This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Separate handling of ${^PREMATCH} from $` etc
[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*/
2c7b5d76
DM
184#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
185#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
186#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
187#define RX_BUFF_IDX_PREMATCH -2 /* $` */
188#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
189#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
190
191/*
192 Flags that are passed to the named_buff and named_buff_iter
193 callbacks above. Those routines are called from universal.c via the
194 Tie::Hash::NamedCapture interface for %+ and %- and the re::
195 functions in the same file.
196*/
197
198/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
199#define RXapif_FETCH 0x0001
200#define RXapif_STORE 0x0002
201#define RXapif_DELETE 0x0004
202#define RXapif_CLEAR 0x0008
203#define RXapif_EXISTS 0x0010
204#define RXapif_SCALAR 0x0020
205#define RXapif_FIRSTKEY 0x0040
206#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
207
208/* Whether %+ or %- is being operated on */
f1b875a0
YO
209#define RXapif_ONE 0x0100 /* %+ */
210#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
211
212/* Whether this is being called from a re:: function */
f1b875a0
YO
213#define RXapif_REGNAME 0x0400
214#define RXapif_REGNAMES 0x0800
7948fc08 215#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 216
f7e71195
AB
217/*
218=head1 REGEXP Functions
219
220=for apidoc Am|REGEXP *|SvRX|SV *sv
221
222Convenience macro to get the REGEXP from a SV. This is approximately
223equivalent to the following snippet:
224
225 if (SvMAGICAL(sv))
226 mg_get(sv);
1af910ea
DL
227 if (SvROK(sv))
228 sv = MUTABLE_SV(SvRV(sv));
229 if (SvTYPE(sv) == SVt_REGEXP)
230 return (REGEXP*) sv;
f7e71195
AB
231
232NULL will be returned if a REGEXP* is not found.
233
234=for apidoc Am|bool|SvRXOK|SV* sv
235
1af910ea
DL
236Returns a boolean indicating whether the SV (or the one it references)
237is a REGEXP.
f7e71195
AB
238
239If you want to do something with the REGEXP* later use SvRX instead
240and check for NULL.
241
242=cut
243*/
244
245#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
246#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
247
248
7948fc08 249/* Flags stored in regexp->extflags
bbe252da 250 * These are used by code external to the regexp engine
e357fc67 251 *
5b126c84
KW
252 * Note that the flags whose names start with RXf_PMf_ are defined in
253 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 254 *
eb2624c9
FC
255 * NOTE: if you modify any RXf flags you should run regen.pl or
256 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 257 *
bbe252da
YO
258 */
259
5b126c84
KW
260#include "op_reg_common.h"
261
cde0cee5 262#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 263
a20207d7
YO
264#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
265 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
266 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
267 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
268 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 269
94b03d7d 270/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 271#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 272 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 273
bcdf7404 274/* chars and strings used as regex pattern modifiers
486ec47a 275 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
276 *
277 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
278 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
279 * for its own uses. So 'k' is out as well.
