This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
make OP_REGCRESET only for taint handling
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
1129b882 4 * 2005, 2006, 2007, 2008 by Larry Wall and others
d6376244
JH
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
a0d0e21e
LW
9 */
10
378cc40b
LW
11/*
12 * Definitions etc. for regexp(3) routines.
13 *
14 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
15 * not the System V one.
16 */
be8e71aa 17#ifndef PLUGGABLE_RE_EXTENSION
785a26d5
YO
18/* we don't want to include this stuff if we are inside of
19 an external regex engine based on the core one - like re 'debug'*/
378cc40b 20
c277df42
IZ
21struct regnode {
22 U8 flags;
23 U8 type;
24 U16 next_off;
25};
26
27typedef struct regnode regnode;
28
cad2e5aa 29struct reg_substr_data;
2779dcf1 30
0ee3c8d0
JH
31struct reg_data;
32
f9f4320a 33struct regexp_engine;
28d8d7f4 34struct regexp;
bbe252da 35
785a26d5
YO
36struct reg_substr_datum {
37 I32 min_offset;
38 I32 max_offset;
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
41 I32 end_shift;
42};
43struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
45};
f9f4320a 46
f8c7b90f 47#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
48#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
49#else
50#define SV_SAVED_COPY
ed252734 51#endif
28d8d7f4 52
f0ab9afb
NC
53typedef struct regexp_paren_pair {
54 I32 start;
55 I32 end;
56} regexp_paren_pair;
28d8d7f4 57
52ae8f7e 58#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 59#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 60#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
61
62/* Subtracting b from a leaves in a everything that was there that isn't in b,
63 * that is the intersection of a with b's complement */
64#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
65#endif
66
3d2bd50a
DM
67/* record the position of a (?{...}) within a pattern */
68
69struct reg_code_block {
70 STRLEN start;
71 STRLEN end;
72 OP *block;
b30fcab9 73 REGEXP *src_regex;
3d2bd50a
DM
74};
75
76
882227b7
AB
77/*
78 The regexp/REGEXP struct, see L<perlreapi> for further documentation
79 on the individual fields. The struct is ordered so that the most
80 commonly used fields are placed at the start.
81
82 Any patch that adds items to this struct will need to include
83 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
84 (C<pregfree()>). This involves freeing or cloning items in the
85 regexp's data array based on the data item's type.
86*/
87
08e44740
NC
88#define _REGEXP_COMMON \
89 /* what engine created this regexp? */ \
90 const struct regexp_engine* engine; \
91 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 92 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
93 /* Information about the match that the perl core uses to */ \
94 /* manage things */ \
95 U32 extflags; /* Flags used both externally and internally */ \
96 I32 minlen; /* mininum possible length of string to match */\
97 I32 minlenret; /* mininum possible length of $& */ \
98 U32 gofs; /* chars left of pos that we search from */ \
99 /* substring data about strings that must appear in the */ \
100 /* final match, used for optimisations */ \
101 struct reg_substr_data *substrs; \
102 U32 nparens; /* number of capture buffers */ \
103 /* private engine specific data */ \
104 U32 intflags; /* Engine Specific Internal flags */ \
105 void *pprivate; /* Data private to the regex engine which */ \
106 /* created this object. */ \
107 /* Data about the last/current match. These are modified */ \
108 /* during matching */ \
109 U32 lastparen; /* last open paren matched */ \
110 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 111 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
112 /* Array of offsets for (@-) and (@+) */ \
113 regexp_paren_pair *offs; \
114 /* saved or original string so \digit works forever. */ \
115 char *subbeg; \
116 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
117 I32 sublen; /* Length of string pointed by subbeg */ \
118 /* Information about the match that isn't often used */ \
119 /* offset from wrapped to the start of precomp */ \
654eccd5 120 PERL_BITFIELD32 pre_prefix:4; \
b82ad30c 121 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 122
bbe252da 123typedef struct regexp {
288b8c02 124 _XPV_HEAD;
08e44740 125 _REGEXP_COMMON;
f8fc2ecf
YO
126} regexp;
127
89f6f287 128#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 129
785a26d5 130/* used for high speed searches */
f9f4320a
YO
131typedef struct re_scream_pos_data_s
132{
133 char **scream_olds; /* match pos */
134 I32 *scream_pos; /* Internal iterator of scream. */
135} re_scream_pos_data;
136
785a26d5
YO
137/* regexp_engine structure. This is the dispatch table for regexes.
