This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
make _REGEXP_COMMON work under win32
[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; \
b82ad30c 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
3c13cae6
DM
166 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
167 OP *expr, const struct regexp_engine* eng,
168 REGEXP *VOL old_re,
169 int *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
170} regexp_engine;
171
192b9cd1
AB
172/*
173 These are passed to the numbered capture variable callbacks as the
174 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
175 $2 etc.
176*/
f1b875a0
YO
177#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
178#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
179#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
180
181/*
182 Flags that are passed to the named_buff and named_buff_iter
183 callbacks above. Those routines are called from universal.c via the
184 Tie::Hash::NamedCapture interface for %+ and %- and the re::
185 functions in the same file.
186*/
187
188/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
189#define RXapif_FETCH 0x0001
190#define RXapif_STORE 0x0002
191#define RXapif_DELETE 0x0004
192#define RXapif_CLEAR 0x0008
193#define RXapif_EXISTS 0x0010
194#define RXapif_SCALAR 0x0020
195#define RXapif_FIRSTKEY 0x0040
196#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
197
198/* Whether %+ or %- is being operated on */
f1b875a0
YO
199#define RXapif_ONE 0x0100 /* %+ */
200#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
201
202/* Whether this is being called from a re:: function */
f1b875a0
YO
203#define RXapif_REGNAME 0x0400
204#define RXapif_REGNAMES 0x0800
7948fc08 205#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 206
f7e71195
AB
207/*
208=head1 REGEXP Functions
209
210=for apidoc Am|REGEXP *|SvRX|SV *sv
211
212Convenience macro to get the REGEXP from a SV. This is approximately
213equivalent to the following snippet:
214
215 if (SvMAGICAL(sv))
216 mg_get(sv);
1af910ea
DL
217 if (SvROK(sv))
218 sv = MUTABLE_SV(SvRV(sv));
219 if (SvTYPE(sv) == SVt_REGEXP)
220 return (REGEXP*) sv;
f7e71195
AB
221
222NULL will be returned if a REGEXP* is not found.
223
224=for apidoc Am|bool|SvRXOK|SV* sv
225
1af910ea
DL
226Returns a boolean indicating whether the SV (or the one it references)
227is a REGEXP.
f7e71195
AB
228
229If you want to do something with the REGEXP* later use SvRX instead
230and check for NULL.
231
232=cut
233*/
234
235#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
236#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
237
238
7948fc08 239/* Flags stored in regexp->extflags
bbe252da 240 * These are used by code external to the regexp engine
e357fc67 241 *
5b126c84
KW
242 * Note that the flags whose names start with RXf_PMf_ are defined in
243 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 244 *
eb2624c9
FC
245 * NOTE: if you modify any RXf flags you should run regen.pl or
246 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 247 *
bbe252da
YO
248 */
249
5b126c84
KW
250#include "op_reg_common.h"
251
cde0cee5 252#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 253
a20207d7
YO
254#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
255 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
256 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
257 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
258 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 259
94b03d7d 260/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 261#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 262 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 263
bcdf7404 264/* chars and strings used as regex pattern modifiers
486ec47a 265 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
266 *
267 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
268 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
269 * for its own uses. So 'k' is out as well.
