This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexp.h: Make tentative division of free-bit space
[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 */
3f6e6b29 353#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT + 5)
52d81aa8 354
bbe252da 355/* What we have seen */
3f6e6b29
KW
356#define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+2))
357#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+3))
bbe252da
YO
358
359/* Special */
3f6e6b29
KW
360#define RXf_UNBOUNDED_QUANTIFIER_SEEN (1<<(RXf_BASE_SHIFT+4))
361#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+5))
bbe252da
YO
362
363/* UTF8 related */
3f6e6b29 364#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+6)) /* $1 etc are utf8 */
bbe252da
YO
365
366/* Intuit related */
3f6e6b29
KW
367#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+7))
368#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+8))
369#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+9))
bbe252da
YO
370#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
371
a3b51d37 372/* Do we have some sort of anchor? */
3f6e6b29 373#define RXf_IS_ANCHORED (1<<(RXf_BASE_SHIFT+10))
a3b51d37 374
bbe252da 375/* Copy and tainted info */
3f6e6b29 376#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+11))
ef07e810 377
1738e041 378/* post-execution: $1 et al are tainted */
3f6e6b29 379#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+12))
ef07e810 380/* this pattern was tainted during compilation */
3f6e6b29 381#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+13))
52d81aa8
NC
382
383/* Flags indicating special patterns */
3f6e6b29
KW
384#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+14)) /* Pattern is /^/ */
385#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+15)) /* Pattern is for a split " " */
386#define RXf_WHITE (1<<(RXf_BASE_SHIFT+16)) /* Pattern is /\s+/ */
387#define RXf_NULL (1U<<(RXf_BASE_SHIFT+17)) /* Pattern is // */
388#if RXf_BASE_SHIFT+17 > 31
1850c8f9
KW
389# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
390#endif
c737faaf 391
cb5027f2 392/*
eb2624c9
FC
393 * NOTE: if you modify any RXf flags you should run regen.pl or
394 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
395 *
396 */
bbe252da 397
dc6d7f5c 398#ifdef NO_TAINT_SUPPORT
284167a5
SM
399# define RX_ISTAINTED(prog) 0
400# define RX_TAINT_on(prog) NOOP
272d35c9
DM
401# define RXp_MATCH_TAINTED(prog) 0
402# define RX_MATCH_TAINTED(prog) 0
403# define RXp_MATCH_TAINTED_on(prog) NOOP
404# define RX_MATCH_TAINTED_on(prog) NOOP
405# define RX_MATCH_TAINTED_off(prog) NOOP
284167a5
SM
406#else
407# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
408# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
272d35c9
DM
409# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
410# define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
411# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
412# define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
413# define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
284167a5
SM
414#endif
415
bbe252da 416#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
72311751
GS
417#define RX_MATCH_TAINTED_set(prog, t) ((t) \
418 ? RX_MATCH_TAINTED_on(prog) \
419 : RX_MATCH_TAINTED_off(prog))
c277df42 420
07bc277f
NC
421#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
422#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
423#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
424#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
425#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
426#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
427#define RX_MATCH_COPIED_set(prog,t) ((t) \
428 ? RX_MATCH_COPIED_on(prog) \
429 : RX_MATCH_COPIED_off(prog))
f9f4320a 430
07bc277f 431#define RXp_EXTFLAGS(rx) ((rx)->extflags)
dbc200c5 432#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 433
07bc277f 434/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
435#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
436#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
437/* FIXME? Are we hardcoding too much here and constraining plugin extension
438 writers? Specifically, the value 1 assumes that the wrapped version always
439 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
440#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
441#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
442#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 443#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 444#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 445#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a 446#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
dbc200c5 447#define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog))
8d919b0a
FC
448#define RX_ENGINE(prog) (ReANY(prog)->engine)
449#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
450#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
451#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
452#define RX_OFFS(prog) (ReANY(prog)->offs)
453#define RX_NPARENS(prog) (ReANY(prog)->nparens)
454#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
455#define RX_MINLEN(prog) (ReANY(prog)->minlen)
