This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
reduce size of struct regmatch_state
[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
b3fd53f3
DM
53/* offsets within a string of a particular /(.)/ capture */
54
f0ab9afb
NC
55typedef struct regexp_paren_pair {
56 I32 start;
57 I32 end;
b3fd53f3
DM
58 /* 'start_tmp' records a new opening position before the matching end
59 * has been found, so that the old start and end values are still
60 * valid, e.g.
61 * "abc" =~ /(.(?{print "[$1]"}))+/
62 *outputs [][a][b]
63 * This field is not part of the API. */
1ca2007e 64 I32 start_tmp;
f0ab9afb 65} regexp_paren_pair;
28d8d7f4 66
52ae8f7e 67#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 68#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 69#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
70
71/* Subtracting b from a leaves in a everything that was there that isn't in b,
72 * that is the intersection of a with b's complement */
73#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
74#endif
75
3d2bd50a
DM
76/* record the position of a (?{...}) within a pattern */
77
78struct reg_code_block {
79 STRLEN start;
80 STRLEN end;
81 OP *block;
b30fcab9 82 REGEXP *src_regex;
3d2bd50a
DM
83};
84
85
882227b7
AB
86/*
87 The regexp/REGEXP struct, see L<perlreapi> for further documentation
88 on the individual fields. The struct is ordered so that the most
89 commonly used fields are placed at the start.
90
91 Any patch that adds items to this struct will need to include
92 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
93 (C<pregfree()>). This involves freeing or cloning items in the
94 regexp's data array based on the data item's type.
95*/
96
08e44740
NC
97#define _REGEXP_COMMON \
98 /* what engine created this regexp? */ \
99 const struct regexp_engine* engine; \
100 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 101 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
102 /* Information about the match that the perl core uses to */ \
103 /* manage things */ \
104 U32 extflags; /* Flags used both externally and internally */ \
105 I32 minlen; /* mininum possible length of string to match */\
106 I32 minlenret; /* mininum possible length of $& */ \
107 U32 gofs; /* chars left of pos that we search from */ \
108 /* substring data about strings that must appear in the */ \
109 /* final match, used for optimisations */ \
110 struct reg_substr_data *substrs; \
111 U32 nparens; /* number of capture buffers */ \
112 /* private engine specific data */ \
113 U32 intflags; /* Engine Specific Internal flags */ \
114 void *pprivate; /* Data private to the regex engine which */ \
115 /* created this object. */ \
116 /* Data about the last/current match. These are modified */ \
117 /* during matching */ \
118 U32 lastparen; /* last open paren matched */ \
119 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 120 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
121 /* Array of offsets for (@-) and (@+) */ \
122 regexp_paren_pair *offs; \
123 /* saved or original string so \digit works forever. */ \
124 char *subbeg; \
125 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
126 I32 sublen; /* Length of string pointed by subbeg */ \
127 /* Information about the match that isn't often used */ \
128 /* offset from wrapped to the start of precomp */ \
654eccd5 129 PERL_BITFIELD32 pre_prefix:4; \
b82ad30c 130 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 131
bbe252da 132typedef struct regexp {
288b8c02 133 _XPV_HEAD;
08e44740 134 _REGEXP_COMMON;
f8fc2ecf
YO
135} regexp;
136
89f6f287 137#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 138
785a26d5 139/* used for high speed searches */
f9f4320a
YO
140typedef struct re_scream_pos_data_s
141{
142 char **scream_olds; /* match pos */
143 I32 *scream_pos; /* Internal iterator of scream. */
144} re_scream_pos_data;
145
785a26d5
YO
146/* regexp_engine structure. This is the dispatch table for regexes.
147 * Any regex engine implementation must be able to build one of these.
148 */
f9f4320a 149typedef struct regexp_engine {
1593ad57 150 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
151 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
152 char* strbeg, I32 minend, SV* screamer,
153 void* data, U32 flags);
154 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
155 char *strend, const U32 flags,
156 re_scream_pos_data *data);
49d7dfbc
AB
157 SV* (*checkstr) (pTHX_ REGEXP * const rx);
158 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 159 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 160 SV * const sv);
2fdbfb4d
AB
161 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
162 SV const * const value);
163 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
164 const I32 paren);
192b9cd1
AB
165 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
166 SV * const value, const U32 flags);
167 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
168 const U32 flags);
49d7dfbc 169 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 170#ifdef USE_ITHREADS
49d7dfbc 171 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 172#endif
3c13cae6
DM
173 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
174 OP *expr, const struct regexp_engine* eng,
175 REGEXP *VOL old_re,
76ac488f 176 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
177} regexp_engine;
178
192b9cd1
AB
179/*
180 These are passed to the numbered capture variable callbacks as the
181 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
182 $2 etc.
