This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
eliminate RF_warned flag from PL_reg_flags
[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
YO
38struct reg_substr_datum {
39 I32 min_offset;
40 I32 max_offset;
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
43 I32 end_shift;
44};
45struct reg_substr_data {
46 struct reg_substr_datum data[3]; /* Actual array */
47};
f9f4320a 48
db2c6cb3 49#ifdef PERL_ANY_COW
bbe252da
YO
50#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
51#else
52#define SV_SAVED_COPY
ed252734 53#endif
28d8d7f4 54
b3fd53f3
DM
55/* offsets within a string of a particular /(.)/ capture */
56
f0ab9afb
NC
57typedef struct regexp_paren_pair {
58 I32 start;
59 I32 end;
b3fd53f3
DM
60 /* 'start_tmp' records a new opening position before the matching end
61 * has been found, so that the old start and end values are still
62 * valid, e.g.
63 * "abc" =~ /(.(?{print "[$1]"}))+/
64 *outputs [][a][b]
65 * This field is not part of the API. */
1ca2007e 66 I32 start_tmp;
f0ab9afb 67} regexp_paren_pair;
28d8d7f4 68
52ae8f7e 69#if defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_UTF8_C)
164173a2 70#define _invlist_union(a, b, output) _invlist_union_maybe_complement_2nd(a, b, FALSE, output)
52ae8f7e 71#define _invlist_intersection(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, FALSE, output)
3f80b571
KW
72
73/* Subtracting b from a leaves in a everything that was there that isn't in b,
74 * that is the intersection of a with b's complement */
75#define _invlist_subtract(a, b, output) _invlist_intersection_maybe_complement_2nd(a, b, TRUE, output)
52ae8f7e
KW
76#endif
77
3d2bd50a
DM
78/* record the position of a (?{...}) within a pattern */
79
80struct reg_code_block {
81 STRLEN start;
82 STRLEN end;
83 OP *block;
b30fcab9 84 REGEXP *src_regex;
3d2bd50a
DM
85};
86
87
882227b7
AB
88/*
89 The regexp/REGEXP struct, see L<perlreapi> for further documentation
90 on the individual fields. The struct is ordered so that the most
91 commonly used fields are placed at the start.
92
93 Any patch that adds items to this struct will need to include
94 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
95 (C<pregfree()>). This involves freeing or cloning items in the
96 regexp's data array based on the data item's type.
97*/
98
08e44740
NC
99#define _REGEXP_COMMON \
100 /* what engine created this regexp? */ \
101 const struct regexp_engine* engine; \
102 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 103 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
104 /* Information about the match that the perl core uses to */ \
105 /* manage things */ \
106 U32 extflags; /* Flags used both externally and internally */ \
9d1714db
KW
107 I32 minlen; /* mininum possible number of chars in string to match */\
108 I32 minlenret; /* mininum possible number of chars in $& */ \
08e44740
NC
109 U32 gofs; /* chars left of pos that we search from */ \
110 /* substring data about strings that must appear in the */ \
111 /* final match, used for optimisations */ \
112 struct reg_substr_data *substrs; \
113 U32 nparens; /* number of capture buffers */ \
114 /* private engine specific data */ \
115 U32 intflags; /* Engine Specific Internal flags */ \
116 void *pprivate; /* Data private to the regex engine which */ \
117 /* created this object. */ \
118 /* Data about the last/current match. These are modified */ \
119 /* during matching */ \
120 U32 lastparen; /* last open paren matched */ \
121 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 122 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
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 */\
128 I32 sublen; /* Length of string pointed by subbeg */ \
6502e081
DM
129 I32 suboffset; /* byte offset of subbeg from logical start of str */ \
130 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740
NC
131 /* Information about the match that isn't often used */ \
132 /* offset from wrapped to the start of precomp */ \
654eccd5 133 PERL_BITFIELD32 pre_prefix:4; \
b82ad30c 134 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 135
bbe252da 136typedef struct regexp {
288b8c02 137 _XPV_HEAD;
08e44740 138 _REGEXP_COMMON;
f8fc2ecf
YO
139} regexp;
140
89f6f287 141#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 142
785a26d5 143/* used for high speed searches */
f9f4320a
YO
144typedef struct re_scream_pos_data_s
145{
146 char **scream_olds; /* match pos */
147 I32 *scream_pos; /* Internal iterator of scream. */
148} re_scream_pos_data;
149
785a26d5
YO
150/* regexp_engine structure. This is the dispatch table for regexes.