bcdf7404 280 */
85508812 281#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 282#define EXEC_PAT_MOD 'e'
87e95b7f 283#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
284#define ONCE_PAT_MOD 'o'
285#define GLOBAL_PAT_MOD 'g'
286#define CONTINUE_PAT_MOD 'c'
287#define MULTILINE_PAT_MOD 'm'
288#define SINGLE_PAT_MOD 's'
289#define IGNORE_PAT_MOD 'i'
290#define XTENDED_PAT_MOD 'x'
4f4d7508 291#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
292#define LOCALE_PAT_MOD 'l'
293#define UNICODE_PAT_MOD 'u'
50e91148 294#define DEPENDS_PAT_MOD 'd'
cfaf538b 295#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
296
297#define ONCE_PAT_MODS "o"
87e95b7f 298#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
299#define EXEC_PAT_MODS "e"
300#define LOOP_PAT_MODS "gc"
4f4d7508 301#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
302#define LOCALE_PAT_MODS "l"
303#define UNICODE_PAT_MODS "u"
50e91148 304#define DEPENDS_PAT_MODS "d"
cfaf538b 305#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 306#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 307
78fecfc8 308/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
309 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
310 * character is bit +1, etc. */
bcdf7404
YO
311#define STD_PAT_MODS "msix"
312
94b03d7d
KW
313#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
314
78fecfc8 315/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
316 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
317 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
318#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
319
320#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 321#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 322#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 323#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 324
cb5027f2 325/*
eb2624c9
FC
326 * NOTE: if you modify any RXf flags you should run regen.pl or
327 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
328 *
329 */
bcdf7404 330
e795e964
KW
331/* Leave some space, so future bit allocations can go either in the shared or
332 * unshared area without affecting binary compatibility */
df7a8460 333#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 334
bf28aba1
KW
335/* Manually decorate this function with gcc-style attributes just to
336 * avoid having to restructure the header files and their called order,
337 * as proto.h would have to be included before this file, and isn't */
338
a62b1201
KW
339PERL_STATIC_INLINE const char *
340get_regex_charset_name(const U32 flags, STRLEN* const lenp)
341 __attribute__warn_unused_result__;
342
df7a8460 343#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
344
345PERL_STATIC_INLINE const char *
346get_regex_charset_name(const U32 flags, STRLEN* const lenp)
347{
348 /* Returns a string that corresponds to the name of the regex character set
349 * given by 'flags', and *lenp is set the length of that string, which
350 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
351
352 *lenp = 1;
353 switch (get_regex_charset(flags)) {
50e91148 354 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
355 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
356 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 357 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 358 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
359 *lenp = 2;
360 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
361 default:
362 return "?"; /* Unknown */
a62b1201 363 }
a62b1201
KW
364}
365
52d81aa8 366/* Anchor and GPOS related stuff */
1ad4ec54
KW
367#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
368#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
369#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
370#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
371#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
372#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
373/* two bits here */
374#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
375#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
376#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
377
bbe252da 378/* What we have seen */
1ad4ec54
KW
379#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
380#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
381#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
382
383/* Special */
1ad4ec54
KW
384#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
385#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
386
387/* UTF8 related */
e795e964 388#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
389
390/* Intuit related */
e795e964
KW
391#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
392#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
393#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
394
395/*
396 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
397 be used by regex engines to check whether they should set
398 RXf_SKIPWHITE
399*/
e795e964 400#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 401
bbe252da
YO
402#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
403
404/* Copy and tainted info */
e795e964 405#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
406
407/* during execution: pattern temporarily tainted by executing locale ops;
