This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add RXf_UNBOUNDED_QUANTIFIER and regexp->maxlen
[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 {
49f55535
FC
39 SSize_t min_offset;
40 SSize_t max_offset;
785a26d5
YO
41 SV *substr; /* non-utf8 variant */
42 SV *utf8_substr; /* utf8 variant */
ea3daa5d 43 SSize_t end_shift;
785a26d5
YO
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 57typedef struct regexp_paren_pair {
99a90e59
FC
58 SSize_t start;
59 SSize_t 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. */
ea3daa5d 66 SSize_t 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 */ \
ea3daa5d
FC
107 SSize_t minlen; /* mininum possible number of chars in string to match */\
108 SSize_t minlenret; /* mininum possible number of chars in $& */ \
109 STRLEN gofs; /* chars left of pos that we search from */ \
08e44740
NC
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 */ \
08e44740
NC
122 /* Array of offsets for (@-) and (@+) */ \
123 regexp_paren_pair *offs; \
124 /* saved or original string so \digit works forever. */ \
125 char *subbeg; \
126 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
19394178 127 SSize_t sublen; /* Length of string pointed by subbeg */ \
ea3daa5d
FC
128 SSize_t suboffset; /* byte offset of subbeg from logical start of str */ \
129 SSize_t subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740 130 /* Information about the match that isn't often used */ \
ee273784 131 SSize_t maxlen; /* mininum possible number of chars in string to match */\
08e44740 132 /* offset from wrapped to the start of precomp */ \
654eccd5 133 PERL_BITFIELD32 pre_prefix:4; \
dbc200c5
YO
134 /* original flags used to compile the pattern, may differ */ \
135 /* from extflags in various ways */ \
136 PERL_BITFIELD32 compflags:9; \
b82ad30c 137 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 138
bbe252da 139typedef struct regexp {
288b8c02 140 _XPV_HEAD;
08e44740 141 _REGEXP_COMMON;
f8fc2ecf
YO
142} regexp;
143
89f6f287 144#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 145
785a26d5 146/* used for high speed searches */
f9f4320a
YO
147typedef struct re_scream_pos_data_s
148{
149 char **scream_olds; /* match pos */
ea3daa5d 150 SSize_t *scream_pos; /* Internal iterator of scream. */
f9f4320a
YO
151} re_scream_pos_data;
152
785a26d5
YO
153/* regexp_engine structure. This is the dispatch table for regexes.
154 * Any regex engine implementation must be able to build one of these.
155 */
f9f4320a 156typedef struct regexp_engine {
1593ad57 157 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc 158 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
ea3daa5d 159 char* strbeg, SSize_t minend, SV* sv,
49d7dfbc 160 void* data, U32 flags);
52a21eb3
DM
161 char* (*intuit) (pTHX_
162 REGEXP * const rx,
163 SV *sv,
164 const char * const strbeg,
165 char *strpos,
166 char *strend,
167 const U32 flags,
9f61653a 168 re_scream_pos_data *data);
49d7dfbc
AB
169 SV* (*checkstr) (pTHX_ REGEXP * const rx);
170 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 171 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 172 SV * const sv);
2fdbfb4d
AB
173 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
174 SV const * const value);
175 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
176 const I32 paren);
192b9cd1
AB
177 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
178 SV * const value, const U32 flags);
179 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
180 const U32 flags);
49d7dfbc 181 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 182#ifdef USE_ITHREADS
49d7dfbc 183 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 184#endif
3c13cae6
DM
185 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
186 OP *expr, const struct regexp_engine* eng,
187 REGEXP *VOL old_re,
76ac488f 188 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
189} regexp_engine;
190
192b9cd1
AB
191/*
192 These are passed to the numbered capture variable callbacks as the
193 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
194 $2 etc.
195*/
2c7b5d76
DM
196#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
197#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
198#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
199#define RX_BUFF_IDX_PREMATCH -2 /* $` */
200#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
201#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
202
203/*
204 Flags that are passed to the named_buff and named_buff_iter
205 callbacks above. Those routines are called from universal.c via the
206 Tie::Hash::NamedCapture interface for %+ and %- and the re::
207 functions in the same file.