151 * Any regex engine implementation must be able to build one of these.
152 */
f9f4320a 153typedef struct regexp_engine {
1593ad57 154 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
155 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
156 char* strbeg, I32 minend, SV* screamer,
157 void* data, U32 flags);
158 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
159 char *strend, const U32 flags,
160 re_scream_pos_data *data);
49d7dfbc
AB
161 SV* (*checkstr) (pTHX_ REGEXP * const rx);
162 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 163 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 164 SV * const sv);
2fdbfb4d
AB
165 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
166 SV const * const value);
167 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
168 const I32 paren);
192b9cd1
AB
169 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
170 SV * const value, const U32 flags);
171 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
172 const U32 flags);
49d7dfbc 173 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 174#ifdef USE_ITHREADS
49d7dfbc 175 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 176#endif
3c13cae6
DM
177 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
178 OP *expr, const struct regexp_engine* eng,
179 REGEXP *VOL old_re,
76ac488f 180 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
181} regexp_engine;
182
192b9cd1
AB
183/*
184 These are passed to the numbered capture variable callbacks as the
185 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
186 $2 etc.
187*/
2c7b5d76
DM
188#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
189#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
190#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
191#define RX_BUFF_IDX_PREMATCH -2 /* $` */
192#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
193#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
194
195/*
196 Flags that are passed to the named_buff and named_buff_iter
197 callbacks above. Those routines are called from universal.c via the
198 Tie::Hash::NamedCapture interface for %+ and %- and the re::
199 functions in the same file.
200*/
201
202/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
203#define RXapif_FETCH 0x0001
204#define RXapif_STORE 0x0002
205#define RXapif_DELETE 0x0004
206#define RXapif_CLEAR 0x0008
207#define RXapif_EXISTS 0x0010
208#define RXapif_SCALAR 0x0020
209#define RXapif_FIRSTKEY 0x0040
210#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
211
212/* Whether %+ or %- is being operated on */
f1b875a0
YO
213#define RXapif_ONE 0x0100 /* %+ */
214#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
215
216/* Whether this is being called from a re:: function */
f1b875a0
YO
217#define RXapif_REGNAME 0x0400
218#define RXapif_REGNAMES 0x0800
7948fc08 219#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 220
f7e71195
AB
221/*
222=head1 REGEXP Functions
223
224=for apidoc Am|REGEXP *|SvRX|SV *sv
225
226Convenience macro to get the REGEXP from a SV. This is approximately
227equivalent to the following snippet:
228
229 if (SvMAGICAL(sv))
230 mg_get(sv);
1af910ea
DL
231 if (SvROK(sv))
232 sv = MUTABLE_SV(SvRV(sv));
233 if (SvTYPE(sv) == SVt_REGEXP)
234 return (REGEXP*) sv;
f7e71195
AB
235
236NULL will be returned if a REGEXP* is not found.
237
238=for apidoc Am|bool|SvRXOK|SV* sv
239
1af910ea
DL
240Returns a boolean indicating whether the SV (or the one it references)
241is a REGEXP.
f7e71195
AB
242
243If you want to do something with the REGEXP* later use SvRX instead
244and check for NULL.
245
246=cut
247*/
248
249#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
250#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
251
252
7948fc08 253/* Flags stored in regexp->extflags
bbe252da 254 * These are used by code external to the regexp engine
e357fc67 255 *
5b126c84
KW
256 * Note that the flags whose names start with RXf_PMf_ are defined in
257 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 258 *
eb2624c9
FC
259 * NOTE: if you modify any RXf flags you should run regen.pl or
260 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 261 *
bbe252da
YO
262 */
263
5b126c84
KW
264#include "op_reg_common.h"
265
cde0cee5 266#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 267
a20207d7
YO
268#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
269 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
270 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
271 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
272 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 273
94b03d7d 274/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 275#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 276 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 277
bcdf7404 278/* chars and strings used as regex pattern modifiers
486ec47a 279 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
280 *
281 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
282 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
283 * for its own uses. So 'k' is out as well.
