This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Define RXf_SPLIT and RXf_SKIPWHITE as 0
[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 */ \
6502e081
DM
127 I32 suboffset; /* byte offset of subbeg from logical start of str */ \
128 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740
NC
129 /* Information about the match that isn't often used */ \
130 /* offset from wrapped to the start of precomp */ \
654eccd5 131 PERL_BITFIELD32 pre_prefix:4; \
b82ad30c 132 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 133
bbe252da 134typedef struct regexp {
288b8c02 135 _XPV_HEAD;
08e44740 136 _REGEXP_COMMON;
f8fc2ecf
YO
137} regexp;
138
89f6f287 139#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 140
785a26d5 141/* used for high speed searches */
f9f4320a
YO
142typedef struct re_scream_pos_data_s
143{
144 char **scream_olds; /* match pos */
145 I32 *scream_pos; /* Internal iterator of scream. */
146} re_scream_pos_data;
147
785a26d5
YO
148/* regexp_engine structure. This is the dispatch table for regexes.
149 * Any regex engine implementation must be able to build one of these.
150 */
f9f4320a 151typedef struct regexp_engine {
1593ad57 152 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
153 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
154 char* strbeg, I32 minend, SV* screamer,
155 void* data, U32 flags);
156 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
157 char *strend, const U32 flags,
158 re_scream_pos_data *data);
49d7dfbc
AB
159 SV* (*checkstr) (pTHX_ REGEXP * const rx);
160 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 161 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 162 SV * const sv);
2fdbfb4d
AB
163 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
164 SV const * const value);
165 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
166 const I32 paren);
192b9cd1
AB
167 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
168 SV * const value, const U32 flags);
169 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
170 const U32 flags);
49d7dfbc 171 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 172#ifdef USE_ITHREADS
49d7dfbc 173 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 174#endif
3c13cae6
DM
175 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
176 OP *expr, const struct regexp_engine* eng,
177 REGEXP *VOL old_re,
76ac488f 178 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
179} regexp_engine;
180
192b9cd1
AB
181/*
182 These are passed to the numbered capture variable callbacks as the
183 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
184 $2 etc.
185*/
2c7b5d76
DM
186#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
187#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
188#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
189#define RX_BUFF_IDX_PREMATCH -2 /* $` */
190#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
191#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
192
193/*
194 Flags that are passed to the named_buff and named_buff_iter
195 callbacks above. Those routines are called from universal.c via the
196 Tie::Hash::NamedCapture interface for %+ and %- and the re::
197 functions in the same file.
198*/
199
200/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
201#define RXapif_FETCH 0x0001
202#define RXapif_STORE 0x0002
203#define RXapif_DELETE 0x0004
204#define RXapif_CLEAR 0x0008
205#define RXapif_EXISTS 0x0010
206#define RXapif_SCALAR 0x0020
207#define RXapif_FIRSTKEY 0x0040
208#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
209
210/* Whether %+ or %- is being operated on */
f1b875a0
YO
211#define RXapif_ONE 0x0100 /* %+ */
212#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
213
214/* Whether this is being called from a re:: function */
f1b875a0
YO
215#define RXapif_REGNAME 0x0400
216#define RXapif_REGNAMES 0x0800
7948fc08 217#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 218
f7e71195
AB
219/*
220=head1 REGEXP Functions
221
222=for apidoc Am|REGEXP *|SvRX|SV *sv
223
224Convenience macro to get the REGEXP from a SV. This is approximately
225equivalent to the following snippet:
226
227 if (SvMAGICAL(sv))
228 mg_get(sv);
1af910ea
DL
229 if (SvROK(sv))
230 sv = MUTABLE_SV(SvRV(sv));
231 if (SvTYPE(sv) == SVt_REGEXP)
232 return (REGEXP*) sv;
f7e71195
AB
233
234NULL will be returned if a REGEXP* is not found.
235
236=for apidoc Am|bool|SvRXOK|SV* sv
237
1af910ea
DL
238Returns a boolean indicating whether the SV (or the one it references)
239is a REGEXP.
f7e71195
AB
240
241If you want to do something with the REGEXP* later use SvRX instead
242and check for NULL.