183*/
f1b875a0
YO
184#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
185#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
186#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
187
188/*
189 Flags that are passed to the named_buff and named_buff_iter
190 callbacks above. Those routines are called from universal.c via the
191 Tie::Hash::NamedCapture interface for %+ and %- and the re::
192 functions in the same file.
193*/
194
195/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
196#define RXapif_FETCH 0x0001
197#define RXapif_STORE 0x0002
198#define RXapif_DELETE 0x0004
199#define RXapif_CLEAR 0x0008
200#define RXapif_EXISTS 0x0010
201#define RXapif_SCALAR 0x0020
202#define RXapif_FIRSTKEY 0x0040
203#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
204
205/* Whether %+ or %- is being operated on */
f1b875a0
YO
206#define RXapif_ONE 0x0100 /* %+ */
207#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
208
209/* Whether this is being called from a re:: function */
f1b875a0
YO
210#define RXapif_REGNAME 0x0400
211#define RXapif_REGNAMES 0x0800
7948fc08 212#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 213
f7e71195
AB
214/*
215=head1 REGEXP Functions
216
217=for apidoc Am|REGEXP *|SvRX|SV *sv
218
219Convenience macro to get the REGEXP from a SV. This is approximately
220equivalent to the following snippet:
221
222 if (SvMAGICAL(sv))
223 mg_get(sv);
1af910ea
DL
224 if (SvROK(sv))
225 sv = MUTABLE_SV(SvRV(sv));
226 if (SvTYPE(sv) == SVt_REGEXP)
227 return (REGEXP*) sv;
f7e71195
AB
228
229NULL will be returned if a REGEXP* is not found.
230
231=for apidoc Am|bool|SvRXOK|SV* sv
232
1af910ea
DL
233Returns a boolean indicating whether the SV (or the one it references)
234is a REGEXP.
f7e71195
AB
235
236If you want to do something with the REGEXP* later use SvRX instead
237and check for NULL.
238
239=cut
240*/
241
242#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
243#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
244
245
7948fc08 246/* Flags stored in regexp->extflags
bbe252da 247 * These are used by code external to the regexp engine
e357fc67 248 *
5b126c84
KW
249 * Note that the flags whose names start with RXf_PMf_ are defined in
250 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 251 *
eb2624c9
FC
252 * NOTE: if you modify any RXf flags you should run regen.pl or
253 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 254 *
bbe252da
YO
255 */
256
5b126c84
KW
257#include "op_reg_common.h"
258
cde0cee5 259#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 260
a20207d7
YO
261#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
262 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
263 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
264 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
265 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 266
94b03d7d 267/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 268#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 269 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 270
bcdf7404 271/* chars and strings used as regex pattern modifiers
486ec47a 272 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
273 *
274 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
275 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
276 * for its own uses. So 'k' is out as well.
bcdf7404 277 */
85508812 278#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 279#define EXEC_PAT_MOD 'e'
87e95b7f 280#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
281#define ONCE_PAT_MOD 'o'
282#define GLOBAL_PAT_MOD 'g'
283#define CONTINUE_PAT_MOD 'c'
284#define MULTILINE_PAT_MOD 'm'
285#define SINGLE_PAT_MOD 's'
286#define IGNORE_PAT_MOD 'i'
287#define XTENDED_PAT_MOD 'x'
4f4d7508 288#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
289#define LOCALE_PAT_MOD 'l'
290#define UNICODE_PAT_MOD 'u'
50e91148 291#define DEPENDS_PAT_MOD 'd'
cfaf538b 292#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
293
294#define ONCE_PAT_MODS "o"
87e95b7f 295#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
296#define EXEC_PAT_MODS "e"
297#define LOOP_PAT_MODS "gc"
4f4d7508 298#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
299#define LOCALE_PAT_MODS "l"
300#define UNICODE_PAT_MODS "u"
50e91148 301#define DEPENDS_PAT_MODS "d"
cfaf538b 302#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 303#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 304
78fecfc8 305/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
306 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
307 * character is bit +1, etc. */
bcdf7404
YO
308#define STD_PAT_MODS "msix"
309
94b03d7d
KW
310#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
311
78fecfc8 312/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
313 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
314 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
315#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
316
317#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 318#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 319#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 320#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 321
cb5027f2 322/*
eb2624c9
FC
323 * NOTE: if you modify any RXf flags you should run regen.pl or
324 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
325 *
326 */
bcdf7404 327
e795e964
KW
328/* Leave some space, so future bit allocations can go either in the shared or
329 * unshared area without affecting binary compatibility */
df7a8460 330#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 331
bf28aba1
KW
332/* Manually decorate this function with gcc-style attributes just to
333 * avoid having to restructure the header files and their called order,
