RX_NPARENS(prog) \
473 (*({ \
7948fc08
RU
474 const REGEXP *const _rx_nparens = (prog); \
475 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
476 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
477 }))
478#else
479# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
480# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
481# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
482# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
483# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
484#endif
288b8c02 485#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
486#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
487#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
488#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
489#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
490#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
491#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 492#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 493
be8e71aa
YO
494#endif /* PLUGGABLE_RE_EXTENSION */
495
486ec47a 496/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 497
f8c7b90f 498#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 499#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
500 STMT_START {if (RX_SAVED_COPY(rx)) { \
501 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
502 } \
503 if (RX_MATCH_COPIED(rx)) { \
07bc277f 504 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
505 RX_MATCH_COPIED_off(rx); \
506 }} STMT_END
507#else
508#define RX_MATCH_COPY_FREE(rx) \
509 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 510 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
511 RX_MATCH_COPIED_off(rx); \
512 }} STMT_END
513#endif
514
07bc277f
NC
515#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
516#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
517#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
518#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
519#define RX_MATCH_UTF8_set(prog, t) ((t) \
520 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
521 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
522
523/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 524#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 525
cad2e5aa
JH
526#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
527#define REXEC_CHECKED 0x02 /* check_substr already checked. */
528#define REXEC_SCREAM 0x04 /* use scream table. */
529#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 530#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 531
7fe44985 532#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
533# define ReREFCNT_inc(re) \
534 ({ \
535 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
536 REGEXP *const _rerefcnt_inc = (re); \
537 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
538 SvREFCNT_inc(_rerefcnt_inc); \
539 _rerefcnt_inc; \
288b8c02
NC
540 })
541# define ReREFCNT_dec(re) \
542 ({ \
543 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
544 REGEXP *const _rerefcnt_dec = (re); \
545 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
546 })
547#else
548# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 549# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
550#endif
551
552/* FIXME for plugins. */
cf93c79d
IZ
553
554#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
555#define FBMcf_TAIL_DOLLARM 2
556#define FBMcf_TAIL_Z 4
557#define FBMcf_TAIL_z 8
558#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
559
560#define FBMrf_MULTILINE 1
cad2e5aa 561
3b0527fe
DM
562/* some basic information about the current match that is created by
563 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
564
565typedef struct {
288b8c02 566 REGEXP *prog;
3b0527fe
DM
567 char *bol;
568 char *till;
569 SV *sv;
570 char *ganch;
24b23f37 571 char *cutpoint;
3b0527fe
DM
572} regmatch_info;
573
5d9a96ca
DM
574
575/* structures for holding and saving the state maintained by regmatch() */
576
d56b3014 577#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 578#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 579#endif
6bda09f9 580
5d9a96ca
DM
581typedef I32 CHECKPOINT;
582
a0374537 583typedef struct regmatch_state {
40a82448 584 int resume_state; /* where to jump to on return */
24d3c4a9 585 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
586
587 union {
77cb431f
DM
588
589 /* this is a fake union member that matches the first element
590 * of each member that needs to store positive backtrack
591 * information */
592 struct {
593 struct regmatch_state *prev_yes_state;
594 } yes;
595
fae667d5
YO
596 /* branchlike members */
597 /* this is a fake union member that matches the first elements
598 * of each member that needs to behave like a branch */
599 struct {
5d458dd8
YO
600 /* this first element must match u.yes */
601 struct regmatch_state *prev_yes_state;
fae667d5
YO
602 U32 lastparen;
603 CHECKPOINT cp;
604
605 } branchlike;
606
607 struct {
608 /* the first elements must match u.branchlike */
609 struct regmatch_state *prev_yes_state;
610 U32 lastparen;
611 CHECKPOINT cp;
612
613 regnode *next_branch; /* next branch node */
614 } branch;
615
616 struct {
617 /* the first elements must match u.branchlike */
618 struct regmatch_state *prev_yes_state;
619 U32 lastparen;
620 CHECKPOINT cp;
621
2e64971a 622 U32 accepted; /* how many accepting states left */
7f69552c 623 U16 *jump; /* positive offsets from me */
166ba7cd 624 regnode *B; /* node following the trie */
7f69552c 625 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
626 U8 *firstpos;/* pos in string of first trie match */
627 U32 firstchars;/* len in chars of firstpos from start */
628 U16 nextword;/* next word to try */
629 U16 topword; /* longest accepted word */
630 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
631 } trie;
632
fae667d5
YO
633 /* special types - these members are used to store state for special
634 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 635 struct {
77cb431f
DM
636 /* this first element must match u.yes */
637 struct regmatch_state *prev_yes_state;
faec1544
DM
638 struct regmatch_state *prev_eval;
639 struct regmatch_state *prev_curlyx;
288b8c02 640 REGEXP *prev_rex;
faec1544
DM
641 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
642 flip when transitioning between
643 inner and outer rexen */
aa283a38
DM
644 CHECKPOINT cp; /* remember current savestack indexes */