243
244=cut
245*/
246
247#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
248#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
249
250
7948fc08 251/* Flags stored in regexp->extflags
bbe252da 252 * These are used by code external to the regexp engine
e357fc67 253 *
5b126c84
KW
254 * Note that the flags whose names start with RXf_PMf_ are defined in
255 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 256 *
eb2624c9
FC
257 * NOTE: if you modify any RXf flags you should run regen.pl or
258 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 259 *
bbe252da
YO
260 */
261
5b126c84
KW
262#include "op_reg_common.h"
263
cde0cee5 264#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 265
a20207d7
YO
266#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
267 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
268 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
269 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
270 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 271
94b03d7d 272/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 273#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 274 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 275
bcdf7404 276/* chars and strings used as regex pattern modifiers
486ec47a 277 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
278 *
279 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
280 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
281 * for its own uses. So 'k' is out as well.
bcdf7404 282 */
85508812 283#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 284#define EXEC_PAT_MOD 'e'
87e95b7f 285#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
286#define ONCE_PAT_MOD 'o'
287#define GLOBAL_PAT_MOD 'g'
288#define CONTINUE_PAT_MOD 'c'
289#define MULTILINE_PAT_MOD 'm'
290#define SINGLE_PAT_MOD 's'
291#define IGNORE_PAT_MOD 'i'
292#define XTENDED_PAT_MOD 'x'
4f4d7508 293#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
294#define LOCALE_PAT_MOD 'l'
295#define UNICODE_PAT_MOD 'u'
50e91148 296#define DEPENDS_PAT_MOD 'd'
cfaf538b 297#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
298
299#define ONCE_PAT_MODS "o"
87e95b7f 300#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
301#define EXEC_PAT_MODS "e"
302#define LOOP_PAT_MODS "gc"
4f4d7508 303#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
304#define LOCALE_PAT_MODS "l"
305#define UNICODE_PAT_MODS "u"
50e91148 306#define DEPENDS_PAT_MODS "d"
cfaf538b 307#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 308#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 309
78fecfc8 310/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
311 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
312 * character is bit +1, etc. */
bcdf7404
YO
313#define STD_PAT_MODS "msix"
314
94b03d7d
KW
315#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
316
78fecfc8 317/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
318 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
319 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
320#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
321
322#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 323#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 324#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 325#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 326
cb5027f2 327/*
eb2624c9
FC
328 * NOTE: if you modify any RXf flags you should run regen.pl or
329 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
330 *
331 */
bcdf7404 332
e795e964
KW
333/* Leave some space, so future bit allocations can go either in the shared or
334 * unshared area without affecting binary compatibility */
df7a8460 335#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 336
bf28aba1
KW
337/* Manually decorate this function with gcc-style attributes just to
338 * avoid having to restructure the header files and their called order,
339 * as proto.h would have to be included before this file, and isn't */
340
a62b1201
KW
341PERL_STATIC_INLINE const char *
342get_regex_charset_name(const U32 flags, STRLEN* const lenp)
343 __attribute__warn_unused_result__;
344
df7a8460 345#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
346
347PERL_STATIC_INLINE const char *
348get_regex_charset_name(const U32 flags, STRLEN* const lenp)
349{
350 /* Returns a string that corresponds to the name of the regex character set
351 * given by 'flags', and *lenp is set the length of that string, which
352 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
353
354 *lenp = 1;
355 switch (get_regex_charset(flags)) {
50e91148 356 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
357 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
358 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 359 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 360 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
361 *lenp = 2;
362 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
363 default:
364 return "?"; /* Unknown */
a62b1201 365 }
a62b1201
KW
366}
367
cccd1425
FC
368/*
369 Two flags no longer used.
370 RXf_SPLIT used to be set in Perl_pmruntime if op_flags & OPf_SPECIAL,
371 i.e., split. It was used by the regex engine to check whether it should
372 set RXf_SKIPWHITE. Regexp plugins on CPAN also have done the same thing
373 historically, so we leave these flags defined.