138 * Any regex engine implementation must be able to build one of these.
139 */
f9f4320a 140typedef struct regexp_engine {
1593ad57 141 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
142 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
143 char* strbeg, I32 minend, SV* screamer,
144 void* data, U32 flags);
145 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
146 char *strend, const U32 flags,
147 re_scream_pos_data *data);
49d7dfbc
AB
148 SV* (*checkstr) (pTHX_ REGEXP * const rx);
149 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 150 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 151 SV * const sv);
2fdbfb4d
AB
152 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
153 SV const * const value);
154 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
155 const I32 paren);
192b9cd1
AB
156 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
157 SV * const value, const U32 flags);
158 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
159 const U32 flags);
49d7dfbc 160 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 161#ifdef USE_ITHREADS
49d7dfbc 162 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 163#endif
3c13cae6
DM
164 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
165 OP *expr, const struct regexp_engine* eng,
166 REGEXP *VOL old_re,
167 int *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
168} regexp_engine;
169
192b9cd1
AB
170/*
171 These are passed to the numbered capture variable callbacks as the
172 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
173 $2 etc.
174*/
f1b875a0
YO
175#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
176#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
177#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
178
179/*
180 Flags that are passed to the named_buff and named_buff_iter
181 callbacks above. Those routines are called from universal.c via the
182 Tie::Hash::NamedCapture interface for %+ and %- and the re::
183 functions in the same file.
184*/
185
186/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
187#define RXapif_FETCH 0x0001
188#define RXapif_STORE 0x0002
189#define RXapif_DELETE 0x0004
190#define RXapif_CLEAR 0x0008
191#define RXapif_EXISTS 0x0010
192#define RXapif_SCALAR 0x0020
193#define RXapif_FIRSTKEY 0x0040
194#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
195
196/* Whether %+ or %- is being operated on */
f1b875a0
YO
197#define RXapif_ONE 0x0100 /* %+ */
198#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
199
200/* Whether this is being called from a re:: function */
f1b875a0
YO
201#define RXapif_REGNAME 0x0400
202#define RXapif_REGNAMES 0x0800
7948fc08 203#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 204
f7e71195
AB
205/*
206=head1 REGEXP Functions
207
208=for apidoc Am|REGEXP *|SvRX|SV *sv
209
210Convenience macro to get the REGEXP from a SV. This is approximately
211equivalent to the following snippet:
212
213 if (SvMAGICAL(sv))
214 mg_get(sv);
1af910ea
DL
215 if (SvROK(sv))
216 sv = MUTABLE_SV(SvRV(sv));
217 if (SvTYPE(sv) == SVt_REGEXP)
218 return (REGEXP*) sv;
f7e71195
AB
219
220NULL will be returned if a REGEXP* is not found.
221
222=for apidoc Am|bool|SvRXOK|SV* sv
223
1af910ea
DL
224Returns a boolean indicating whether the SV (or the one it references)
225is a REGEXP.
f7e71195
AB
226
227If you want to do something with the REGEXP* later use SvRX instead
228and check for NULL.
229
230=cut
231*/
232
233#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
234#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
235
236
7948fc08 237/* Flags stored in regexp->extflags
bbe252da 238 * These are used by code external to the regexp engine
e357fc67 239 *
5b126c84
KW
240 * Note that the flags whose names start with RXf_PMf_ are defined in
241 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 242 *
eb2624c9
FC
243 * NOTE: if you modify any RXf flags you should run regen.pl or
244 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 245 *
bbe252da
YO
246 */
247
5b126c84
KW
248#include "op_reg_common.h"
249
cde0cee5 250#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 251
a20207d7
YO
252#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
253 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
254 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
255 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
256 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 257
94b03d7d 258/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 259#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 260 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 261
bcdf7404 262/* chars and strings used as regex pattern modifiers
486ec47a 263 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
264 *
265 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
266 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
267 * for its own uses. So 'k' is out as well.