208*/
209
210/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
211#define RXapif_FETCH 0x0001
212#define RXapif_STORE 0x0002
213#define RXapif_DELETE 0x0004
214#define RXapif_CLEAR 0x0008
215#define RXapif_EXISTS 0x0010
216#define RXapif_SCALAR 0x0020
217#define RXapif_FIRSTKEY 0x0040
218#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
219
220/* Whether %+ or %- is being operated on */
f1b875a0
YO
221#define RXapif_ONE 0x0100 /* %+ */
222#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
223
224/* Whether this is being called from a re:: function */
f1b875a0
YO
225#define RXapif_REGNAME 0x0400
226#define RXapif_REGNAMES 0x0800
7948fc08 227#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 228
f7e71195
AB
229/*
230=head1 REGEXP Functions
231
232=for apidoc Am|REGEXP *|SvRX|SV *sv
233
72d33970 234Convenience macro to get the REGEXP from a SV. This is approximately
f7e71195
AB
235equivalent to the following snippet:
236
237 if (SvMAGICAL(sv))
238 mg_get(sv);
1af910ea
DL
239 if (SvROK(sv))
240 sv = MUTABLE_SV(SvRV(sv));
241 if (SvTYPE(sv) == SVt_REGEXP)
242 return (REGEXP*) sv;
f7e71195
AB
243
244NULL will be returned if a REGEXP* is not found.
245
246=for apidoc Am|bool|SvRXOK|SV* sv
247
1af910ea
DL
248Returns a boolean indicating whether the SV (or the one it references)
249is a REGEXP.
f7e71195
AB
250
251If you want to do something with the REGEXP* later use SvRX instead
252and check for NULL.
253
254=cut
255*/
256
257#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
258#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
259
260
7948fc08 261/* Flags stored in regexp->extflags
bbe252da 262 * These are used by code external to the regexp engine
e357fc67 263 *
5b126c84
KW
264 * Note that the flags whose names start with RXf_PMf_ are defined in
265 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 266 *
eb2624c9
FC
267 * NOTE: if you modify any RXf flags you should run regen.pl or
268 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 269 *
bbe252da
YO
270 */
271
5b126c84
KW
272#include "op_reg_common.h"
273
cde0cee5 274#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 275
a20207d7
YO
276#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
277 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
278 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
279 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
280 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 281
94b03d7d 282/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 283#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 284 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 285
bcdf7404 286/* chars and strings used as regex pattern modifiers
486ec47a 287 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
288 *
289 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
290 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
291 * for its own uses. So 'k' is out as well.
bcdf7404 292 */
85508812 293#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 294#define EXEC_PAT_MOD 'e'
87e95b7f 295#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
296#define ONCE_PAT_MOD 'o'
297#define GLOBAL_PAT_MOD 'g'
298#define CONTINUE_PAT_MOD 'c'
299#define MULTILINE_PAT_MOD 'm'
300#define SINGLE_PAT_MOD 's'
301#define IGNORE_PAT_MOD 'i'
302#define XTENDED_PAT_MOD 'x'
4f4d7508 303#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
304#define LOCALE_PAT_MOD 'l'
305#define UNICODE_PAT_MOD 'u'
50e91148 306#define DEPENDS_PAT_MOD 'd'
cfaf538b 307#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
308
309#define ONCE_PAT_MODS "o"
87e95b7f 310#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
311#define EXEC_PAT_MODS "e"
312#define LOOP_PAT_MODS "gc"
4f4d7508 313#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
314#define LOCALE_PAT_MODS "l"
315#define UNICODE_PAT_MODS "u"
50e91148 316#define DEPENDS_PAT_MODS "d"
cfaf538b 317#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 318#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 319
78fecfc8 320/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
321 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
322 * character is bit +1, etc. */
bcdf7404
YO
323#define STD_PAT_MODS "msix"
324
94b03d7d
KW
325#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
326
78fecfc8 327/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
328 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
329 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
330#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
331
332#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 333#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 334#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 335#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 336
cb5027f2 337/*
eb2624c9
FC
338 * NOTE: if you modify any RXf flags you should run regen.pl or
339 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
340 *
341 */
bcdf7404 342
e795e964
KW
343/* Leave some space, so future bit allocations can go either in the shared or
344 * unshared area without affecting binary compatibility */
dbc200c5
YO
345#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT)
346
347/*
348 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
349 be used by regex engines to check whether they should set
350 RXf_SKIPWHITE
351*/
352#define RXf_SPLIT (1<<(RXf_BASE_SHIFT-1))
353#if RXf_SPLIT != RXf_PMf_SPLIT
354# error "RXf_SPLIT does not match RXf_PMf_SPLIT"
355#endif
f4a4cecb 356
bf28aba1
KW
357/* Manually decorate this function with gcc-style attributes just to
358 * avoid having to restructure the header files and their called order,
359 * as proto.h would have to be included before this file, and isn't */