374*/
375#ifndef PERL_CORE
376# define RXf_SPLIT 0
377# define RXf_SKIPWHITE 0
378#endif
379
380
52d81aa8 381/* Anchor and GPOS related stuff */
1ad4ec54
KW
382#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
383#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
384#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
385#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
386#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
387#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
388/* two bits here */
389#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
390#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
391#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
392
bbe252da 393/* What we have seen */
1ad4ec54
KW
394#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
395#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
396#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
397
398/* Special */
1ad4ec54
KW
399#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
400#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
401
402/* UTF8 related */
e795e964 403#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
404
405/* Intuit related */
e795e964
KW
406#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
407#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
408#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
409#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
410
411/* Copy and tainted info */
e795e964 412#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
413
414/* during execution: pattern temporarily tainted by executing locale ops;
415 * post-execution: $1 et al are tainted */
e795e964 416#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
417/* this pattern was tainted during compilation */
418#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
419
420/* Flags indicating special patterns */
e795e964 421#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
e795e964 422#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 423#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 424#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
425# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
426#endif
c737faaf 427
cb5027f2 428/*
eb2624c9
FC
429 * NOTE: if you modify any RXf flags you should run regen.pl or
430 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
431 *
432 */
bbe252da 433
bbe252da 434#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
435#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
436#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
437#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
438#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
439#define RX_MATCH_TAINTED_set(prog, t) ((t) \
440 ? RX_MATCH_TAINTED_on(prog) \
441 : RX_MATCH_TAINTED_off(prog))
c277df42 442
07bc277f
NC
443#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
444#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
445#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
446#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
447#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
448#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
449#define RX_MATCH_COPIED_set(prog,t) ((t) \
450 ? RX_MATCH_COPIED_on(prog) \
451 : RX_MATCH_COPIED_off(prog))
f9f4320a 452
07bc277f 453#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 454
07bc277f 455/* For source compatibility. We used to store these explicitly. */
f7c278bf 456#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 457#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
458/* FIXME? Are we hardcoding too much here and constraining plugin extension
459 writers? Specifically, the value 1 assumes that the wrapped version always
460 has exactly one character at the end, a ')'. Will that always be true? */
461#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 462#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 463#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 464#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 465#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 466#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 467#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
468# define RX_EXTFLAGS(prog) \
469 (*({ \
7948fc08
RU
470 const REGEXP *const _rx_extflags = (prog); \
471 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
472 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
473 }))
474# define RX_ENGINE(prog) \
475 (*({ \
7948fc08
RU
476 const REGEXP *const _rx_engine = (prog); \
477 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
478 &SvANY(_rx_engine)->engine; \
4450b630
NC
479 }))
480# define RX_SUBBEG(prog) \
481 (*({ \
7948fc08
RU
482 const REGEXP *const _rx_subbeg = (prog); \
483 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
484 &SvANY(_rx_subbeg)->subbeg; \
4450b630 485 }))
6502e081
DM
486# define RX_SUBOFFSET(prog) \
487 (*({ \
488 const REGEXP *const _rx_suboffset = (prog); \
489 assert(SvTYPE(_rx_suboffset) == SVt_REGEXP); \
490 &SvANY(_rx_suboffset)->suboffset; \
491 }))
492# define RX_SUBCOFFSET(prog) \
493 (*({ \
494 const REGEXP *const _rx_subcoffset = (prog); \
495 assert(SvTYPE(_rx_subcoffset) == SVt_REGEXP); \
496 &SvANY(_rx_subcoffset)->subcoffset; \
497 }))
200ff815 498# define RX_OFFS(prog) \
4450b630 499 (*({ \
7948fc08
RU
500 const REGEXP *const _rx_offs = (prog); \
501 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
502 &SvANY(_rx_offs)->offs; \
4450b630
NC
503 }))
504# define RX_NPARENS(prog) \
505 (*({ \
7948fc08
RU
506 const REGEXP *const _rx_nparens = (prog); \
507 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
508 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
509 }))
510#else
511# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
512# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
513# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
6502e081
DM
514# define RX_SUBOFFSET(prog) (((struct regexp *)SvANY(prog))->suboffset)
515# define RX_SUBCOFFSET(prog) (((struct regexp *)SvANY(prog))->subcoffset)
4450b630
NC
516# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
517# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