645 CHECKPOINT lastcp;
6bda09f9 646 U32 close_paren; /* which close bracket is our end */
1b5c067c 647 regnode *B; /* the node following us */
e822a8b4
DM
648 } eval;
649
650 struct {
c476f425 651 /* this first element must match u.yes */
262b90c4 652 struct regmatch_state *prev_yes_state;
fae667d5
YO
653 I32 wanted;
654 I32 logical; /* saved copy of 'logical' var */
655 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
656 } ifmatch; /* and SUSPEND/UNLESSM */
657
658 struct {
659 /* this first element must match u.yes */
660 struct regmatch_state *prev_yes_state;
661 struct regmatch_state *prev_mark;
662 SV* mark_name;
663 char *mark_loc;
664 } mark;
665
666 struct {
667 int val;
668 } keeper;
669
670 /* quantifiers - these members are used for storing state for
671 for the regops used to implement quantifiers */
672 struct {
673 /* this first element must match u.yes */
674 struct regmatch_state *prev_yes_state;
c476f425 675 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
676 regnode *me; /* the CURLYX node */
677 regnode *B; /* the B node in /A*B/ */
c476f425
DM
678 CHECKPOINT cp; /* remember current savestack index */
679 bool minmod;
a0374537 680 int parenfloor;/* how far back to strip paren data */
c476f425
DM
681
682 /* these two are modified by WHILEM */
683 int count; /* how many instances of A we've matched */
684 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
685 } curlyx;
686
687 struct {
c476f425 688 /* this first element must match u.yes */
262b90c4 689 struct regmatch_state *prev_yes_state;
c476f425
DM
690 struct regmatch_state *save_curlyx;
691 CHECKPOINT cp; /* remember current savestack indexes */
692 CHECKPOINT lastcp;
693 char *save_lastloc; /* previous curlyx.lastloc */
694 I32 cache_offset;
695 I32 cache_mask;
e822a8b4
DM
696 } whilem;
697
698 struct {
5d458dd8
YO
699 /* this first element must match u.yes */
700 struct regmatch_state *prev_yes_state;
e822a8b4 701 I32 c1, c2; /* case fold search */
40a82448
DM
702 CHECKPOINT cp;
703 I32 alen; /* length of first-matched A string */
704 I32 count;
0cadcf80 705 bool minmod;
40a82448
DM
706 regnode *A, *B; /* the nodes corresponding to /A*B/ */
707 regnode *me; /* the curlym node */
e822a8b4
DM
708 } curlym;
709
710 struct {
3b6647e0 711 U32 paren;
c255a977 712 CHECKPOINT cp;
e822a8b4 713 I32 c1, c2; /* case fold search */
c255a977
DM
714 char *maxpos; /* highest possible point in string to match */
715 char *oldloc; /* the previous locinput */
e822a8b4 716 int count;
c255a977
DM
717 int min, max; /* {m,n} */
718 regnode *A, *B; /* the nodes corresponding to /A*B/ */
719 } curly; /* and CURLYN/PLUS/STAR */
dad79028 720
d8319b27 721 } u;
5d9a96ca
DM
722} regmatch_state;
723
724/* how many regmatch_state structs to allocate as a single slab.
725 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
726 * pointers, plus 1 for any mythical malloc overhead. */
727
728#define PERL_REGMATCH_SLAB_SLOTS \
729 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
730
731typedef struct regmatch_slab {
732 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
733 struct regmatch_slab *prev, *next;
734} regmatch_slab;
1ade1aa1 735
46ab3289
NC
736#define PL_reg_flags PL_reg_state.re_state_reg_flags
737#define PL_bostr PL_reg_state.re_state_bostr
738#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 739#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 740#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
741#define PL_reglastparen PL_reg_state.re_state_reglastparen
742#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
743#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
744#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
745#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
746#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
747#define PL_reg_magic PL_reg_state.re_state_reg_magic
748#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
749#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
750#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
751#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
752#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
753#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
754#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
755#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
756#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
757#define PL_regsize PL_reg_state.re_state_regsize
758#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
759#define PL_nrs PL_reg_state.re_state_nrs
760
1ade1aa1
NC
761struct re_save_state {
762 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
763 U32 re_state_reg_start_tmpl; /* from regexec.c */
764 I32 re_state_reg_eval_set; /* from regexec.c */
765 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
766 char *re_state_bostr;
767 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 768 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 769 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
770 U32 *re_state_reglastparen; /* Similarly for lastparen. */
771 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 772 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 773 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
774 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
775 PMOP *re_state_reg_curpm; /* from regexec.c */
776 char *re_state_reg_oldsaved; /* old saved substr during match */
777 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
778 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
779 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
780 I32 re_state_reg_maxiter; /* max wait until caching pos */
781 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 782 U32 re_state_regsize; /* from regexec.c */
1b5c067c 783 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
784 char *re_state_reg_starttry; /* from regexec.c */
785#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 786 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
787#endif
788};
789
790#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
791 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 792
1ade1aa1
NC
793/*
794 * Local variables:
795 * c-indentation-style: bsd
796 * c-basic-offset: 4
14d04a33 797 * indent-tabs-mode: nil
1ade1aa1
NC
798 * End:
799 *
14d04a33 800 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 801 */