334 * as proto.h would have to be included before this file, and isn't */
335
a62b1201
KW
336PERL_STATIC_INLINE const char *
337get_regex_charset_name(const U32 flags, STRLEN* const lenp)
338 __attribute__warn_unused_result__;
339
df7a8460 340#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
341
342PERL_STATIC_INLINE const char *
343get_regex_charset_name(const U32 flags, STRLEN* const lenp)
344{
345 /* Returns a string that corresponds to the name of the regex character set
346 * given by 'flags', and *lenp is set the length of that string, which
347 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
348
349 *lenp = 1;
350 switch (get_regex_charset(flags)) {
50e91148 351 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
352 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
353 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 354 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 355 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
356 *lenp = 2;
357 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
358 default:
359 return "?"; /* Unknown */
a62b1201 360 }
a62b1201
KW
361}
362
52d81aa8 363/* Anchor and GPOS related stuff */
1ad4ec54
KW
364#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
365#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
366#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
367#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
368#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
369#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
370/* two bits here */
371#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
372#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
373#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
374
bbe252da 375/* What we have seen */
1ad4ec54
KW
376#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
377#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
378#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
379
380/* Special */
1ad4ec54
KW
381#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
382#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
383
384/* UTF8 related */
e795e964 385#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
386
387/* Intuit related */
e795e964
KW
388#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
389#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
390#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
391
392/*
393 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
394 be used by regex engines to check whether they should set
395 RXf_SKIPWHITE
396*/
e795e964 397#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 398
bbe252da
YO
399#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
400
401/* Copy and tainted info */
e795e964 402#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
403
404/* during execution: pattern temporarily tainted by executing locale ops;
405 * post-execution: $1 et al are tainted */
e795e964 406#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
407/* this pattern was tainted during compilation */
408#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
409
410/* Flags indicating special patterns */
e795e964
KW
411#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
412#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
413#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 414#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 415#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
416# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
417#endif
c737faaf 418
cb5027f2 419/*
eb2624c9
FC
420 * NOTE: if you modify any RXf flags you should run regen.pl or
421 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
422 *
423 */
bbe252da 424
bbe252da 425#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
426#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
427#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
428#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
429#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
430#define RX_MATCH_TAINTED_set(prog, t) ((t) \
431 ? RX_MATCH_TAINTED_on(prog) \
432 : RX_MATCH_TAINTED_off(prog))
c277df42 433
07bc277f
NC
434#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
435#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
436#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
437#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
438#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
439#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
440#define RX_MATCH_COPIED_set(prog,t) ((t) \
441 ? RX_MATCH_COPIED_on(prog) \
442 : RX_MATCH_COPIED_off(prog))
f9f4320a 443
07bc277f 444#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 445
07bc277f 446/* For source compatibility. We used to store these explicitly. */
f7c278bf 447#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 448#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
449/* FIXME? Are we hardcoding too much here and constraining plugin extension
450 writers? Specifically, the value 1 assumes that the wrapped version always
451 has exactly one character at the end, a ')'. Will that always be true? */
452#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 453#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 454#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 455#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 456#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 457#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 458#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
459# define RX_EXTFLAGS(prog) \
460 (*({ \
7948fc08
RU
461 const REGEXP *const _rx_extflags = (prog); \
462 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
463 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
464 }))
465# define RX_ENGINE(prog) \
466 (*({ \
7948fc08
RU