518#endif
288b8c02 519#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
520#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
521#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
522#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
523#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
524#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
bdd9a1b1 525#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 526
be8e71aa
YO
527#endif /* PLUGGABLE_RE_EXTENSION */
528
486ec47a 529/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 530
f8c7b90f 531#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 532#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
533 STMT_START {if (RX_SAVED_COPY(rx)) { \
534 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
535 } \
536 if (RX_MATCH_COPIED(rx)) { \
07bc277f 537 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
538 RX_MATCH_COPIED_off(rx); \
539 }} STMT_END
540#else
541#define RX_MATCH_COPY_FREE(rx) \
542 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 543 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
544 RX_MATCH_COPIED_off(rx); \
545 }} STMT_END
546#endif
547
07bc277f
NC
548#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
549#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
550#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
551#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
552#define RX_MATCH_UTF8_set(prog, t) ((t) \
553 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
554 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
555
556/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 557#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 558
cad2e5aa
JH
559#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
560#define REXEC_CHECKED 0x02 /* check_substr already checked. */
561#define REXEC_SCREAM 0x04 /* use scream table. */
562#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 563#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
6502e081
DM
564 /* under REXEC_COPY_STR, it's ok for the
565 * engine (modulo PL_sawamperand etc)
566 * to skip copying ... */
567#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
568#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
c277df42 569
7fe44985 570#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
571# define ReREFCNT_inc(re) \
572 ({ \
573 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
574 REGEXP *const _rerefcnt_inc = (re); \
575 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
576 SvREFCNT_inc(_rerefcnt_inc); \
577 _rerefcnt_inc; \
288b8c02
NC
578 })
579# define ReREFCNT_dec(re) \
580 ({ \
581 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
582 REGEXP *const _rerefcnt_dec = (re); \
583 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
584 })
585#else
586# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 587# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
588#endif
589
590/* FIXME for plugins. */
cf93c79d
IZ
591
592#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
593#define FBMcf_TAIL_DOLLARM 2
594#define FBMcf_TAIL_Z 4
595#define FBMcf_TAIL_z 8
596#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
597
598#define FBMrf_MULTILINE 1
cad2e5aa 599
3b0527fe
DM
600/* some basic information about the current match that is created by
601 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
602
603typedef struct {
288b8c02 604 REGEXP *prog;
3b0527fe
DM
605 char *bol;
606 char *till;
607 SV *sv;
608 char *ganch;
24b23f37 609 char *cutpoint;
3b0527fe
DM
610} regmatch_info;
611
5d9a96ca
DM
612
613/* structures for holding and saving the state maintained by regmatch() */
614
d56b3014 615#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 616#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 617#endif
6bda09f9 618
5d9a96ca
DM
619typedef I32 CHECKPOINT;
620
a0374537 621typedef struct regmatch_state {
40a82448 622 int resume_state; /* where to jump to on return */
24d3c4a9 623 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
624
625 union {
77cb431f
DM
626
627 /* this is a fake union member that matches the first element
628 * of each member that needs to store positive backtrack
629 * information */
630 struct {
631 struct regmatch_state *prev_yes_state;
632 } yes;
633
fae667d5
YO
634 /* branchlike members */
635 /* this is a fake union member that matches the first elements
636 * of each member that needs to behave like a branch */
637 struct {
5d458dd8
YO
638 /* this first element must match u.yes */
639 struct regmatch_state *prev_yes_state;
fae667d5 640 U32 lastparen;
f6033a9d 641 U32 lastcloseparen;
fae667d5
YO
642 CHECKPOINT cp;
643
644 } branchlike;
645
646 struct {
647 /* the first elements must match u.branchlike */
648 struct regmatch_state *prev_yes_state;
649 U32 lastparen;
f6033a9d 650 U32 lastcloseparen;
fae667d5
YO
651 CHECKPOINT cp;
652
653 regnode *next_branch; /* next branch node */
654 } branch;
655
656 struct {
657 /* the first elements must match u.branchlike */
658 struct regmatch_state *prev_yes_state;
659 U32 lastparen;
f6033a9d 660 U32 lastcloseparen;
fae667d5
YO
661 CHECKPOINT cp;
662
2e64971a 663 U32 accepted; /* how many accepting states left */
6603fe3e 664 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 665 U16 *jump; /* positive offsets from me */
7f69552c 666 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
667 U8 *firstpos;/* pos in string of first trie match */
668 U32 firstchars;/* len in chars of firstpos from start */
669 U16 nextword;/* next word to try */
670 U16 topword; /* longest accepted word */
e822a8b4
DM
671 } trie;
672
fae667d5
YO
673 /* special types - these members are used to store state for special
674 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 675 struct {
77cb431f
DM
676 /* this first element must match u.yes */
677 struct regmatch_state *prev_yes_state;
faec1544
DM
678 struct regmatch_state *prev_eval;
679 struct regmatch_state *prev_curlyx;
288b8c02 680 REGEXP *prev_rex;
faec1544
DM
681 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
682 flip when transitioning between
683 inner and outer rexen */
aa283a38
DM