408 * post-execution: $1 et al are tainted */
e795e964 409#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
410/* this pattern was tainted during compilation */
411#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
412
413/* Flags indicating special patterns */
e795e964
KW
414#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
415#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
416#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 417#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 418#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
419# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
420#endif
c737faaf 421
cb5027f2 422/*
eb2624c9
FC
423 * NOTE: if you modify any RXf flags you should run regen.pl or
424 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
425 *
426 */
bbe252da 427
bbe252da 428#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
429#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
430#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
431#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
432#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
433#define RX_MATCH_TAINTED_set(prog, t) ((t) \
434 ? RX_MATCH_TAINTED_on(prog) \
435 : RX_MATCH_TAINTED_off(prog))
c277df42 436
07bc277f
NC
437#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
438#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
439#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
440#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
441#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
442#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
443#define RX_MATCH_COPIED_set(prog,t) ((t) \
444 ? RX_MATCH_COPIED_on(prog) \
445 : RX_MATCH_COPIED_off(prog))
f9f4320a 446
07bc277f 447#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 448
07bc277f 449/* For source compatibility. We used to store these explicitly. */
f7c278bf 450#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 451#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
452/* FIXME? Are we hardcoding too much here and constraining plugin extension
453 writers? Specifically, the value 1 assumes that the wrapped version always
454 has exactly one character at the end, a ')'. Will that always be true? */
455#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 456#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 457#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 458#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 459#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 460#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 461#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
462# define RX_EXTFLAGS(prog) \
463 (*({ \
7948fc08
RU
464 const REGEXP *const _rx_extflags = (prog); \
465 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
466 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
467 }))
468# define RX_ENGINE(prog) \
469 (*({ \
7948fc08
RU
470 const REGEXP *const _rx_engine = (prog); \
471 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
472 &SvANY(_rx_engine)->engine; \
4450b630
NC
473 }))
474# define RX_SUBBEG(prog) \
475 (*({ \
7948fc08
RU
476 const REGEXP *const _rx_subbeg = (prog); \
477 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
478 &SvANY(_rx_subbeg)->subbeg; \
4450b630 479 }))
200ff815 480# define RX_OFFS(prog) \
4450b630 481 (*({ \
7948fc08
RU
482 const REGEXP *const _rx_offs = (prog); \
483 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
484 &SvANY(_rx_offs)->offs; \
4450b630
NC
485 }))
486# define RX_NPARENS(prog) \
487 (*({ \
7948fc08
RU
488 const REGEXP *const _rx_nparens = (prog); \
489 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
490 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
491 }))
492#else
493# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
494# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
495# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
496# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
497# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
498#endif
288b8c02 499#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
500#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
501#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
502#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
503#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
504#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
bdd9a1b1 505#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 506
be8e71aa
YO
507#endif /* PLUGGABLE_RE_EXTENSION */
508
486ec47a 509/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 510
f8c7b90f 511#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 512#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
513 STMT_START {if (RX_SAVED_COPY(rx)) { \
514 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
515 } \
516 if (RX_MATCH_COPIED(rx)) { \
07bc277f 517 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
518 RX_MATCH_COPIED_off(rx); \
519 }} STMT_END
520#else
521#define RX_MATCH_COPY_FREE(rx) \
522 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 523 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
524 RX_MATCH_COPIED_off(rx); \
525 }} STMT_END
526#endif
527
07bc277f
NC
528#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