bcdf7404 268 */
85508812 269#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 270#define EXEC_PAT_MOD 'e'
87e95b7f 271#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
272#define ONCE_PAT_MOD 'o'
273#define GLOBAL_PAT_MOD 'g'
274#define CONTINUE_PAT_MOD 'c'
275#define MULTILINE_PAT_MOD 'm'
276#define SINGLE_PAT_MOD 's'
277#define IGNORE_PAT_MOD 'i'
278#define XTENDED_PAT_MOD 'x'
4f4d7508 279#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
280#define LOCALE_PAT_MOD 'l'
281#define UNICODE_PAT_MOD 'u'
50e91148 282#define DEPENDS_PAT_MOD 'd'
cfaf538b 283#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
284
285#define ONCE_PAT_MODS "o"
87e95b7f 286#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
287#define EXEC_PAT_MODS "e"
288#define LOOP_PAT_MODS "gc"
4f4d7508 289#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
290#define LOCALE_PAT_MODS "l"
291#define UNICODE_PAT_MODS "u"
50e91148 292#define DEPENDS_PAT_MODS "d"
cfaf538b 293#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 294#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 295
78fecfc8 296/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
297 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
298 * character is bit +1, etc. */
bcdf7404
YO
299#define STD_PAT_MODS "msix"
300
94b03d7d
KW
301#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
302
78fecfc8 303/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
304 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
305 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
306#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
307
308#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 309#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 310#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 311#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 312
cb5027f2 313/*
eb2624c9
FC
314 * NOTE: if you modify any RXf flags you should run regen.pl or
315 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
316 *
317 */
bcdf7404 318
e795e964
KW
319/* Leave some space, so future bit allocations can go either in the shared or
320 * unshared area without affecting binary compatibility */
df7a8460 321#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 322
bf28aba1
KW
323/* Manually decorate this function with gcc-style attributes just to
324 * avoid having to restructure the header files and their called order,
325 * as proto.h would have to be included before this file, and isn't */
326
a62b1201
KW
327PERL_STATIC_INLINE const char *
328get_regex_charset_name(const U32 flags, STRLEN* const lenp)
329 __attribute__warn_unused_result__;
330
df7a8460 331#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
332
333PERL_STATIC_INLINE const char *
334get_regex_charset_name(const U32 flags, STRLEN* const lenp)
335{
336 /* Returns a string that corresponds to the name of the regex character set
337 * given by 'flags', and *lenp is set the length of that string, which
338 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
339
340 *lenp = 1;
341 switch (get_regex_charset(flags)) {
50e91148 342 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
343 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
344 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 345 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 346 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
347 *lenp = 2;
348 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
349 default:
350 return "?"; /* Unknown */
a62b1201 351 }
a62b1201
KW
352}
353
52d81aa8 354/* Anchor and GPOS related stuff */
1ad4ec54
KW
355#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
356#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
357#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
358#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
359#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
360#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
361/* two bits here */
362#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
363#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
364#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
365
bbe252da 366/* What we have seen */
1ad4ec54
KW
367#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
368#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
369#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
370
371/* Special */
1ad4ec54
KW
372#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
373#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
374
375/* UTF8 related */
e795e964 376#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
377
378/* Intuit related */
e795e964
KW
379#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
380#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
381#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
0ac6acae
AB
382
383/*
384 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
385 be used by regex engines to check whether they should set
386 RXf_SKIPWHITE
387*/
e795e964 388#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15))
0ac6acae 389
bbe252da
YO
390#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
391
392/* Copy and tainted info */
e795e964 393#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
394
395/* during execution: pattern temporarily tainted by executing locale ops;
396 * post-execution: $1 et al are tainted */