bcdf7404 270 */
85508812 271#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 272#define EXEC_PAT_MOD 'e'
87e95b7f 273#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
274#define ONCE_PAT_MOD 'o'
275#define GLOBAL_PAT_MOD 'g'
276#define CONTINUE_PAT_MOD 'c'
277#define MULTILINE_PAT_MOD 'm'
278#define SINGLE_PAT_MOD 's'
279#define IGNORE_PAT_MOD 'i'
280#define XTENDED_PAT_MOD 'x'
4f4d7508 281#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
282#define LOCALE_PAT_MOD 'l'
283#define UNICODE_PAT_MOD 'u'
50e91148 284#define DEPENDS_PAT_MOD 'd'
cfaf538b 285#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
286
287#define ONCE_PAT_MODS "o"
87e95b7f 288#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
289#define EXEC_PAT_MODS "e"
290#define LOOP_PAT_MODS "gc"
4f4d7508 291#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
292#define LOCALE_PAT_MODS "l"
293#define UNICODE_PAT_MODS "u"
50e91148 294#define DEPENDS_PAT_MODS "d"
cfaf538b 295#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 296#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 297
78fecfc8 298/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
299 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
300 * character is bit +1, etc. */
bcdf7404
YO
301#define STD_PAT_MODS "msix"
302
94b03d7d
KW
303#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
304
78fecfc8 305/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
306 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
307 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
308#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
309
310#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 311#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 312#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 313#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 314
cb5027f2 315/*
eb2624c9
FC
316 * NOTE: if you modify any RXf flags you should run regen.pl or
317 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
318 *
319 */
bcdf7404 320
e795e964
KW
321/* Leave some space, so future bit allocations can go either in the shared or
322 * unshared area without affecting binary compatibility */
df7a8460 323#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 324
bf28aba1
KW
325/* Manually decorate this function with gcc-style attributes just to
326 * avoid having to restructure the header files and their called order,
327 * as proto.h would have to be included before this file, and isn't */
328
a62b1201
KW
329PERL_STATIC_INLINE const char *
330get_regex_charset_name(const U32 flags, STRLEN* const lenp)
331 __attribute__warn_unused_result__;
332
df7a8460 333#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
334
335PERL_STATIC_INLINE const char *
336get_regex_charset_name(const U32 flags, STRLEN* const lenp)
337{
338 /* Returns a string that corresponds to the name of the regex character set
339 * given by 'flags', and *lenp is set the length of that string, which
340 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
341
342 *lenp = 1;
343 switch (get_regex_charset(flags)) {
50e91148 344 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
345 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
346 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 347 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 348 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
349 *lenp = 2;
350 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
351 default:
352 return "?"; /* Unknown */
a62b1201 353 }
a62b1201
KW
354}
355
52d81aa8 356/* Anchor and GPOS related stuff */
1ad4ec54
KW
357#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
358#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
359#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
360#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
361#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
362#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
363/* two bits here */
364#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
365#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
366#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
367
bbe252da 368/* What we have seen */
1ad4ec54
KW
369#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
370#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
371#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
372
373/* Special */
1ad4ec54
KW
374#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
375#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
376
377/* UTF8 related */
e795e964 378#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
379
380/* Intuit related */
e795e964
KW
381#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
382#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
383#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
384
385/*
386 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
387 be used by regex engines to check whether they should set
388 RXf_SKIPWHITE
389*/
e795e964 390#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 391
bbe252da
YO
392#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
393
394/* Copy and tainted info */
e795e964 395#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
396
397/* during execution: pattern temporarily tainted by executing locale ops;
398 * post-execution: $1 et al are tainted */
e795e964 399#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