456#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
457#define RX_GOFS(prog) (ReANY(prog)->gofs)
458#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
459#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
460#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
03c83e26
DM
461/* last match was zero-length */
462#define RX_ZERO_LEN(prog) \
99a90e59
FC
463 (RX_OFFS(prog)[0].start + (SSize_t)RX_GOFS(prog) \
464 == RX_OFFS(prog)[0].end)
220fc49f 465
be8e71aa
YO
466#endif /* PLUGGABLE_RE_EXTENSION */
467
486ec47a 468/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 469
db2c6cb3 470#ifdef PERL_ANY_COW
ed252734 471#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
472 STMT_START {if (RX_SAVED_COPY(rx)) { \
473 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
474 } \
475 if (RX_MATCH_COPIED(rx)) { \
07bc277f 476 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
477 RX_MATCH_COPIED_off(rx); \
478 }} STMT_END
479#else
480#define RX_MATCH_COPY_FREE(rx) \
481 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 482 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
483 RX_MATCH_COPIED_off(rx); \
484 }} STMT_END
485#endif
486
07bc277f
NC
487#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
488#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
489#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
490#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f 491#define RX_MATCH_UTF8_set(prog, t) ((t) \
0603fe5c
DM
492 ? RX_MATCH_UTF8_on(prog) \
493 : RX_MATCH_UTF8_off(prog))
efd26800
NC
494
495/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 496#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 497
a340edde
DM
498
499/* bits in flags arg of Perl_regexec_flags() */
500
501#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
502#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
503#define REXEC_SCREAM 0x04 /* currently unused. */
504#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
505#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
506 no need to copy string again */
507
508 /* under REXEC_COPY_STR, it's ok for the
509 engine (modulo PL_sawamperand etc)
510 to skip copying: ... */
511#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
512#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
513#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
514 the start pos (so s/.\G// would fail
515 on second iteration */
c277df42 516
7fe44985 517#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
518# define ReREFCNT_inc(re) \
519 ({ \
520 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
521 REGEXP *const _rerefcnt_inc = (re); \
522 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
523 SvREFCNT_inc(_rerefcnt_inc); \
524 _rerefcnt_inc; \
288b8c02
NC
525 })
526# define ReREFCNT_dec(re) \
527 ({ \
528 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
529 REGEXP *const _rerefcnt_dec = (re); \
530 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
531 })
532#else
533# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 534# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 535#endif
8d919b0a 536#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
537
538/* FIXME for plugins. */
cf93c79d
IZ
539
540#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
541#define FBMcf_TAIL_DOLLARM 2
542#define FBMcf_TAIL_Z 4
543#define FBMcf_TAIL_z 8
544#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
545
546#define FBMrf_MULTILINE 1
cad2e5aa 547
331b2dcc
DM
548struct regmatch_state;
549struct regmatch_slab;
8adc0f72 550
bf2039a9
DM
551/* like regmatch_info_aux, but contains extra fields only needed if the
552 * pattern contains (?{}). If used, is snuck into the second slot in the
553 * regmatch_state stack at the start of execution */
8adc0f72
DM
554
555typedef struct {
556 regexp *rex;
557 PMOP *curpm; /* saved PL_curpm */
558#ifdef PERL_ANY_COW
559 SV *saved_copy; /* saved saved_copy field from rex */
560#endif
561 char *subbeg; /* saved subbeg field from rex */
562 STRLEN sublen; /* saved sublen field from rex */
563 STRLEN suboffset; /* saved suboffset field from rex */
564 STRLEN subcoffset; /* saved subcoffset field from rex */
565 MAGIC *pos_magic; /* pos() magic attached to $_ */
ea3daa5d 566 SSize_t pos; /* the original value of pos() in pos_magic */
25fdce4a 567 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
568} regmatch_info_aux_eval;
569
570
571/* fields that logically live in regmatch_info, but which need cleaning
572 * up on croak(), and so are instead are snuck into the first slot in
573 * the regmatch_state stack at the start of execution */
574
575typedef struct {
576 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
577 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
578 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 579 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
580} regmatch_info_aux;
581
8adc0f72 582
3b0527fe 583/* some basic information about the current match that is created by
bf2039a9
DM
584 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
585 * It is allocated as a local var on the stack, so nothing should be
586 * stored in it that needs preserving or clearing up on croak().