e795e964 397#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
398/* this pattern was tainted during compilation */
399#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
400
401/* Flags indicating special patterns */
e795e964
KW
402#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
403#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */
404#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 405#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 406#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
407# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
408#endif
c737faaf 409
cb5027f2 410/*
eb2624c9
FC
411 * NOTE: if you modify any RXf flags you should run regen.pl or
412 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
413 *
414 */
bbe252da 415
bbe252da 416#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
417#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
418#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
419#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
420#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
421#define RX_MATCH_TAINTED_set(prog, t) ((t) \
422 ? RX_MATCH_TAINTED_on(prog) \
423 : RX_MATCH_TAINTED_off(prog))
c277df42 424
07bc277f
NC
425#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
426#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
427#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
428#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
429#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
430#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
431#define RX_MATCH_COPIED_set(prog,t) ((t) \
432 ? RX_MATCH_COPIED_on(prog) \
433 : RX_MATCH_COPIED_off(prog))
f9f4320a 434
07bc277f 435#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 436
07bc277f 437/* For source compatibility. We used to store these explicitly. */
f7c278bf 438#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 439#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
440/* FIXME? Are we hardcoding too much here and constraining plugin extension
441 writers? Specifically, the value 1 assumes that the wrapped version always
442 has exactly one character at the end, a ')'. Will that always be true? */
443#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 444#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 445#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 446#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 447#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 448#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 449#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
450# define RX_EXTFLAGS(prog) \
451 (*({ \
7948fc08
RU
452 const REGEXP *const _rx_extflags = (prog); \
453 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
454 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
455 }))
456# define RX_ENGINE(prog) \
457 (*({ \
7948fc08
RU
458 const REGEXP *const _rx_engine = (prog); \
459 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
460 &SvANY(_rx_engine)->engine; \
4450b630
NC
461 }))
462# define RX_SUBBEG(prog) \
463 (*({ \
7948fc08
RU
464 const REGEXP *const _rx_subbeg = (prog); \
465 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
466 &SvANY(_rx_subbeg)->subbeg; \
4450b630 467 }))
200ff815 468# define RX_OFFS(prog) \
4450b630 469 (*({ \
7948fc08
RU
470 const REGEXP *const _rx_offs = (prog); \
471 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
472 &SvANY(_rx_offs)->offs; \
4450b630
NC
473 }))
474# define RX_NPARENS(prog) \
475 (*({ \
7948fc08
RU
476 const REGEXP *const _rx_nparens = (prog); \
477 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
478 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
479 }))
480#else
481# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
482# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
483# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
484# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
485# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
486#endif
288b8c02 487#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
488#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
489#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
490#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
491#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
492#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
bdd9a1b1 493#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 494
be8e71aa
YO
495#endif /* PLUGGABLE_RE_EXTENSION */
496
486ec47a 497/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 498
f8c7b90f 499#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 500#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
501 STMT_START {if (RX_SAVED_COPY(rx)) { \
502 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
503 } \
504 if (RX_MATCH_COPIED(rx)) { \
07bc277f 505 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
506 RX_MATCH_COPIED_off(rx); \
507 }} STMT_END
508#else
509#define RX_MATCH_COPY_FREE(rx) \
510 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 511 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
512 RX_MATCH_COPIED_off(rx); \
513 }} STMT_END
514#endif
515
07bc277f
NC
516#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
517#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
518#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
519#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
