This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexp.h: Define flag bit directly, not indirectly
[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
79a2a0e8
KW
21#include "utf8.h"
22
c277df42
IZ
23struct regnode {
24 U8 flags;
25 U8 type;
26 U16 next_off;
27};
28
29typedef struct regnode regnode;
30
cad2e5aa 31struct reg_substr_data;
2779dcf1 32
0ee3c8d0
JH
33struct reg_data;
34
f9f4320a 35struct regexp_engine;
28d8d7f4 36struct regexp;
bbe252da 37
785a26d5 38struct reg_substr_datum {
c4828ca0
DM
39 SSize_t min_offset; /* min pos (in chars) that substr must appear */
40 SSize_t max_offset /* max pos (in chars) that substr must appear */;
785a26d5
YO
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
c4828ca0 43 SSize_t end_shift; /* how many fixed chars must end the string */
785a26d5
YO
44};
45struct reg_substr_data {
6480a6c4 46 U8 check_ix; /* index into data[] of check substr */
785a26d5
YO
47 struct reg_substr_datum data[3]; /* Actual array */
48};
f9f4320a 49
db2c6cb3 50#ifdef PERL_ANY_COW
bbe252da
YO
51#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
52#else
53#define SV_SAVED_COPY
ed252734 54#endif
28d8d7f4 55
b3fd53f3
DM
56/* offsets within a string of a particular /(.)/ capture */
57
f0ab9afb 58typedef struct regexp_paren_pair {
99a90e59
FC
59 SSize_t start;
60 SSize_t end;
b3fd53f3
DM
61 /* 'start_tmp' records a new opening position before the matching end
62 * has been found, so that the old start and end values are still
63 * valid, e.g.
64 * "abc" =~ /(.(?{print "[$1]"}))+/
65 *outputs [][a][b]
66 * This field is not part of the API. */
ea3daa5d 67 SSize_t start_tmp;
f0ab9afb 68} regexp_paren_pair;
28d8d7f4 69
52ae8f7e 70#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 71#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 72#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
73
74/* Subtracting b from a leaves in a everything that was there that isn't in b,
75 * that is the intersection of a with b's complement */
76#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
77#endif
78
3d2bd50a
DM
79/* record the position of a (?{...}) within a pattern */
80
81struct reg_code_block {
82 STRLEN start;
83 STRLEN end;
84 OP *block;
b30fcab9 85 REGEXP *src_regex;
3d2bd50a
DM
86};
87
88
882227b7
AB
89/*
90 The regexp/REGEXP struct, see L<perlreapi> for further documentation
91 on the individual fields. The struct is ordered so that the most
92 commonly used fields are placed at the start.
93
94 Any patch that adds items to this struct will need to include
95 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
96 (C<pregfree()>). This involves freeing or cloning items in the
97 regexp's data array based on the data item's type.