529#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
530#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
531#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
532#define RX_MATCH_UTF8_set(prog, t) ((t) \
533 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
534 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
535
536/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 537#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 538
cad2e5aa
JH
539#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
540#define REXEC_CHECKED 0x02 /* check_substr already checked. */
541#define REXEC_SCREAM 0x04 /* use scream table. */
542#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 543#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 544
7fe44985 545#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
546# define ReREFCNT_inc(re) \
547 ({ \
548 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
549 REGEXP *const _rerefcnt_inc = (re); \
550 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
551 SvREFCNT_inc(_rerefcnt_inc); \
552 _rerefcnt_inc; \
288b8c02
NC
553 })
554# define ReREFCNT_dec(re) \
555 ({ \
556 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
557 REGEXP *const _rerefcnt_dec = (re); \
558 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
559 })
560#else
561# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 562# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
563#endif
564
565/* FIXME for plugins. */
cf93c79d
IZ
566
567#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
568#define FBMcf_TAIL_DOLLARM 2
569#define FBMcf_TAIL_Z 4
570#define FBMcf_TAIL_z 8
571#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
572
573#define FBMrf_MULTILINE 1
cad2e5aa 574
3b0527fe
DM
575/* some basic information about the current match that is created by
576 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
577
578typedef struct {
288b8c02 579 REGEXP *prog;
3b0527fe
DM
580 char *bol;
581 char *till;
582 SV *sv;
583 char *ganch;
24b23f37 584 char *cutpoint;
3b0527fe
DM
585} regmatch_info;
586
5d9a96ca
DM
587
588/* structures for holding and saving the state maintained by regmatch() */
589
d56b3014 590#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 591#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 592#endif
6bda09f9 593
5d9a96ca
DM
594typedef I32 CHECKPOINT;
595
a0374537 596typedef struct regmatch_state {
40a82448 597 int resume_state; /* where to jump to on return */
24d3c4a9 598 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
599
600 union {
77cb431f
DM
601
602 /* this is a fake union member that matches the first element
603 * of each member that needs to store positive backtrack
604 * information */
605 struct {
606 struct regmatch_state *prev_yes_state;
607 } yes;
608
fae667d5
YO
609 /* branchlike members */
610 /* this is a fake union member that matches the first elements
611 * of each member that needs to behave like a branch */
612 struct {
5d458dd8
YO
613 /* this first element must match u.yes */
614 struct regmatch_state *prev_yes_state;
fae667d5 615 U32 lastparen;
f6033a9d 616 U32 lastcloseparen;
fae667d5
YO
617 CHECKPOINT cp;
618
619 } branchlike;
620
621 struct {
622 /* the first elements must match u.branchlike */
623 struct regmatch_state *prev_yes_state;
624 U32 lastparen;
f6033a9d 625 U32 lastcloseparen;
fae667d5
YO
626 CHECKPOINT cp;
627
628 regnode *next_branch; /* next branch node */
629 } branch;
630
631 struct {
632 /* the first elements must match u.branchlike */
633 struct regmatch_state *prev_yes_state;
634 U32 lastparen;
f6033a9d 635 U32 lastcloseparen;
fae667d5
YO
636 CHECKPOINT cp;
637
2e64971a 638 U32 accepted; /* how many accepting states left */
6603fe3e 639 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 640 U16 *jump; /* positive offsets from me */
7f69552c 641 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
642 U8 *firstpos;/* pos in string of first trie match */
643 U32 firstchars;/* len in chars of firstpos from start */
644 U16 nextword;/* next word to try */
645 U16 topword; /* longest accepted word */
e822a8b4
DM
646 } trie;
647
fae667d5
YO
648 /* special types - these members are used to store state for special
649 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 650 struct {
77cb431f
DM
651 /* this first element must match u.yes */
652 struct regmatch_state *prev_yes_state;
faec1544
DM
653 struct regmatch_state *prev_eval;
654 struct regmatch_state *prev_curlyx;
288b8c02 655 REGEXP *prev_rex;
faec1544
DM
656 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
657 flip when transitioning between
658 inner and outer rexen */
aa283a38
DM
659 CHECKPOINT cp; /* remember current savestack indexes */
660 CHECKPOINT lastcp;
6bda09f9 661 U32 close_paren; /* which close bracket is our end */
1b5c067c 662 regnode *B; /* the node following us */
e822a8b4
DM
663 } eval;
664
665 struct {
c476f425 666 /* this first element must match u.yes */
262b90c4 667 struct regmatch_state *prev_yes_state;
fae667d5
YO
668 I32 wanted;
669 I32 logical; /* saved copy of 'logical' var */
670 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
671 } ifmatch; /* and SUSPEND/UNLESSM */
672
673 struct {
674 /* this first element must match u.yes */
675 struct regmatch_state *prev_yes_state;
676 struct regmatch_state *prev_mark;
677 SV* mark_name;