587 * For that, see the aux_info and aux_info_eval members of the
588 * regmatch_state union. */
3b0527fe
DM
589
590typedef struct {
28d03b21 591 REGEXP *prog; /* the regex being executed */
9d9163fb 592 const char * strbeg; /* real start of string */
28d03b21
DM
593 char *strend; /* one byte beyond last char of match string */
594 char *till; /* matches shorter than this fail (see minlen arg) */
595 SV *sv; /* the SV string currently being matched */
596 char *ganch; /* position of \G anchor */
597 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
598 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
599 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
600 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
601 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 602 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 603 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
604 bool is_utf8_pat; /* regex is utf8 */
605 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 606 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
607} regmatch_info;
608
5d9a96ca
DM
609
610/* structures for holding and saving the state maintained by regmatch() */
611
d56b3014 612#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 613#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 614#endif
6bda09f9 615
5d9a96ca
DM
616typedef I32 CHECKPOINT;
617
a0374537 618typedef struct regmatch_state {
40a82448 619 int resume_state; /* where to jump to on return */
24d3c4a9 620 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
621
622 union {
77cb431f 623
bf2039a9
DM
624 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
625 * the nest. They aren't saved backtrack state; rather they
626 * represent one or two extra chunks of data that need allocating
627 * at the start of a match. These fields would logically live in
628 * the regmatch_info struct, except that is allocated on the
629 * C stack, and these fields are all things that require cleanup
630 * after a croak(), when the stack is lost.
631 * As a convenience, we just use the first 1 or 2 regmatch_state
632 * slots to store this info, as we will be allocating a slab of
633 * these anyway. Otherwise we'd have to malloc and then free them,
634 * or allocate them on the save stack (where they will get
635 * realloced if the save stack grows).
636 * info_aux contains the extra fields that are always needed;
637 * info_aux_eval contains extra fields that only needed if
638 * the pattern contains code blocks
639 * We split them into two separate structs to avoid increasing
640 * the size of the union.
641 */
642
643 regmatch_info_aux info_aux;
644
645 regmatch_info_aux_eval info_aux_eval;
646
77cb431f
DM
647 /* this is a fake union member that matches the first element
648 * of each member that needs to store positive backtrack
649 * information */
650 struct {
651 struct regmatch_state *prev_yes_state;
652 } yes;
653
fae667d5
YO
654 /* branchlike members */
655 /* this is a fake union member that matches the first elements
656 * of each member that needs to behave like a branch */
657 struct {
5d458dd8
YO
658 /* this first element must match u.yes */
659 struct regmatch_state *prev_yes_state;
fae667d5 660 U32 lastparen;
f6033a9d 661 U32 lastcloseparen;
fae667d5
YO
662 CHECKPOINT cp;
663
664 } branchlike;
665
666 struct {
667 /* the first elements must match u.branchlike */
668 struct regmatch_state *prev_yes_state;
669 U32 lastparen;
f6033a9d 670 U32 lastcloseparen;
fae667d5
YO
671 CHECKPOINT cp;
672
673 regnode *next_branch; /* next branch node */
674 } branch;
675
676 struct {
677 /* the first elements must match u.branchlike */
678 struct regmatch_state *prev_yes_state;
679 U32 lastparen;
f6033a9d 680 U32 lastcloseparen;
fae667d5
YO
681 CHECKPOINT cp;
682
2e64971a 683 U32 accepted; /* how many accepting states left */
6603fe3e 684 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 685 U16 *jump; /* positive offsets from me */
7f69552c 686 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
687 U8 *firstpos;/* pos in string of first trie match */
688 U32 firstchars;/* len in chars of firstpos from start */
689 U16 nextword;/* next word to try */
690 U16 topword; /* longest accepted word */
e822a8b4