98*/
99
08e44740
NC
100#define _REGEXP_COMMON \
101 /* what engine created this regexp? */ \
102 const struct regexp_engine* engine; \
103 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 104 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
105 /* Information about the match that the perl core uses to */ \
106 /* manage things */ \
107 U32 extflags; /* Flags used both externally and internally */ \
ea3daa5d
FC
108 SSize_t minlen; /* mininum possible number of chars in string to match */\
109 SSize_t minlenret; /* mininum possible number of chars in $& */ \
110 STRLEN gofs; /* chars left of pos that we search from */ \
08e44740
NC
111 /* substring data about strings that must appear in the */ \
112 /* final match, used for optimisations */ \
113 struct reg_substr_data *substrs; \
114 U32 nparens; /* number of capture buffers */ \
115 /* private engine specific data */ \
116 U32 intflags; /* Engine Specific Internal flags */ \
117 void *pprivate; /* Data private to the regex engine which */ \
118 /* created this object. */ \
119 /* Data about the last/current match. These are modified */ \
120 /* during matching */ \
121 U32 lastparen; /* last open paren matched */ \
122 U32 lastcloseparen; /* last close paren matched */ \
08e44740
NC
123 /* Array of offsets for (@-) and (@+) */ \
124 regexp_paren_pair *offs; \
125 /* saved or original string so \digit works forever. */ \
126 char *subbeg; \
127 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
19394178 128 SSize_t sublen; /* Length of string pointed by subbeg */ \
ea3daa5d
FC
129 SSize_t suboffset; /* byte offset of subbeg from logical start of str */ \
130 SSize_t subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740 131 /* Information about the match that isn't often used */ \
ee273784 132 SSize_t maxlen; /* mininum possible number of chars in string to match */\
08e44740 133 /* offset from wrapped to the start of precomp */ \
654eccd5 134 PERL_BITFIELD32 pre_prefix:4; \
dbc200c5
YO
135 /* original flags used to compile the pattern, may differ */ \
136 /* from extflags in various ways */ \
137 PERL_BITFIELD32 compflags:9; \
b82ad30c 138 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 139
bbe252da 140typedef struct regexp {
288b8c02 141 _XPV_HEAD;
08e44740 142 _REGEXP_COMMON;
f8fc2ecf
YO
143} regexp;
144
89f6f287 145#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 146
785a26d5 147/* used for high speed searches */
f9f4320a
YO
148typedef struct re_scream_pos_data_s
149{
150 char **scream_olds; /* match pos */
ea3daa5d 151 SSize_t *scream_pos; /* Internal iterator of scream. */
f9f4320a
YO
152} re_scream_pos_data;
153
785a26d5
YO
154/* regexp_engine structure. This is the dispatch table for regexes.
155 * Any regex engine implementation must be able to build one of these.
156 */
f9f4320a 157typedef struct regexp_engine {
1593ad57 158 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc 159 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
ea3daa5d 160 char* strbeg, SSize_t minend, SV* sv,
49d7dfbc 161 void* data, U32 flags);
52a21eb3
DM
162 char* (*intuit) (pTHX_
163 REGEXP * const rx,
164 SV *sv,
165 const char * const strbeg,
166 char *strpos,
167 char *strend,
168 const U32 flags,
9f61653a 169 re_scream_pos_data *data);
49d7dfbc
AB
170 SV* (*checkstr) (pTHX_ REGEXP * const rx);
171 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 172 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 173 SV * const sv);
2fdbfb4d
AB
174 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
175 SV const * const value);
176 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
177 const I32 paren);
192b9cd1
AB
178 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
179 SV * const value, const U32 flags);
180 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
181 const U32 flags);
49d7dfbc 182 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 183#ifdef USE_ITHREADS
49d7dfbc 184 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 185#endif
3c13cae6
DM
186 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
187 OP *expr, const struct regexp_engine* eng,
188 REGEXP *VOL old_re,
76ac488f 189 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
190} regexp_engine;
191
192b9cd1
AB
192/*
193 These are passed to the numbered capture variable callbacks as the
194 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
195 $2 etc.
196*/
2c7b5d76
DM
197#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
198#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
199#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
200#define RX_BUFF_IDX_PREMATCH -2 /* $` */
201#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
202#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
203
204/*
205 Flags that are passed to the named_buff and named_buff_iter
206 callbacks above. Those routines are called from universal.c via the
207 Tie::Hash::NamedCapture interface for %+ and %- and the re::
208 functions in the same file.