520#define RX_MATCH_UTF8_set(prog, t) ((t) \
521 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
522 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
523
524/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 525#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 526
cad2e5aa
JH
527#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
528#define REXEC_CHECKED 0x02 /* check_substr already checked. */
529#define REXEC_SCREAM 0x04 /* use scream table. */
530#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 531#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 532
7fe44985 533#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
534# define ReREFCNT_inc(re) \
535 ({ \
536 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
537 REGEXP *const _rerefcnt_inc = (re); \
538 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
539 SvREFCNT_inc(_rerefcnt_inc); \
540 _rerefcnt_inc; \
288b8c02
NC
541 })
542# define ReREFCNT_dec(re) \
543 ({ \
544 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
545 REGEXP *const _rerefcnt_dec = (re); \
546 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
547 })
548#else
549# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 550# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
551#endif
552
553/* FIXME for plugins. */
cf93c79d
IZ
554
555#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
556#define FBMcf_TAIL_DOLLARM 2
557#define FBMcf_TAIL_Z 4
558#define FBMcf_TAIL_z 8
559#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
560
561#define FBMrf_MULTILINE 1
cad2e5aa 562
3b0527fe
DM
563/* some basic information about the current match that is created by
564 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
565
566typedef struct {
288b8c02 567 REGEXP *prog;
3b0527fe
DM
568 char *bol;
569 char *till;
570 SV *sv;
571 char *ganch;
24b23f37 572 char *cutpoint;
3b0527fe
DM
573} regmatch_info;
574
5d9a96ca
DM
575
576/* structures for holding and saving the state maintained by regmatch() */
577
d56b3014 578#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 579#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 580#endif
6bda09f9 581
5d9a96ca
DM
582typedef I32 CHECKPOINT;
583
a0374537 584typedef struct regmatch_state {
40a82448 585 int resume_state; /* where to jump to on return */
24d3c4a9 586 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
587
588 union {
77cb431f
DM
589
590 /* this is a fake union member that matches the first element
591 * of each member that needs to store positive backtrack
592 * information */
593 struct {
594 struct regmatch_state *prev_yes_state;
595 } yes;
596
fae667d5
YO
597 /* branchlike members */
598 /* this is a fake union member that matches the first elements
599 * of each member that needs to behave like a branch */
600 struct {
5d458dd8
YO
601 /* this first element must match u.yes */
602 struct regmatch_state *prev_yes_state;
fae667d5
YO
603 U32 lastparen;
604 CHECKPOINT cp;
605
606 } branchlike;
607
608 struct {
609 /* the first elements must match u.branchlike */
610 struct regmatch_state *prev_yes_state;
611 U32 lastparen;
612 CHECKPOINT cp;
613
614 regnode *next_branch; /* next branch node */
615 } branch;
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
2e64971a 623 U32 accepted; /* how many accepting states left */
7f69552c 624 U16 *jump; /* positive offsets from me */
166ba7cd 625 regnode *B; /* node following the trie */
7f69552c 626 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
627 U8 *firstpos;/* pos in string of first trie match */
628 U32 firstchars;/* len in chars of firstpos from start */
629 U16 nextword;/* next word to try */
630 U16 topword; /* longest accepted word */
631 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
632 } trie;
633
fae667d5
YO
634 /* special types - these members are used to store state for special
635 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 636 struct {
77cb431f
DM
637 /* this first element must match u.yes */
638 struct regmatch_state *prev_yes_state;
faec1544
DM
639 struct regmatch_state *prev_eval;
640 struct regmatch_state *prev_curlyx;
288b8c02 641 REGEXP *prev_rex;
faec1544
DM
642 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
643 flip when transitioning between
644 inner and outer rexen */
aa283a38
DM
645 CHECKPOINT cp; /* remember current savestack indexes */
646 CHECKPOINT lastcp;
6bda09f9 647 U32 close_paren; /* which close bracket is our end */
1b5c067c 648 regnode *B; /* the node following us */
e822a8b4
DM
649 } eval;
650
651 struct {
c476f425 652 /* this first element must match u.yes */
262b90c4 653 struct regmatch_state *prev_yes_state;
fae667d5
YO
654 I32 wanted;
655 I32 logical; /* saved copy of 'logical' var */
656 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
657 } ifmatch; /* and SUSPEND/UNLESSM */
658
659 struct {
660 /* this first element must match u.yes */
661 struct regmatch_state *prev_yes_state;
662 struct regmatch_state *prev_mark;
663 SV* mark_name;
664 char *mark_loc;
665 } mark;
666
667 struct {
668 int val;
669 } keeper;
670
671 /* quantifiers - these members are used for storing state for
672 for the regops used to implement quantifiers */
673 struct {
674 /* this first element must match u.yes */
675 struct regmatch_state *prev_yes_state;
c476f425 676 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
677 regnode *me; /* the CURLYX node */