bcdf7404 284 */
85508812 285#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 286#define EXEC_PAT_MOD 'e'
87e95b7f 287#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
288#define ONCE_PAT_MOD 'o'
289#define GLOBAL_PAT_MOD 'g'
290#define CONTINUE_PAT_MOD 'c'
291#define MULTILINE_PAT_MOD 'm'
292#define SINGLE_PAT_MOD 's'
293#define IGNORE_PAT_MOD 'i'
294#define XTENDED_PAT_MOD 'x'
4f4d7508 295#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
296#define LOCALE_PAT_MOD 'l'
297#define UNICODE_PAT_MOD 'u'
50e91148 298#define DEPENDS_PAT_MOD 'd'
cfaf538b 299#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
300
301#define ONCE_PAT_MODS "o"
87e95b7f 302#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
303#define EXEC_PAT_MODS "e"
304#define LOOP_PAT_MODS "gc"
4f4d7508 305#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
306#define LOCALE_PAT_MODS "l"
307#define UNICODE_PAT_MODS "u"
50e91148 308#define DEPENDS_PAT_MODS "d"
cfaf538b 309#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 310#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 311
78fecfc8 312/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
313 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
314 * character is bit +1, etc. */
bcdf7404
YO
315#define STD_PAT_MODS "msix"
316
94b03d7d
KW
317#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
318
78fecfc8 319/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
320 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
321 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
322#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
323
324#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 325#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 326#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 327#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 328
cb5027f2 329/*
eb2624c9
FC
330 * NOTE: if you modify any RXf flags you should run regen.pl or
331 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
332 *
333 */
bcdf7404 334
e795e964
KW
335/* Leave some space, so future bit allocations can go either in the shared or
336 * unshared area without affecting binary compatibility */
df7a8460 337#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 338
bf28aba1
KW
339/* Manually decorate this function with gcc-style attributes just to
340 * avoid having to restructure the header files and their called order,
341 * as proto.h would have to be included before this file, and isn't */
342
a62b1201
KW
343PERL_STATIC_INLINE const char *
344get_regex_charset_name(const U32 flags, STRLEN* const lenp)
345 __attribute__warn_unused_result__;
346
df7a8460 347#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
348
349PERL_STATIC_INLINE const char *
350get_regex_charset_name(const U32 flags, STRLEN* const lenp)
351{
352 /* Returns a string that corresponds to the name of the regex character set
353 * given by 'flags', and *lenp is set the length of that string, which
354 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
355
356 *lenp = 1;
357 switch (get_regex_charset(flags)) {
50e91148 358 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
359 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
360 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 361 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 362 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
363 *lenp = 2;
364 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
365 default:
366 return "?"; /* Unknown */
a62b1201 367 }
a62b1201
KW
368}
369
cccd1425
FC
370/*
371 Two flags no longer used.
372 RXf_SPLIT used to be set in Perl_pmruntime if op_flags & OPf_SPECIAL,
373 i.e., split. It was used by the regex engine to check whether it should
374 set RXf_SKIPWHITE. Regexp plugins on CPAN also have done the same thing
375 historically, so we leave these flags defined.