209*/
210
211/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
212#define RXapif_FETCH 0x0001
213#define RXapif_STORE 0x0002
214#define RXapif_DELETE 0x0004
215#define RXapif_CLEAR 0x0008
216#define RXapif_EXISTS 0x0010
217#define RXapif_SCALAR 0x0020
218#define RXapif_FIRSTKEY 0x0040
219#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
220
221/* Whether %+ or %- is being operated on */
f1b875a0
YO
222#define RXapif_ONE 0x0100 /* %+ */
223#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
224
225/* Whether this is being called from a re:: function */
f1b875a0
YO
226#define RXapif_REGNAME 0x0400
227#define RXapif_REGNAMES 0x0800
7948fc08 228#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 229
f7e71195
AB
230/*
231=head1 REGEXP Functions
232
233=for apidoc Am|REGEXP *|SvRX|SV *sv
234
72d33970 235Convenience macro to get the REGEXP from a SV. This is approximately
f7e71195
AB
236equivalent to the following snippet:
237
238 if (SvMAGICAL(sv))
239 mg_get(sv);
1af910ea
DL
240 if (SvROK(sv))
241 sv = MUTABLE_SV(SvRV(sv));
242 if (SvTYPE(sv) == SVt_REGEXP)
243 return (REGEXP*) sv;
f7e71195
AB
244
245NULL will be returned if a REGEXP* is not found.
246
247=for apidoc Am|bool|SvRXOK|SV* sv
248
1af910ea
DL
249Returns a boolean indicating whether the SV (or the one it references)
250is a REGEXP.
f7e71195
AB
251
252If you want to do something with the REGEXP* later use SvRX instead
253and check for NULL.
254
255=cut
256*/
257
258#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
259#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
260
261
7948fc08 262/* Flags stored in regexp->extflags
bbe252da 263 * These are used by code external to the regexp engine
e357fc67 264 *
5b126c84
KW
265 * Note that the flags whose names start with RXf_PMf_ are defined in
266 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 267 *
eb2624c9
FC
268 * NOTE: if you modify any RXf flags you should run regen.pl or
269 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 270 *
bbe252da
YO
271 */
272
5b126c84
KW
273#include "op_reg_common.h"
274
cde0cee5 275#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 276
a20207d7
YO
277#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
278 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
279 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
280 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
281 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 282
94b03d7d 283/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 284#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 285 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 286
bcdf7404 287/* chars and strings used as regex pattern modifiers
486ec47a 288 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
289 *
290 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
291 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
292 * for its own uses. So 'k' is out as well.
bcdf7404 293 */
85508812 294#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 295#define EXEC_PAT_MOD 'e'
87e95b7f 296#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
297#define ONCE_PAT_MOD 'o'
298#define GLOBAL_PAT_MOD 'g'
299#define CONTINUE_PAT_MOD 'c'
300#define MULTILINE_PAT_MOD 'm'
301#define SINGLE_PAT_MOD 's'
302#define IGNORE_PAT_MOD 'i'
303#define XTENDED_PAT_MOD 'x'
4f4d7508 304#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
305#define LOCALE_PAT_MOD 'l'
306#define UNICODE_PAT_MOD 'u'
50e91148 307#define DEPENDS_PAT_MOD 'd'
cfaf538b 308#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
309
310#define ONCE_PAT_MODS "o"
87e95b7f 311#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
312#define EXEC_PAT_MODS "e"
313#define LOOP_PAT_MODS "gc"
4f4d7508 314#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
315#define LOCALE_PAT_MODS "l"
316#define UNICODE_PAT_MODS "u"
50e91148 317#define DEPENDS_PAT_MODS "d"
cfaf538b 318#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 319#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 320
78fecfc8 321/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
322 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
323 * character is bit +1, etc. */
bcdf7404
YO
324#define STD_PAT_MODS "msix"
325
94b03d7d
KW
326#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
327
78fecfc8 328/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
329 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
330 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
331#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
332
333#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 334#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 335#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 336#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 337
cb5027f2 338/*
eb2624c9
FC
339 * NOTE: if you modify any RXf flags you should run regen.pl or
340 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
341 *
342 */
bcdf7404 343
dbc200c5
YO
344/*
345 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
346 be used by regex engines to check whether they should set
347 RXf_SKIPWHITE
348*/
32f62247
KW