400/* this pattern was tainted during compilation */
401#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
402
403/* Flags indicating special patterns */
e795e964
KW
404#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
405#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
406#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 407#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 408#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
409# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
410#endif
c737faaf 411
cb5027f2 412/*
eb2624c9
FC
413 * NOTE: if you modify any RXf flags you should run regen.pl or
414 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
415 *
416 */
bbe252da 417
bbe252da 418#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
419#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
420#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
421#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
422#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
423#define RX_MATCH_TAINTED_set(prog, t) ((t) \
424 ? RX_MATCH_TAINTED_on(prog) \
425 : RX_MATCH_TAINTED_off(prog))
c277df42 426
07bc277f
NC
427#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
428#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
429#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
430#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
431#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
432#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
433#define RX_MATCH_COPIED_set(prog,t) ((t) \
434 ? RX_MATCH_COPIED_on(prog) \
435 : RX_MATCH_COPIED_off(prog))
f9f4320a 436
07bc277f 437#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 438
07bc277f 439/* For source compatibility. We used to store these explicitly. */
f7c278bf 440#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 441#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
442/* FIXME? Are we hardcoding too much here and constraining plugin extension
443 writers? Specifically, the value 1 assumes that the wrapped version always
444 has exactly one character at the end, a ')'. Will that always be true? */
445#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 446#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 447#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 448#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 449#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 450#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 451#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
452# define RX_EXTFLAGS(prog) \
453 (*({ \
7948fc08
RU
454 const REGEXP *const _rx_extflags = (prog); \
455 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
456 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
457 }))
458# define RX_ENGINE(prog) \
459 (*({ \
7948fc08
RU
460 const REGEXP *const _rx_engine = (prog); \
461 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
462 &SvANY(_rx_engine)->engine; \
4450b630
NC
463 }))
464# define RX_SUBBEG(prog) \
465 (*({ \
7948fc08
RU
466 const REGEXP *const _rx_subbeg = (prog); \
467 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
468 &SvANY(_rx_subbeg)->subbeg; \
4450b630 469 }))
200ff815 470# define RX_OFFS(prog) \
4450b630 471 (*({ \
7948fc08
RU
472 const REGEXP *const _rx_offs = (prog); \
473 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
474 &SvANY(_rx_offs)->offs; \
4450b630
NC
475 }))
476# define RX_NPARENS(prog) \
477 (*({ \
7948fc08
RU
478 const REGEXP *const _rx_nparens = (prog); \
479 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
480 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
481 }))
482#else
483# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
484# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
485# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
486# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
487# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
488#endif
288b8c02 489#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
490#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
491#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
492#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
493#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
494#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
495#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 496#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 497
be8e71aa
YO
498#endif /* PLUGGABLE_RE_EXTENSION */
499
486ec47a 500/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 501
f8c7b90f 502#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 503#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
504 STMT_START {if (RX_SAVED_COPY(rx)) { \
505 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
506 } \
507 if (RX_MATCH_COPIED(rx)) { \
07bc277f 508 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
509 RX_MATCH_COPIED_off(rx); \
510 }} STMT_END
511#else
512#define RX_MATCH_COPY_FREE(rx) \
513 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 514 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
515 RX_MATCH_COPIED_off(rx); \
516 }} STMT_END
517#endif
518
07bc277f
NC
519#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
520#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
521#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
522#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