678 regnode *B; /* the B node in /A*B/ */
c476f425
DM
679 CHECKPOINT cp; /* remember current savestack index */
680 bool minmod;
a0374537 681 int parenfloor;/* how far back to strip paren data */
c476f425
DM
682
683 /* these two are modified by WHILEM */
684 int count; /* how many instances of A we've matched */
685 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
686 } curlyx;
687
688 struct {
c476f425 689 /* this first element must match u.yes */
262b90c4 690 struct regmatch_state *prev_yes_state;
c476f425
DM
691 struct regmatch_state *save_curlyx;
692 CHECKPOINT cp; /* remember current savestack indexes */
693 CHECKPOINT lastcp;
694 char *save_lastloc; /* previous curlyx.lastloc */
695 I32 cache_offset;
696 I32 cache_mask;
e822a8b4
DM
697 } whilem;
698
699 struct {
5d458dd8
YO
700 /* this first element must match u.yes */
701 struct regmatch_state *prev_yes_state;
e822a8b4 702 I32 c1, c2; /* case fold search */
40a82448
DM
703 CHECKPOINT cp;
704 I32 alen; /* length of first-matched A string */
705 I32 count;
0cadcf80 706 bool minmod;
40a82448
DM
707 regnode *A, *B; /* the nodes corresponding to /A*B/ */
708 regnode *me; /* the curlym node */
e822a8b4
DM
709 } curlym;
710
711 struct {
3b6647e0 712 U32 paren;
c255a977 713 CHECKPOINT cp;
e822a8b4 714 I32 c1, c2; /* case fold search */
c255a977
DM
715 char *maxpos; /* highest possible point in string to match */
716 char *oldloc; /* the previous locinput */
e822a8b4 717 int count;
c255a977
DM
718 int min, max; /* {m,n} */
719 regnode *A, *B; /* the nodes corresponding to /A*B/ */
720 } curly; /* and CURLYN/PLUS/STAR */
dad79028 721
d8319b27 722 } u;
5d9a96ca
DM
723} regmatch_state;
724
725/* how many regmatch_state structs to allocate as a single slab.
726 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
727 * pointers, plus 1 for any mythical malloc overhead. */
728
729#define PERL_REGMATCH_SLAB_SLOTS \
730 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
731
732typedef struct regmatch_slab {
733 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
734 struct regmatch_slab *prev, *next;
735} regmatch_slab;
1ade1aa1 736
46ab3289
NC
737#define PL_reg_flags PL_reg_state.re_state_reg_flags
738#define PL_bostr PL_reg_state.re_state_bostr
739#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 740#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 741#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
742#define PL_reglastparen PL_reg_state.re_state_reglastparen
743#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
744#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
745#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
46ab3289
NC
746#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
747#define PL_reg_magic PL_reg_state.re_state_reg_magic
748#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
749#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
750#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
751#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
752#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
753#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
754#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
755#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
756#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
757#define PL_regsize PL_reg_state.re_state_regsize
758#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
759#define PL_nrs PL_reg_state.re_state_nrs
760
1ade1aa1
NC
761struct re_save_state {
762 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c 763 U32 re_state_reg_start_tmpl; /* from regexec.c */
ed301438 764 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 765 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 766 bool re_reparsing; /* runtime (?{}) fed back into parser */
1ade1aa1
NC
767 char *re_state_bostr;
768 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 769 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 770 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
771 U32 *re_state_reglastparen; /* Similarly for lastparen. */
772 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 773 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 774 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
775 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
776 PMOP *re_state_reg_curpm; /* from regexec.c */
777 char *re_state_reg_oldsaved; /* old saved substr during match */
778 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
779 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
780 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
781 I32 re_state_reg_maxiter; /* max wait until caching pos */
782 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 783 U32 re_state_regsize; /* from regexec.c */
1b5c067c 784 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
785 char *re_state_reg_starttry; /* from regexec.c */
786#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 787 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
788#endif
789};
790
791#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
792 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 793
1ade1aa1
NC
794/*
795 * Local variables:
796 * c-indentation-style: bsd
797 * c-basic-offset: 4
14d04a33 798 * indent-tabs-mode: nil
1ade1aa1
NC
799 * End:
800 *
14d04a33 801 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 802 */