349#define RXf_SPLIT RXf_PMf_SPLIT
350
351/* Leave some space, so future bit allocations can go either in the shared or
352 * unshared area without affecting binary compatibility */
353#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT)
f4a4cecb 354
1d32d911 355/* Unused +0 - +6 */
52d81aa8 356
bbe252da 357/* What we have seen */
a3b51d37
KW
358#define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+7))
359#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
360
361/* Special */
ee273784 362#define RXf_UNBOUNDED_QUANTIFIER_SEEN (1<<(RXf_BASE_SHIFT+9))
1ad4ec54 363#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
364
365/* UTF8 related */
0603fe5c 366#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) /* $1 etc are utf8 */
bbe252da
YO
367
368/* Intuit related */
e795e964
KW
369#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
370#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
371#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
372#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
373
a3b51d37
KW
374/* Do we have some sort of anchor? */
375#define RXf_IS_ANCHORED (1<<(RXf_BASE_SHIFT+15))
376
bbe252da 377/* Copy and tainted info */
e795e964 378#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810 379
1738e041 380/* post-execution: $1 et al are tainted */
e795e964 381#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
382/* this pattern was tainted during compilation */
383#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
384
385/* Flags indicating special patterns */
e795e964 386#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
dbc200c5 387#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split " " */
e795e964 388#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 389#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 390#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
391# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
392#endif
c737faaf 393
cb5027f2 394/*
eb2624c9
FC
395 * NOTE: if you modify any RXf flags you should run regen.pl or
396 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
397 *
398 */
bbe252da 399
dc6d7f5c 400#ifdef NO_TAINT_SUPPORT
284167a5
SM
401# define RX_ISTAINTED(prog) 0
402# define RX_TAINT_on(prog) NOOP
272d35c9
DM
403# define RXp_MATCH_TAINTED(prog) 0
404# define RX_MATCH_TAINTED(prog) 0
405# define RXp_MATCH_TAINTED_on(prog) NOOP
406# define RX_MATCH_TAINTED_on(prog) NOOP
407# define RX_MATCH_TAINTED_off(prog) NOOP
284167a5
SM
408#else
409# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
410# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
272d35c9
DM
411# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
412# define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
413# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
414# define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
415# define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
284167a5
SM
416#endif
417
bbe252da 418#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
72311751
GS
419#define RX_MATCH_TAINTED_set(prog, t) ((t) \
420 ? RX_MATCH_TAINTED_on(prog) \
421 : RX_MATCH_TAINTED_off(prog))
c277df42 422
07bc277f
NC
423#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
424#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
425#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
426#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
427#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
428#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
429#define RX_MATCH_COPIED_set(prog,t) ((t) \
430 ? RX_MATCH_COPIED_on(prog) \
431 : RX_MATCH_COPIED_off(prog))
f9f4320a 432
07bc277f 433#define RXp_EXTFLAGS(rx) ((rx)->extflags)
dbc200c5 434#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 435
07bc277f 436/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
437#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
438#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
439/* FIXME? Are we hardcoding too much here and constraining plugin extension
440 writers? Specifically, the value 1 assumes that the wrapped version always
441 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
442#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
443#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
444#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 445#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 446#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 447#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a 448#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
dbc200c5 449#define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog))
8d919b0a
FC
450#define RX_ENGINE(prog) (ReANY(prog)->engine)
451#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
452#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
453#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
454#define RX_OFFS(prog) (ReANY(prog)->offs)
455#define RX_NPARENS(prog) (ReANY(prog)->nparens)
456#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
457#define RX_MINLEN(prog) (ReANY(prog)->minlen)
458#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