523#define RX_MATCH_UTF8_set(prog, t) ((t) \
524 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
525 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
526
527/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 528#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 529
cad2e5aa
JH
530#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
531#define REXEC_CHECKED 0x02 /* check_substr already checked. */
532#define REXEC_SCREAM 0x04 /* use scream table. */
533#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 534#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 535
7fe44985 536#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
537# define ReREFCNT_inc(re) \
538 ({ \
539 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
540 REGEXP *const _rerefcnt_inc = (re); \
541 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
542 SvREFCNT_inc(_rerefcnt_inc); \
543 _rerefcnt_inc; \
288b8c02
NC
544 })
545# define ReREFCNT_dec(re) \
546 ({ \
547 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
548 REGEXP *const _rerefcnt_dec = (re); \
549 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
550 })
551#else
552# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 553# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
554#endif
555
556/* FIXME for plugins. */
cf93c79d
IZ
557
558#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
559#define FBMcf_TAIL_DOLLARM 2
560#define FBMcf_TAIL_Z 4
561#define FBMcf_TAIL_z 8
562#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
563
564#define FBMrf_MULTILINE 1
cad2e5aa 565
3b0527fe
DM
566/* some basic information about the current match that is created by
567 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
568
569typedef struct {
288b8c02 570 REGEXP *prog;
3b0527fe
DM
571 char *bol;
572 char *till;
573 SV *sv;
574 char *ganch;
24b23f37 575 char *cutpoint;
3b0527fe
DM
576} regmatch_info;
577
5d9a96ca
DM
578
579/* structures for holding and saving the state maintained by regmatch() */
580
d56b3014 581#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 582#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 583#endif
6bda09f9 584
5d9a96ca
DM
585typedef I32 CHECKPOINT;
586
a0374537 587typedef struct regmatch_state {
40a82448 588 int resume_state; /* where to jump to on return */
24d3c4a9 589 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
590
591 union {
77cb431f
DM
592
593 /* this is a fake union member that matches the first element
594 * of each member that needs to store positive backtrack
595 * information */
596 struct {
597 struct regmatch_state *prev_yes_state;
598 } yes;
599
fae667d5
YO
600 /* branchlike members */
601 /* this is a fake union member that matches the first elements
602 * of each member that needs to behave like a branch */
603 struct {
5d458dd8
YO
604 /* this first element must match u.yes */
605 struct regmatch_state *prev_yes_state;
fae667d5
YO
606 U32 lastparen;
607 CHECKPOINT cp;
608
609 } branchlike;
610
611 struct {
612 /* the first elements must match u.branchlike */
613 struct regmatch_state *prev_yes_state;
614 U32 lastparen;
615 CHECKPOINT cp;
616
617 regnode *next_branch; /* next branch node */
618 } branch;
619
620 struct {
621 /* the first elements must match u.branchlike */
622 struct regmatch_state *prev_yes_state;
623 U32 lastparen;
624 CHECKPOINT cp;
625
2e64971a 626 U32 accepted; /* how many accepting states left */
7f69552c 627 U16 *jump; /* positive offsets from me */
166ba7cd 628 regnode *B; /* node following the trie */
7f69552c 629 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
630 U8 *firstpos;/* pos in string of first trie match */
631 U32 firstchars;/* len in chars of firstpos from start */
632 U16 nextword;/* next word to try */
633 U16 topword; /* longest accepted word */
634 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
635 } trie;
636
fae667d5
YO
637 /* special types - these members are used to store state for special
638 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 639 struct {
77cb431f
DM
640 /* this first element must match u.yes */
641 struct regmatch_state *prev_yes_state;
faec1544
DM
642 struct regmatch_state *prev_eval;
643 struct regmatch_state *prev_curlyx;
288b8c02 644 REGEXP *prev_rex;
faec1544
DM
645 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
646 flip when transitioning between
647 inner and outer rexen */
aa283a38
DM
648 CHECKPOINT cp; /* remember current savestack indexes */
649 CHECKPOINT lastcp;
6bda09f9 650 U32 close_paren; /* which close bracket is our end */
1b5c067c 651 regnode *B; /* the node following us */
e822a8b4
DM
652 } eval;
653
654 struct {
c476f425 655 /* this first element must match u.yes */
262b90c4 656 struct regmatch_state *prev_yes_state;
fae667d5
YO
657 I32 wanted;
658 I32 logical; /* saved copy of 'logical' var */
659 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
660 } ifmatch; /* and SUSPEND/UNLESSM */
661
662 struct {
663 /* this first element must match u.yes */
664 struct regmatch_state *prev_yes_state;
665 struct regmatch_state *prev_mark;
666 SV* mark_name;
667 char *mark_loc;
668 } mark;
669
670 struct {
671 int val;
672 } keeper;
673
674 /* quantifiers - these members are used for storing state for
675 for the regops used to implement quantifiers */
676 struct {
677 /* this first element must match u.yes */
678 struct regmatch_state *prev_yes_state;