467 const REGEXP *const _rx_engine = (prog); \
468 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
469 &SvANY(_rx_engine)->engine; \
4450b630
NC
470 }))
471# define RX_SUBBEG(prog) \
472 (*({ \
7948fc08
RU
473 const REGEXP *const _rx_subbeg = (prog); \
474 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
475 &SvANY(_rx_subbeg)->subbeg; \
4450b630 476 }))
200ff815 477# define RX_OFFS(prog) \
4450b630 478 (*({ \
7948fc08
RU
479 const REGEXP *const _rx_offs = (prog); \
480 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
481 &SvANY(_rx_offs)->offs; \
4450b630
NC
482 }))
483# define RX_NPARENS(prog) \
484 (*({ \
7948fc08
RU
485 const REGEXP *const _rx_nparens = (prog); \
486 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
487 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
488 }))
489#else
490# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
491# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
492# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
493# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
494# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
495#endif
288b8c02 496#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
497#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
498#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
499#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
500#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
501#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
bdd9a1b1 502#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 503
be8e71aa
YO
504#endif /* PLUGGABLE_RE_EXTENSION */
505
486ec47a 506/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 507
f8c7b90f 508#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 509#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
510 STMT_START {if (RX_SAVED_COPY(rx)) { \
511 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
512 } \
513 if (RX_MATCH_COPIED(rx)) { \
07bc277f 514 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
515 RX_MATCH_COPIED_off(rx); \
516 }} STMT_END
517#else
518#define RX_MATCH_COPY_FREE(rx) \
519 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 520 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
521 RX_MATCH_COPIED_off(rx); \
522 }} STMT_END
523#endif
524
07bc277f
NC
525#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
526#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
527#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
528#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
529#define RX_MATCH_UTF8_set(prog, t) ((t) \
530 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
531 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
532
533/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 534#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 535
cad2e5aa
JH
536#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
537#define REXEC_CHECKED 0x02 /* check_substr already checked. */
538#define REXEC_SCREAM 0x04 /* use scream table. */
539#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 540#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 541
7fe44985 542#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
543# define ReREFCNT_inc(re) \
544 ({ \
545 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
546 REGEXP *const _rerefcnt_inc = (re); \
547 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
548 SvREFCNT_inc(_rerefcnt_inc); \
549 _rerefcnt_inc; \
288b8c02
NC
550 })
551# define ReREFCNT_dec(re) \
552 ({ \
553 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
554 REGEXP *const _rerefcnt_dec = (re); \
555 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
556 })
557#else
558# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 559# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
560#endif
561
562/* FIXME for plugins. */
cf93c79d
IZ
563
564#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
565#define FBMcf_TAIL_DOLLARM 2
566#define FBMcf_TAIL_Z 4
567#define FBMcf_TAIL_z 8
568#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
569
570#define FBMrf_MULTILINE 1
cad2e5aa 571
3b0527fe
DM
572/* some basic information about the current match that is created by
573 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
574
575typedef struct {
288b8c02 576 REGEXP *prog;
3b0527fe
DM
577 char *bol;
578 char *till;
579 SV *sv;
580 char *ganch;
24b23f37 581 char *cutpoint;
3b0527fe
DM
582} regmatch_info;
583
5d9a96ca
DM
584
585/* structures for holding and saving the state maintained by regmatch() */
586
d56b3014 587#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 588#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 589#endif
6bda09f9 590
5d9a96ca
DM
591typedef I32 CHECKPOINT;
592
a0374537 593typedef struct regmatch_state {
40a82448 594 int resume_state; /* where to jump to on return */
24d3c4a9 595 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
596
597 union {
77cb431f
DM
598
599 /* this is a fake union member that matches the first element
600 * of each member that needs to store positive backtrack
601 * information */
602 struct {
603 struct regmatch_state *prev_yes_state;
604 } yes;
605
fae667d5
YO
606 /* branchlike members */
607 /* this is a fake union member that matches the first elements
608 * of each member that needs to behave like a branch */
609 struct {
5d458dd8
YO
610 /* this first element must match u.yes */
611 struct regmatch_state *prev_yes_state;
fae667d5
YO
612 U32 lastparen;
613 CHECKPOINT cp;
614
615 } branchlike;
616
617 struct {
618 /* the first elements must match u.branchlike */
619 struct regmatch_state *prev_yes_state;
620 U32 lastparen;
621 CHECKPOINT cp;
622
623 regnode *next_branch; /* next branch node */
624 } branch;
625
626 struct {
627 /* the first elements must match u.branchlike */
628 struct regmatch_state *prev_yes_state;