459#define RX_GOFS(prog) (ReANY(prog)->gofs)
460#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
461#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
462#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
03c83e26
DM
463/* last match was zero-length */
464#define RX_ZERO_LEN(prog) \
99a90e59
FC
465 (RX_OFFS(prog)[0].start + (SSize_t)RX_GOFS(prog) \
466 == RX_OFFS(prog)[0].end)
220fc49f 467
be8e71aa
YO
468#endif /* PLUGGABLE_RE_EXTENSION */
469
486ec47a 470/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 471
db2c6cb3 472#ifdef PERL_ANY_COW
ed252734 473#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
474 STMT_START {if (RX_SAVED_COPY(rx)) { \
475 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
476 } \
477 if (RX_MATCH_COPIED(rx)) { \
07bc277f 478 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
479 RX_MATCH_COPIED_off(rx); \
480 }} STMT_END
481#else
482#define RX_MATCH_COPY_FREE(rx) \
483 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 484 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
485 RX_MATCH_COPIED_off(rx); \
486 }} STMT_END
487#endif
488
07bc277f
NC
489#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
490#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
491#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
492#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f 493#define RX_MATCH_UTF8_set(prog, t) ((t) \
0603fe5c
DM
494 ? RX_MATCH_UTF8_on(prog) \
495 : RX_MATCH_UTF8_off(prog))
efd26800
NC
496
497/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 498#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 499
a340edde
DM
500
501/* bits in flags arg of Perl_regexec_flags() */
502
503#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
504#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
505#define REXEC_SCREAM 0x04 /* currently unused. */
506#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
507#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
508 no need to copy string again */
509
510 /* under REXEC_COPY_STR, it's ok for the
511 engine (modulo PL_sawamperand etc)
512 to skip copying: ... */
513#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
514#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
515#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
516 the start pos (so s/.\G// would fail
517 on second iteration */
c277df42 518
7fe44985 519#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
520# define ReREFCNT_inc(re) \
521 ({ \
522 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
523 REGEXP *const _rerefcnt_inc = (re); \
524 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
525 SvREFCNT_inc(_rerefcnt_inc); \
526 _rerefcnt_inc; \
288b8c02
NC
527 })
528# define ReREFCNT_dec(re) \
529 ({ \
530 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
531 REGEXP *const _rerefcnt_dec = (re); \
532 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
533 })
534#else
535# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 536# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 537#endif
8d919b0a 538#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
539
540/* FIXME for plugins. */
cf93c79d
IZ
541
542#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
543#define FBMcf_TAIL_DOLLARM 2
544#define FBMcf_TAIL_Z 4
545#define FBMcf_TAIL_z 8
546#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
547
548#define FBMrf_MULTILINE 1
cad2e5aa 549
331b2dcc
DM
550struct regmatch_state;
551struct regmatch_slab;
8adc0f72 552
bf2039a9
DM
553/* like regmatch_info_aux, but contains extra fields only needed if the
554 * pattern contains (?{}). If used, is snuck into the second slot in the
555 * regmatch_state stack at the start of execution */
8adc0f72
DM
556
557typedef struct {
558 regexp *rex;
559 PMOP *curpm; /* saved PL_curpm */
560#ifdef PERL_ANY_COW
561 SV *saved_copy; /* saved saved_copy field from rex */
562#endif
563 char *subbeg; /* saved subbeg field from rex */
564 STRLEN sublen; /* saved sublen field from rex */
565 STRLEN suboffset; /* saved suboffset field from rex */
566 STRLEN subcoffset; /* saved subcoffset field from rex */
567 MAGIC *pos_magic; /* pos() magic attached to $_ */
ea3daa5d 568 SSize_t pos; /* the original value of pos() in pos_magic */
25fdce4a 569 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
570} regmatch_info_aux_eval;
571
572
573/* fields that logically live in regmatch_info, but which need cleaning
574 * up on croak(), and so are instead are snuck into the first slot in
575 * the regmatch_state stack at the start of execution */
576
577typedef struct {
578 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
579 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
580 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 581 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
582} regmatch_info_aux;
583
8adc0f72 584
3b0527fe 585/* some basic information about the current match that is created by
bf2039a9
DM
586 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
587 * It is allocated as a local var on the stack, so nothing should be
588 * stored in it that needs preserving or clearing up on croak().