c476f425 679 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
680 regnode *me; /* the CURLYX node */
681 regnode *B; /* the B node in /A*B/ */
c476f425
DM
682 CHECKPOINT cp; /* remember current savestack index */
683 bool minmod;
a0374537 684 int parenfloor;/* how far back to strip paren data */
c476f425
DM
685
686 /* these two are modified by WHILEM */
687 int count; /* how many instances of A we've matched */
688 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
689 } curlyx;
690
691 struct {
c476f425 692 /* this first element must match u.yes */
262b90c4 693 struct regmatch_state *prev_yes_state;
c476f425
DM
694 struct regmatch_state *save_curlyx;
695 CHECKPOINT cp; /* remember current savestack indexes */
696 CHECKPOINT lastcp;
697 char *save_lastloc; /* previous curlyx.lastloc */
698 I32 cache_offset;
699 I32 cache_mask;
e822a8b4
DM
700 } whilem;
701
702 struct {
5d458dd8
YO
703 /* this first element must match u.yes */
704 struct regmatch_state *prev_yes_state;
e822a8b4 705 I32 c1, c2; /* case fold search */
40a82448
DM
706 CHECKPOINT cp;
707 I32 alen; /* length of first-matched A string */
708 I32 count;
0cadcf80 709 bool minmod;
40a82448
DM
710 regnode *A, *B; /* the nodes corresponding to /A*B/ */
711 regnode *me; /* the curlym node */
e822a8b4
DM
712 } curlym;
713
714 struct {
3b6647e0 715 U32 paren;
c255a977 716 CHECKPOINT cp;
e822a8b4 717 I32 c1, c2; /* case fold search */
c255a977
DM
718 char *maxpos; /* highest possible point in string to match */
719 char *oldloc; /* the previous locinput */
e822a8b4 720 int count;
c255a977
DM
721 int min, max; /* {m,n} */
722 regnode *A, *B; /* the nodes corresponding to /A*B/ */
723 } curly; /* and CURLYN/PLUS/STAR */
dad79028 724
d8319b27 725 } u;
5d9a96ca
DM
726} regmatch_state;
727
728/* how many regmatch_state structs to allocate as a single slab.
729 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
730 * pointers, plus 1 for any mythical malloc overhead. */
731
732#define PERL_REGMATCH_SLAB_SLOTS \
733 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
734
735typedef struct regmatch_slab {
736 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
737 struct regmatch_slab *prev, *next;
738} regmatch_slab;
1ade1aa1 739
46ab3289
NC
740#define PL_reg_flags PL_reg_state.re_state_reg_flags
741#define PL_bostr PL_reg_state.re_state_bostr
742#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 743#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 744#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
745#define PL_reglastparen PL_reg_state.re_state_reglastparen
746#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
747#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
748#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
749#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
750#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
751#define PL_reg_magic PL_reg_state.re_state_reg_magic
752#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
753#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
754#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
755#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
756#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
757#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
758#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
759#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
760#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
761#define PL_regsize PL_reg_state.re_state_regsize
762#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
763#define PL_nrs PL_reg_state.re_state_nrs
764
1ade1aa1
NC
765struct re_save_state {
766 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
767 U32 re_state_reg_start_tmpl; /* from regexec.c */
768 I32 re_state_reg_eval_set; /* from regexec.c */
769 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
770 char *re_state_bostr;
771 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 772 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 773 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
774 U32 *re_state_reglastparen; /* Similarly for lastparen. */
775 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 776 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 777 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
778 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
779 PMOP *re_state_reg_curpm; /* from regexec.c */
780 char *re_state_reg_oldsaved; /* old saved substr during match */
781 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
782 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
783 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
784 I32 re_state_reg_maxiter; /* max wait until caching pos */
785 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 786 U32 re_state_regsize; /* from regexec.c */
1b5c067c 787 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
788 char *re_state_reg_starttry; /* from regexec.c */
789#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 790 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
791#endif
792};
793
794#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
795 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 796
1ade1aa1
NC
797/*
798 * Local variables:
799 * c-indentation-style: bsd
800 * c-basic-offset: 4
14d04a33 801 * indent-tabs-mode: nil
1ade1aa1
NC
802 * End:
803 *
14d04a33 804 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 805 */