678 char *mark_loc;
679 } mark;
680
681 struct {
682 int val;
683 } keeper;
684
685 /* quantifiers - these members are used for storing state for
686 for the regops used to implement quantifiers */
687 struct {
688 /* this first element must match u.yes */
689 struct regmatch_state *prev_yes_state;
c476f425 690 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
691 regnode *me; /* the CURLYX node */
692 regnode *B; /* the B node in /A*B/ */
c476f425
DM
693 CHECKPOINT cp; /* remember current savestack index */
694 bool minmod;
a0374537 695 int parenfloor;/* how far back to strip paren data */
c476f425
DM
696
697 /* these two are modified by WHILEM */
698 int count; /* how many instances of A we've matched */
699 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
700 } curlyx;
701
702 struct {
c476f425 703 /* this first element must match u.yes */
262b90c4 704 struct regmatch_state *prev_yes_state;
c476f425
DM
705 struct regmatch_state *save_curlyx;
706 CHECKPOINT cp; /* remember current savestack indexes */
707 CHECKPOINT lastcp;
708 char *save_lastloc; /* previous curlyx.lastloc */
709 I32 cache_offset;
710 I32 cache_mask;
e822a8b4
DM
711 } whilem;
712
713 struct {
5d458dd8
YO
714 /* this first element must match u.yes */
715 struct regmatch_state *prev_yes_state;
e822a8b4 716 I32 c1, c2; /* case fold search */
40a82448 717 CHECKPOINT cp;
f6033a9d
DM
718 U32 lastparen;
719 U32 lastcloseparen;
40a82448
DM
720 I32 alen; /* length of first-matched A string */
721 I32 count;
0cadcf80 722 bool minmod;
40a82448
DM
723 regnode *A, *B; /* the nodes corresponding to /A*B/ */
724 regnode *me; /* the curlym node */
e822a8b4
DM
725 } curlym;
726
727 struct {
3b6647e0 728 U32 paren;
c255a977 729 CHECKPOINT cp;
f6033a9d
DM
730 U32 lastparen;
731 U32 lastcloseparen;
e822a8b4 732 I32 c1, c2; /* case fold search */
c255a977
DM
733 char *maxpos; /* highest possible point in string to match */
734 char *oldloc; /* the previous locinput */
e822a8b4 735 int count;
c255a977
DM
736 int min, max; /* {m,n} */
737 regnode *A, *B; /* the nodes corresponding to /A*B/ */
738 } curly; /* and CURLYN/PLUS/STAR */
dad79028 739
d8319b27 740 } u;
5d9a96ca
DM
741} regmatch_state;
742
743/* how many regmatch_state structs to allocate as a single slab.
744 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
745 * pointers, plus 1 for any mythical malloc overhead. */
746
747#define PERL_REGMATCH_SLAB_SLOTS \
748 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
749
750typedef struct regmatch_slab {
751 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
752 struct regmatch_slab *prev, *next;
753} regmatch_slab;
1ade1aa1 754
46ab3289
NC
755#define PL_reg_flags PL_reg_state.re_state_reg_flags
756#define PL_bostr PL_reg_state.re_state_bostr
757#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 758#define PL_regeol PL_reg_state.re_state_regeol
46ab3289
NC
759#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
760#define PL_reg_magic PL_reg_state.re_state_reg_magic
761#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
762#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
763#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
764#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
765#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
766#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
767#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
768#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
769#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
770#define PL_regsize PL_reg_state.re_state_regsize
771#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
772#define PL_nrs PL_reg_state.re_state_nrs
773
1ade1aa1
NC
774struct re_save_state {
775 U32 re_state_reg_flags; /* from regexec.c */
ed301438 776 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 777 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 778 bool re_reparsing; /* runtime (?{}) fed back into parser */
1ade1aa1
NC
779 char *re_state_bostr;
780 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 781 char *re_state_regeol; /* End of input, for $ check. */
1ade1aa1 782 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
783 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
784 PMOP *re_state_reg_curpm; /* from regexec.c */
785 char *re_state_reg_oldsaved; /* old saved substr during match */
786 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
787 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
788 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
789 I32 re_state_reg_maxiter; /* max wait until caching pos */
790 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 791 U32 re_state_regsize; /* from regexec.c */
1b5c067c 792 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
793 char *re_state_reg_starttry; /* from regexec.c */
794#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 795 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
796#endif
797};
798
799#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
800 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 801
1ade1aa1
NC
802/*
803 * Local variables:
804 * c-indentation-style: bsd
805 * c-basic-offset: 4
14d04a33 806 * indent-tabs-mode: nil
1ade1aa1
NC
807 * End:
808 *
14d04a33 809 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 810 */