684 CHECKPOINT cp; /* remember current savestack indexes */
685 CHECKPOINT lastcp;
6bda09f9 686 U32 close_paren; /* which close bracket is our end */
1b5c067c 687 regnode *B; /* the node following us */
e822a8b4
DM
688 } eval;
689
690 struct {
c476f425 691 /* this first element must match u.yes */
262b90c4 692 struct regmatch_state *prev_yes_state;
fae667d5
YO
693 I32 wanted;
694 I32 logical; /* saved copy of 'logical' var */
695 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
696 } ifmatch; /* and SUSPEND/UNLESSM */
697
698 struct {
699 /* this first element must match u.yes */
700 struct regmatch_state *prev_yes_state;
701 struct regmatch_state *prev_mark;
702 SV* mark_name;
703 char *mark_loc;
704 } mark;
705
706 struct {
707 int val;
708 } keeper;
709
710 /* quantifiers - these members are used for storing state for
711 for the regops used to implement quantifiers */
712 struct {
713 /* this first element must match u.yes */
714 struct regmatch_state *prev_yes_state;
c476f425 715 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
716 regnode *me; /* the CURLYX node */
717 regnode *B; /* the B node in /A*B/ */
c476f425
DM
718 CHECKPOINT cp; /* remember current savestack index */
719 bool minmod;
a0374537 720 int parenfloor;/* how far back to strip paren data */
c476f425
DM
721
722 /* these two are modified by WHILEM */
723 int count; /* how many instances of A we've matched */
724 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
725 } curlyx;
726
727 struct {
c476f425 728 /* this first element must match u.yes */
262b90c4 729 struct regmatch_state *prev_yes_state;
c476f425
DM
730 struct regmatch_state *save_curlyx;
731 CHECKPOINT cp; /* remember current savestack indexes */
732 CHECKPOINT lastcp;
733 char *save_lastloc; /* previous curlyx.lastloc */
734 I32 cache_offset;
735 I32 cache_mask;
e822a8b4
DM
736 } whilem;
737
738 struct {
5d458dd8
YO
739 /* this first element must match u.yes */
740 struct regmatch_state *prev_yes_state;
e822a8b4 741 I32 c1, c2; /* case fold search */
40a82448 742 CHECKPOINT cp;
f6033a9d
DM
743 U32 lastparen;
744 U32 lastcloseparen;
40a82448
DM
745 I32 alen; /* length of first-matched A string */
746 I32 count;
0cadcf80 747 bool minmod;
40a82448
DM
748 regnode *A, *B; /* the nodes corresponding to /A*B/ */
749 regnode *me; /* the curlym node */
e822a8b4
DM
750 } curlym;
751
752 struct {
3b6647e0 753 U32 paren;
c255a977 754 CHECKPOINT cp;
f6033a9d
DM
755 U32 lastparen;
756 U32 lastcloseparen;
e822a8b4 757 I32 c1, c2; /* case fold search */
c255a977
DM
758 char *maxpos; /* highest possible point in string to match */
759 char *oldloc; /* the previous locinput */
e822a8b4 760 int count;
c255a977
DM
761 int min, max; /* {m,n} */
762 regnode *A, *B; /* the nodes corresponding to /A*B/ */
763 } curly; /* and CURLYN/PLUS/STAR */
dad79028 764
d8319b27 765 } u;
5d9a96ca
DM
766} regmatch_state;
767
768/* how many regmatch_state structs to allocate as a single slab.
769 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
770 * pointers, plus 1 for any mythical malloc overhead. */
771
772#define PERL_REGMATCH_SLAB_SLOTS \
773 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
774
775typedef struct regmatch_slab {
776 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
777 struct regmatch_slab *prev, *next;
778} regmatch_slab;
1ade1aa1 779
46ab3289
NC
780#define PL_reg_flags PL_reg_state.re_state_reg_flags
781#define PL_bostr PL_reg_state.re_state_bostr
46ab3289 782#define PL_regeol PL_reg_state.re_state_regeol
46ab3289
NC
783#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
784#define PL_reg_magic PL_reg_state.re_state_reg_magic
785#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
786#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
787#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
788#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
789#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
6502e081
DM
790#define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset
791#define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset
46ab3289
NC
792#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
793#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
794#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
795#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
796#define PL_regsize PL_reg_state.re_state_regsize
797#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
798#define PL_nrs PL_reg_state.re_state_nrs
799
1ade1aa1
NC
800struct re_save_state {
801 U32 re_state_reg_flags; /* from regexec.c */
ed301438 802 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 803 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 804 bool re_reparsing; /* runtime (?{}) fed back into parser */
1ade1aa1 805 char *re_state_bostr;
1ade1aa1 806 char *re_state_regeol; /* End of input, for $ check. */
1ade1aa1 807 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
808 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
809 PMOP *re_state_reg_curpm; /* from regexec.c */
810 char *re_state_reg_oldsaved; /* old saved substr during match */
811 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
6502e081
DM
812 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */
813 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */
1b5c067c
NC
814 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
815 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
816 I32 re_state_reg_maxiter; /* max wait until caching pos */
817 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 818 U32 re_state_regsize; /* from regexec.c */
1b5c067c 819 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
820 char *re_state_reg_starttry; /* from regexec.c */
821#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 822 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
823#endif
824};
825
826#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
827 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 828
1ade1aa1
NC
829/*
830 * Local variables:
831 * c-indentation-style: bsd
832 * c-basic-offset: 4
14d04a33 833 * indent-tabs-mode: nil
1ade1aa1
NC
834 * End:
835 *
14d04a33 836 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 837 */