376*/
377#ifndef PERL_CORE
378# define RXf_SPLIT 0
379# define RXf_SKIPWHITE 0
380#endif
381
382
52d81aa8 383/* Anchor and GPOS related stuff */
1ad4ec54
KW
384#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
385#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
386#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
387#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
388#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
389#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
390/* two bits here */
391#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
392#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
393#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
394
bbe252da 395/* What we have seen */
1ad4ec54
KW
396#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
397#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
398#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
399
400/* Special */
1ad4ec54
KW
401#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
402#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
403
404/* UTF8 related */
e795e964 405#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
406
407/* Intuit related */
e795e964
KW
408#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
409#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
410#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
411#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
412
2ed8c61f
FC
413#define RXf_MODIFIES_VARS (1<<(RXf_BASE_SHIFT+15))
414
bbe252da 415/* Copy and tainted info */
e795e964 416#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
417
418/* during execution: pattern temporarily tainted by executing locale ops;
419 * post-execution: $1 et al are tainted */
e795e964 420#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
421/* this pattern was tainted during compilation */
422#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
423
424/* Flags indicating special patterns */
e795e964 425#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
e795e964 426#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 427#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 428#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
429# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
430#endif
c737faaf 431
cb5027f2 432/*
eb2624c9
FC
433 * NOTE: if you modify any RXf flags you should run regen.pl or
434 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
435 *
436 */
bbe252da 437
284167a5
SM
438#if NO_TAINT_SUPPORT
439# define RX_ISTAINTED(prog) 0
440# define RX_TAINT_on(prog) NOOP
441#else
442# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
443# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
444#endif
445
bbe252da 446#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
447#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
448#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
449#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
450#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
451#define RX_MATCH_TAINTED_set(prog, t) ((t) \
452 ? RX_MATCH_TAINTED_on(prog) \
453 : RX_MATCH_TAINTED_off(prog))
c277df42 454
07bc277f
NC
455#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
456#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
457#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
458#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
459#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
460#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
461#define RX_MATCH_COPIED_set(prog,t) ((t) \
462 ? RX_MATCH_COPIED_on(prog) \
463 : RX_MATCH_COPIED_off(prog))
f9f4320a 464
07bc277f 465#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 466
07bc277f 467/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
468#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
469#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
470/* FIXME? Are we hardcoding too much here and constraining plugin extension
471 writers? Specifically, the value 1 assumes that the wrapped version always
472 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
473#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
474#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
475#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 476#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 477#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 478#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a
FC
479#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
480#define RX_ENGINE(prog) (ReANY(prog)->engine)
481#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
482#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
483#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
484#define RX_OFFS(prog) (ReANY(prog)->offs)
485#define RX_NPARENS(prog) (ReANY(prog)->nparens)
486#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
487#define RX_MINLEN(prog) (ReANY(prog)->minlen)
488#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
489#define RX_GOFS(prog) (ReANY(prog)->gofs)
490#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
491#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
492#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
220fc49f 493
be8e71aa
YO
494#endif /* PLUGGABLE_RE_EXTENSION */
495
486ec47a 496/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 497
db2c6cb3 498#ifdef PERL_ANY_COW
ed252734 499#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
500 STMT_START {if (RX_SAVED_COPY(rx)) { \
501 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
502 } \
503 if (RX_MATCH_COPIED(rx)) { \
07bc277f 504 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
505 RX_MATCH_COPIED_off(rx); \
506 }} STMT_END
507#else
508#define RX_MATCH_COPY_FREE(rx) \