DM
691 } trie;
692
fae667d5
YO
693 /* special types - these members are used to store state for special
694 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 695 struct {
77cb431f
DM
696 /* this first element must match u.yes */
697 struct regmatch_state *prev_yes_state;
faec1544
DM
698 struct regmatch_state *prev_eval;
699 struct regmatch_state *prev_curlyx;
288b8c02 700 REGEXP *prev_rex;
aa283a38
DM
701 CHECKPOINT cp; /* remember current savestack indexes */
702 CHECKPOINT lastcp;
6bda09f9 703 U32 close_paren; /* which close bracket is our end */
1b5c067c 704 regnode *B; /* the node following us */
e822a8b4
DM
705 } eval;
706
707 struct {
c476f425 708 /* this first element must match u.yes */
262b90c4 709 struct regmatch_state *prev_yes_state;
fae667d5
YO
710 I32 wanted;
711 I32 logical; /* saved copy of 'logical' var */
712 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
713 } ifmatch; /* and SUSPEND/UNLESSM */
714
715 struct {
716 /* this first element must match u.yes */
717 struct regmatch_state *prev_yes_state;
718 struct regmatch_state *prev_mark;
719 SV* mark_name;
720 char *mark_loc;
721 } mark;
722
723 struct {
724 int val;
725 } keeper;
726
727 /* quantifiers - these members are used for storing state for
728 for the regops used to implement quantifiers */
729 struct {
730 /* this first element must match u.yes */
731 struct regmatch_state *prev_yes_state;
c476f425 732 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
733 regnode *me; /* the CURLYX node */
734 regnode *B; /* the B node in /A*B/ */
c476f425
DM
735 CHECKPOINT cp; /* remember current savestack index */
736 bool minmod;
a0374537 737 int parenfloor;/* how far back to strip paren data */
c476f425
DM
738
739 /* these two are modified by WHILEM */
740 int count; /* how many instances of A we've matched */
741 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
742 } curlyx;
743
744 struct {
c476f425 745 /* this first element must match u.yes */
262b90c4 746 struct regmatch_state *prev_yes_state;
c476f425
DM
747 struct regmatch_state *save_curlyx;
748 CHECKPOINT cp; /* remember current savestack indexes */
749 CHECKPOINT lastcp;
750 char *save_lastloc; /* previous curlyx.lastloc */
751 I32 cache_offset;
752 I32 cache_mask;
e822a8b4
DM
753 } whilem;
754
755 struct {
5d458dd8
YO
756 /* this first element must match u.yes */
757 struct regmatch_state *prev_yes_state;
79a2a0e8 758 int c1, c2; /* case fold search */
40a82448 759 CHECKPOINT cp;
f6033a9d
DM
760 U32 lastparen;
761 U32 lastcloseparen;
40a82448
DM
762 I32 alen; /* length of first-matched A string */
763 I32 count;
0cadcf80 764 bool minmod;
40a82448
DM
765 regnode *A, *B; /* the nodes corresponding to /A*B/ */
766 regnode *me; /* the curlym node */
79a2a0e8
KW
767 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
768 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
769 } curlym;
770
771 struct {
3b6647e0 772 U32 paren;
c255a977 773 CHECKPOINT cp;
f6033a9d
DM
774 U32 lastparen;
775 U32 lastcloseparen;
79a2a0e8 776 int c1, c2; /* case fold search */
c255a977
DM
777 char *maxpos; /* highest possible point in string to match */
778 char *oldloc; /* the previous locinput */
e822a8b4 779 int count;
c255a977
DM
780 int min, max; /* {m,n} */
781 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
782 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
783 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 784 } curly; /* and CURLYN/PLUS/STAR */
dad79028 785
d8319b27 786 } u;
5d9a96ca
DM
787} regmatch_state;
788
789/* how many regmatch_state structs to allocate as a single slab.
790 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
791 * pointers, plus 1 for any mythical malloc overhead. */
792
793#define PERL_REGMATCH_SLAB_SLOTS \
794 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
795
796typedef struct regmatch_slab {
797 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
798 struct regmatch_slab *prev, *next;
799} regmatch_slab;
1ade1aa1 800
46ab3289 801
cde0cee5 802
1ade1aa1
NC
803/*
804 * Local variables:
805 * c-indentation-style: bsd
806 * c-basic-offset: 4
14d04a33 807 * indent-tabs-mode: nil
1ade1aa1
NC
808 * End:
809 *
14d04a33 810 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 811 */