589 * For that, see the aux_info and aux_info_eval members of the
590 * regmatch_state union. */
3b0527fe
DM
591
592typedef struct {
28d03b21 593 REGEXP *prog; /* the regex being executed */
9d9163fb 594 const char * strbeg; /* real start of string */
28d03b21
DM
595 char *strend; /* one byte beyond last char of match string */
596 char *till; /* matches shorter than this fail (see minlen arg) */
597 SV *sv; /* the SV string currently being matched */
598 char *ganch; /* position of \G anchor */
599 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
600 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
601 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
602 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
603 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 604 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 605 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
606 bool is_utf8_pat; /* regex is utf8 */
607 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 608 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
609} regmatch_info;
610
5d9a96ca
DM
611
612/* structures for holding and saving the state maintained by regmatch() */
613
d56b3014 614#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 615#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 616#endif
6bda09f9 617
5d9a96ca
DM
618typedef I32 CHECKPOINT;
619
a0374537 620typedef struct regmatch_state {
40a82448 621 int resume_state; /* where to jump to on return */
24d3c4a9 622 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
623
624 union {
77cb431f 625
bf2039a9
DM
626 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
627 * the nest. They aren't saved backtrack state; rather they
628 * represent one or two extra chunks of data that need allocating
629 * at the start of a match. These fields would logically live in
630 * the regmatch_info struct, except that is allocated on the
631 * C stack, and these fields are all things that require cleanup
632 * after a croak(), when the stack is lost.
633 * As a convenience, we just use the first 1 or 2 regmatch_state
634 * slots to store this info, as we will be allocating a slab of
635 * these anyway. Otherwise we'd have to malloc and then free them,
636 * or allocate them on the save stack (where they will get
637 * realloced if the save stack grows).
638 * info_aux contains the extra fields that are always needed;
639 * info_aux_eval contains extra fields that only needed if
640 * the pattern contains code blocks
641 * We split them into two separate structs to avoid increasing
642 * the size of the union.
643 */
644
645 regmatch_info_aux info_aux;
646
647 regmatch_info_aux_eval info_aux_eval;
648
77cb431f
DM
649 /* this is a fake union member that matches the first element
650 * of each member that needs to store positive backtrack
651 * information */
652 struct {
653 struct regmatch_state *prev_yes_state;
654 } yes;
655
fae667d5
YO
656 /* branchlike members */
657 /* this is a fake union member that matches the first elements
658 * of each member that needs to behave like a branch */
659 struct {
5d458dd8
YO
660 /* this first element must match u.yes */
661 struct regmatch_state *prev_yes_state;
fae667d5 662 U32 lastparen;
f6033a9d 663 U32 lastcloseparen;
fae667d5
YO
664 CHECKPOINT cp;
665
666 } branchlike;
667
668 struct {
669 /* the first elements must match u.branchlike */
670 struct regmatch_state *prev_yes_state;
671 U32 lastparen;
f6033a9d 672 U32 lastcloseparen;
fae667d5
YO
673 CHECKPOINT cp;
674
675 regnode *next_branch; /* next branch node */
676 } branch;
677
678 struct {
679 /* the first elements must match u.branchlike */
680 struct regmatch_state *prev_yes_state;
681 U32 lastparen;
f6033a9d 682 U32 lastcloseparen;
fae667d5
YO
683 CHECKPOINT cp;
684
2e64971a 685 U32 accepted; /* how many accepting states left */
6603fe3e 686 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 687 U16 *jump; /* positive offsets from me */
7f69552c 688 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
689 U8 *firstpos;/* pos in string of first trie match */
690 U32 firstchars;/* len in chars of firstpos from start */
691 U16 nextword;/* next word to try */
692 U16 topword; /* longest accepted word */
e822a8b4