509 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 510 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
511 RX_MATCH_COPIED_off(rx); \
512 }} STMT_END
513#endif
514
07bc277f
NC
515#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
516#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
517#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
518#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
519#define RX_MATCH_UTF8_set(prog, t) ((t) \
520 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
521 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
522
523/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 524#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 525
cad2e5aa
JH
526#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
527#define REXEC_CHECKED 0x02 /* check_substr already checked. */
528#define REXEC_SCREAM 0x04 /* use scream table. */
529#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 530#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
6502e081
DM
531 /* under REXEC_COPY_STR, it's ok for the
532 * engine (modulo PL_sawamperand etc)
533 * to skip copying ... */
534#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
535#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
c277df42 536
7fe44985 537#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
538# define ReREFCNT_inc(re) \
539 ({ \
540 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
541 REGEXP *const _rerefcnt_inc = (re); \
542 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
543 SvREFCNT_inc(_rerefcnt_inc); \
544 _rerefcnt_inc; \
288b8c02
NC
545 })
546# define ReREFCNT_dec(re) \
547 ({ \
548 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
549 REGEXP *const _rerefcnt_dec = (re); \
550 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
551 })
552#else
553# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 554# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 555#endif
8d919b0a 556#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
557
558/* FIXME for plugins. */
cf93c79d
IZ
559
560#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
561#define FBMcf_TAIL_DOLLARM 2
562#define FBMcf_TAIL_Z 4
563#define FBMcf_TAIL_z 8
564#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
565
566#define FBMrf_MULTILINE 1
cad2e5aa 567
3b0527fe
DM
568/* some basic information about the current match that is created by
569 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
570
571typedef struct {
288b8c02 572 REGEXP *prog;
3b0527fe
DM
573 char *bol;
574 char *till;
575 SV *sv;
576 char *ganch;
24b23f37 577 char *cutpoint;
984e6dd1 578 bool is_utf8_pat;
39819bd9 579 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
580} regmatch_info;
581
5d9a96ca
DM
582
583/* structures for holding and saving the state maintained by regmatch() */
584
d56b3014 585#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 586#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 587#endif
6bda09f9 588
5d9a96ca
DM
589typedef I32 CHECKPOINT;
590
a0374537 591typedef struct regmatch_state {
40a82448 592 int resume_state; /* where to jump to on return */
24d3c4a9 593 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
594
595 union {
77cb431f
DM
596
597 /* this is a fake union member that matches the first element
598 * of each member that needs to store positive backtrack
599 * information */
600 struct {
601 struct regmatch_state *prev_yes_state;
602 } yes;
603
fae667d5
YO
604 /* branchlike members */
605 /* this is a fake union member that matches the first elements
606 * of each member that needs to behave like a branch */
607 struct {
5d458dd8
YO
608 /* this first element must match u.yes */
609 struct regmatch_state *prev_yes_state;
fae667d5 610 U32 lastparen;
f6033a9d 611 U32 lastcloseparen;
fae667d5
YO
612 CHECKPOINT cp;
613
614 } branchlike;
615
616 struct {
617 /* the first elements must match u.branchlike */
618 struct regmatch_state *prev_yes_state;
619 U32 lastparen;
f6033a9d 620 U32 lastcloseparen;
fae667d5
YO
621 CHECKPOINT cp;
622
623 regnode *next_branch; /* next branch node */
624 } branch;
625
626 struct {
627 /* the first elements must match u.branchlike */
628 struct regmatch_state *prev_yes_state;
629 U32 lastparen;
f6033a9d 630 U32 lastcloseparen;
fae667d5
YO
631 CHECKPOINT cp;
632
2e64971a 633 U32 accepted; /* how many accepting states left */
6603fe3e 634 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 635 U16 *jump; /* positive offsets from me */
7f69552c 636 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
637 U8 *firstpos;/* pos in string of first trie match */
638 U32 firstchars;/* len in chars of firstpos from start */
639 U16 nextword;/* next word to try */
640 U16 topword; /* longest accepted word */
e822a8b4
DM
641 } trie;
642
fae667d5
YO
643 /* special types - these members are used to store state for special
644 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 645 struct {
77cb431f
DM
646 /* this first element must match u.yes */
647 struct regmatch_state *prev_yes_state;
faec1544
DM
648 struct regmatch_state *prev_eval;
649 struct regmatch_state *prev_curlyx;
288b8c02 650 REGEXP *prev_rex;
984e6dd1 651 bool saved_utf8_pat; /* saved copy of is_utf8_pat */
aa283a38
DM
652 CHECKPOINT cp; /* remember current savestack indexes */
653 CHECKPOINT lastcp;
6bda09f9 654 U32 close_paren; /* which close bracket is our end */
1b5c067c 655 regnode *B; /* the node following us */
e822a8b4
DM
656 } eval;
657
658 struct {
c476f425 659 /* this first element must match u.yes */
262b90c4 660 struct regmatch_state *prev_yes_state;
fae667d5
YO
661 I32 wanted;
662 I32 logical; /* saved copy of 'logical' var */
663 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
664 } ifmatch; /* and SUSPEND/UNLESSM */
665
666 struct {