629 U32 lastparen;
630 CHECKPOINT cp;
631
2e64971a 632 U32 accepted; /* how many accepting states left */
6603fe3e 633 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 634 U16 *jump; /* positive offsets from me */
7f69552c 635 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
636 U8 *firstpos;/* pos in string of first trie match */
637 U32 firstchars;/* len in chars of firstpos from start */
638 U16 nextword;/* next word to try */
639 U16 topword; /* longest accepted word */
e822a8b4
DM
640 } trie;
641
fae667d5
YO
642 /* special types - these members are used to store state for special
643 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 644 struct {
77cb431f
DM
645 /* this first element must match u.yes */
646 struct regmatch_state *prev_yes_state;
faec1544
DM
647 struct regmatch_state *prev_eval;
648 struct regmatch_state *prev_curlyx;
288b8c02 649 REGEXP *prev_rex;
faec1544
DM
650 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
651 flip when transitioning between
652 inner and outer rexen */
aa283a38
DM
653 CHECKPOINT cp; /* remember current savestack indexes */
654 CHECKPOINT lastcp;
6bda09f9 655 U32 close_paren; /* which close bracket is our end */
1b5c067c 656 regnode *B; /* the node following us */
e822a8b4
DM
657 } eval;
658
659 struct {
c476f425 660 /* this first element must match u.yes */
262b90c4 661 struct regmatch_state *prev_yes_state;
fae667d5
YO
662 I32 wanted;
663 I32 logical; /* saved copy of 'logical' var */
664 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
665 } ifmatch; /* and SUSPEND/UNLESSM */
666
667 struct {
668 /* this first element must match u.yes */
669 struct regmatch_state *prev_yes_state;
670 struct regmatch_state *prev_mark;
671 SV* mark_name;
672 char *mark_loc;
673 } mark;
674
675 struct {
676 int val;
677 } keeper;
678
679 /* quantifiers - these members are used for storing state for
680 for the regops used to implement quantifiers */
681 struct {
682 /* this first element must match u.yes */
683 struct regmatch_state *prev_yes_state;
c476f425 684 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
685 regnode *me; /* the CURLYX node */
686 regnode *B; /* the B node in /A*B/ */
c476f425
DM
687 CHECKPOINT cp; /* remember current savestack index */
688 bool minmod;
a0374537 689 int parenfloor;/* how far back to strip paren data */
c476f425
DM
690
691 /* these two are modified by WHILEM */
692 int count; /* how many instances of A we've matched */
693 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
694 } curlyx;
695
696 struct {
c476f425 697 /* this first element must match u.yes */
262b90c4 698 struct regmatch_state *prev_yes_state;
c476f425
DM
699 struct regmatch_state *save_curlyx;
700 CHECKPOINT cp; /* remember current savestack indexes */
701 CHECKPOINT lastcp;
702 char *save_lastloc; /* previous curlyx.lastloc */
703 I32 cache_offset;
704 I32 cache_mask;
e822a8b4
DM
705 } whilem;
706
707 struct {
5d458dd8
YO
708 /* this first element must match u.yes */
709 struct regmatch_state *prev_yes_state;
e822a8b4 710 I32 c1, c2; /* case fold search */
40a82448
DM
711 CHECKPOINT cp;
712 I32 alen; /* length of first-matched A string */
713 I32 count;
0cadcf80 714 bool minmod;
40a82448
DM
715 regnode *A, *B; /* the nodes corresponding to /A*B/ */
716 regnode *me; /* the curlym node */
e822a8b4
DM
717 } curlym;
718
719 struct {
3b6647e0 720 U32 paren;
c255a977 721 CHECKPOINT cp;
e822a8b4 722 I32 c1, c2; /* case fold search */
c255a977
DM
723 char *maxpos; /* highest possible point in string to match */
724 char *oldloc; /* the previous locinput */
e822a8b4 725 int count;
c255a977
DM
726 int min, max; /* {m,n} */
727 regnode *A, *B; /* the nodes corresponding to /A*B/ */
728 } curly; /* and CURLYN/PLUS/STAR */
dad79028 729
d8319b27 730 } u;
5d9a96ca
DM
731} regmatch_state;
732
733/* how many regmatch_state structs to allocate as a single slab.
734 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
735 * pointers, plus 1 for any mythical malloc overhead. */
736
737#define PERL_REGMATCH_SLAB_SLOTS \
738 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
739
740typedef struct regmatch_slab {
741 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
742 struct regmatch_slab *prev, *next;
743} regmatch_slab;
1ade1aa1 744
46ab3289
NC
745#define PL_reg_flags PL_reg_state.re_state_reg_flags
746#define PL_bostr PL_reg_state.re_state_bostr
747#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 748#define PL_regeol PL_reg_state.re_state_regeol
46ab3289
NC
749#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
750#define PL_reg_magic PL_reg_state.re_state_reg_magic
751#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
752#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
753#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
754#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
755#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
756#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
757#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
758#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
759#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
760#define PL_regsize PL_reg_state.re_state_regsize
761#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
762#define PL_nrs PL_reg_state.re_state_nrs
763
1ade1aa1
NC
764struct re_save_state {
765 U32 re_state_reg_flags; /* from regexec.c */
ed301438 766 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 767 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 768 bool re_reparsing; /* runtime (?{}) fed back into parser */
1ade1aa1
NC
769 char *re_state_bostr;
770 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 771 char *re_state_regeol; /* End of input, for $ check. */
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 */