DM
693 } trie;
694
fae667d5
YO
695 /* special types - these members are used to store state for special
696 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 697 struct {
77cb431f
DM
698 /* this first element must match u.yes */
699 struct regmatch_state *prev_yes_state;
faec1544
DM
700 struct regmatch_state *prev_eval;
701 struct regmatch_state *prev_curlyx;
288b8c02 702 REGEXP *prev_rex;
aa283a38
DM
703 CHECKPOINT cp; /* remember current savestack indexes */
704 CHECKPOINT lastcp;
6bda09f9 705 U32 close_paren; /* which close bracket is our end */
1b5c067c 706 regnode *B; /* the node following us */
e822a8b4
DM
707 } eval;
708
709 struct {
c476f425 710 /* this first element must match u.yes */
262b90c4 711 struct regmatch_state *prev_yes_state;
fae667d5
YO
712 I32 wanted;
713 I32 logical; /* saved copy of 'logical' var */
714 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
715 } ifmatch; /* and SUSPEND/UNLESSM */
716
717 struct {
718 /* this first element must match u.yes */
719 struct regmatch_state *prev_yes_state;
720 struct regmatch_state *prev_mark;
721 SV* mark_name;
722 char *mark_loc;
723 } mark;
724
725 struct {
726 int val;
727 } keeper;
728
729 /* quantifiers - these members are used for storing state for
730 for the regops used to implement quantifiers */
731 struct {
732 /* this first element must match u.yes */
733 struct regmatch_state *prev_yes_state;
c476f425 734 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
735 regnode *me; /* the CURLYX node */
736 regnode *B; /* the B node in /A*B/ */
c476f425
DM
737 CHECKPOINT cp; /* remember current savestack index */
738 bool minmod;
a0374537 739 int parenfloor;/* how far back to strip paren data */
c476f425
DM
740
741 /* these two are modified by WHILEM */
742 int count; /* how many instances of A we've matched */
743 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
744 } curlyx;
745
746 struct {
c476f425 747 /* this first element must match u.yes */
262b90c4 748 struct regmatch_state *prev_yes_state;
c476f425
DM
749 struct regmatch_state *save_curlyx;
750 CHECKPOINT cp; /* remember current savestack indexes */
751 CHECKPOINT lastcp;
752 char *save_lastloc; /* previous curlyx.lastloc */
753 I32 cache_offset;
754 I32 cache_mask;
e822a8b4
DM
755 } whilem;
756
757 struct {
5d458dd8
YO
758 /* this first element must match u.yes */
759 struct regmatch_state *prev_yes_state;
79a2a0e8 760 int c1, c2; /* case fold search */
40a82448 761 CHECKPOINT cp;
f6033a9d
DM
762 U32 lastparen;
763 U32 lastcloseparen;
40a82448
DM
764 I32 alen; /* length of first-matched A string */
765 I32 count;
0cadcf80 766 bool minmod;
40a82448
DM
767 regnode *A, *B; /* the nodes corresponding to /A*B/ */
768 regnode *me; /* the curlym node */
79a2a0e8
KW
769 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
770 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
771 } curlym;
772
773 struct {
3b6647e0 774 U32 paren;
c255a977 775 CHECKPOINT cp;
f6033a9d
DM
776 U32 lastparen;
777 U32 lastcloseparen;
79a2a0e8 778 int c1, c2; /* case fold search */
c255a977
DM
779 char *maxpos; /* highest possible point in string to match */
780 char *oldloc; /* the previous locinput */
e822a8b4 781 int count;
c255a977
DM
782 int min, max; /* {m,n} */
783 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
784 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
785 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 786 } curly; /* and CURLYN/PLUS/STAR */
dad79028 787
d8319b27 788 } u;
5d9a96ca
DM
789} regmatch_state;
790
791/* how many regmatch_state structs to allocate as a single slab.
792 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
793 * pointers, plus 1 for any mythical malloc overhead. */
794
795#define PERL_REGMATCH_SLAB_SLOTS \
796 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
797
798typedef struct regmatch_slab {
799 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
800 struct regmatch_slab *prev, *next;
801} regmatch_slab;
1ade1aa1 802
46ab3289 803
cde0cee5 804
1ade1aa1
NC
805/*
806 * Local variables:
807 * c-indentation-style: bsd
808 * c-basic-offset: 4
14d04a33 809 * indent-tabs-mode: nil
1ade1aa1
NC
810 * End:
811 *
14d04a33 812 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 813 */