667 /* this first element must match u.yes */
668 struct regmatch_state *prev_yes_state;
669 struct regmatch_state *prev_mark;
670 SV* mark_name;
671 char *mark_loc;
672 } mark;
673
674 struct {
675 int val;
676 } keeper;
677
678 /* quantifiers - these members are used for storing state for
679 for the regops used to implement quantifiers */
680 struct {
681 /* this first element must match u.yes */
682 struct regmatch_state *prev_yes_state;
c476f425 683 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
684 regnode *me; /* the CURLYX node */
685 regnode *B; /* the B node in /A*B/ */
c476f425
DM
686 CHECKPOINT cp; /* remember current savestack index */
687 bool minmod;
a0374537 688 int parenfloor;/* how far back to strip paren data */
c476f425
DM
689
690 /* these two are modified by WHILEM */
691 int count; /* how many instances of A we've matched */
692 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
693 } curlyx;
694
695 struct {
c476f425 696 /* this first element must match u.yes */
262b90c4 697 struct regmatch_state *prev_yes_state;
c476f425
DM
698 struct regmatch_state *save_curlyx;
699 CHECKPOINT cp; /* remember current savestack indexes */
700 CHECKPOINT lastcp;
701 char *save_lastloc; /* previous curlyx.lastloc */
702 I32 cache_offset;
703 I32 cache_mask;
e822a8b4
DM
704 } whilem;
705
706 struct {
5d458dd8
YO
707 /* this first element must match u.yes */
708 struct regmatch_state *prev_yes_state;
79a2a0e8 709 int c1, c2; /* case fold search */
40a82448 710 CHECKPOINT cp;
f6033a9d
DM
711 U32 lastparen;
712 U32 lastcloseparen;
40a82448
DM
713 I32 alen; /* length of first-matched A string */
714 I32 count;
0cadcf80 715 bool minmod;
40a82448
DM
716 regnode *A, *B; /* the nodes corresponding to /A*B/ */
717 regnode *me; /* the curlym node */
79a2a0e8
KW
718 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
719 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
720 } curlym;
721
722 struct {
3b6647e0 723 U32 paren;
c255a977 724 CHECKPOINT cp;
f6033a9d
DM
725 U32 lastparen;
726 U32 lastcloseparen;
79a2a0e8 727 int c1, c2; /* case fold search */
c255a977
DM
728 char *maxpos; /* highest possible point in string to match */
729 char *oldloc; /* the previous locinput */
e822a8b4 730 int count;
c255a977
DM
731 int min, max; /* {m,n} */
732 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
733 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
734 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 735 } curly; /* and CURLYN/PLUS/STAR */
dad79028 736
d8319b27 737 } u;
5d9a96ca
DM
738} regmatch_state;
739
740/* how many regmatch_state structs to allocate as a single slab.
741 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
742 * pointers, plus 1 for any mythical malloc overhead. */
743
744#define PERL_REGMATCH_SLAB_SLOTS \
745 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
746
747typedef struct regmatch_slab {
748 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
749 struct regmatch_slab *prev, *next;
750} regmatch_slab;
1ade1aa1 751
46ab3289
NC
752#define PL_reg_flags PL_reg_state.re_state_reg_flags
753#define PL_bostr PL_reg_state.re_state_bostr
46ab3289 754#define PL_regeol PL_reg_state.re_state_regeol
46ab3289
NC
755#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
756#define PL_reg_magic PL_reg_state.re_state_reg_magic
757#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
758#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
759#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
760#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
761#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
6502e081
DM
762#define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset
763#define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset
46ab3289
NC
764#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
765#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
766#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
767#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
46ab3289
NC
768#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
769#define PL_nrs PL_reg_state.re_state_nrs
770
1ade1aa1
NC
771struct re_save_state {
772 U32 re_state_reg_flags; /* from regexec.c */
ed301438 773 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 774 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 775 bool re_reparsing; /* runtime (?{}) fed back into parser */
1ade1aa1 776 char *re_state_bostr;
1ade1aa1 777 char *re_state_regeol; /* End of input, for $ check. */
1ade1aa1 778 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
779 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
780 PMOP *re_state_reg_curpm; /* from regexec.c */
781 char *re_state_reg_oldsaved; /* old saved substr during match */
782 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
6502e081
DM
783 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */
784 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */
1b5c067c
NC
785 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
786 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
787 I32 re_state_reg_maxiter; /* max wait until caching pos */
788 I32 re_state_reg_leftiter; /* wait until caching pos */
1b5c067c 789 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1 790 char *re_state_reg_starttry; /* from regexec.c */
db2c6cb3 791#ifdef PERL_ANY_COW
46ab3289 792 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
793#endif
794};
795
796#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
797 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 798
1ade1aa1
NC
799/*
800 * Local variables:
801 * c-indentation-style: bsd
802 * c-basic-offset: 4
14d04a33 803 * indent-tabs-mode: nil
1ade1aa1
NC
804 * End:
805 *
14d04a33 806 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 807 */