360
a62b1201
KW
361PERL_STATIC_INLINE const char *
362get_regex_charset_name(const U32 flags, STRLEN* const lenp)
363 __attribute__warn_unused_result__;
364
df7a8460 365#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
366
367PERL_STATIC_INLINE const char *
368get_regex_charset_name(const U32 flags, STRLEN* const lenp)
369{
370 /* Returns a string that corresponds to the name of the regex character set
371 * given by 'flags', and *lenp is set the length of that string, which
372 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
373
374 *lenp = 1;
375 switch (get_regex_charset(flags)) {
50e91148 376 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
377 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
378 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 379 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 380 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
381 *lenp = 2;
382 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
383 default:
384 return "?"; /* Unknown */
a62b1201 385 }
a62b1201
KW
386}
387
8e1490ee
YO
388/* Do we have some sort of anchor? */
389#define RXf_IS_ANCHORED (1<<(RXf_BASE_SHIFT+0))
390#define RXf_UNUSED1 (1<<(RXf_BASE_SHIFT+1))
391#define RXf_UNUSED2 (1<<(RXf_BASE_SHIFT+2))
392#define RXf_UNUSED3 (1<<(RXf_BASE_SHIFT+3))
5e11cd63
YO
393#define RXf_UNUSED4 (1<<(RXf_BASE_SHIFT+4))
394#define RXf_UNUSED5 (1<<(RXf_BASE_SHIFT+5))
52d81aa8 395
bbe252da 396/* What we have seen */
dbc200c5 397#define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+6))
1ad4ec54 398#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
5e11cd63 399#define RXf_UNUSED8 (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
400
401/* Special */
ee273784 402#define RXf_UNBOUNDED_QUANTIFIER_SEEN (1<<(RXf_BASE_SHIFT+9))
1ad4ec54 403#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
404
405/* UTF8 related */
0603fe5c 406#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) /* $1 etc are utf8 */
bbe252da
YO
407
408/* Intuit related */
e795e964
KW
409#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
410#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
411#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
412#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
413
414/* Copy and tainted info */
e795e964 415#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
416
417/* during execution: pattern temporarily tainted by executing locale ops;
418 * post-execution: $1 et al are tainted */
e795e964 419#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
420/* this pattern was tainted during compilation */
421#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
422
423/* Flags indicating special patterns */
e795e964 424#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
dbc200c5 425#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split " " */
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
272d35c9
DM
441# define RXp_MATCH_TAINTED(prog) 0
442# define RX_MATCH_TAINTED(prog) 0
443# define RXp_MATCH_TAINTED_on(prog) NOOP
444# define RX_MATCH_TAINTED_on(prog) NOOP
445# define RX_MATCH_TAINTED_off(prog) NOOP
284167a5
SM
446#else
447# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
448# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
272d35c9
DM
449# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
450# define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
451# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
452# define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
453# define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
284167a5
SM
454#endif
455
bbe252da 456#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
72311751
GS
457#define RX_MATCH_TAINTED_set(prog, t) ((t) \
458 ? RX_MATCH_TAINTED_on(prog) \
459 : RX_MATCH_TAINTED_off(prog))
c277df42 460
07bc277f
NC
461#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
462#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
463#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
464#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
465#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
466#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
467#define RX_MATCH_COPIED_set(prog,t) ((t) \
468 ? RX_MATCH_COPIED_on(prog) \
469 : RX_MATCH_COPIED_off(prog))
f9f4320a 470
07bc277f 471#define RXp_EXTFLAGS(rx) ((rx)->extflags)
dbc200c5 472#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 473
07bc277f 474/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
475#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
476#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
477/* FIXME? Are we hardcoding too much here and constraining plugin extension
478 writers? Specifically, the value 1 assumes that the wrapped version always
479 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
480#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
481#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
482#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 483#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 484#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 485#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a 486#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
dbc200c5 487#define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog))
8d919b0a
FC
488#define RX_ENGINE(prog) (ReANY(prog)->engine)
489#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
490#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
491#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
492#define RX_OFFS(prog) (ReANY(prog)->offs)
493#define RX_NPARENS(prog) (ReANY(prog)->nparens)
494#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
495#define RX_MINLEN(prog) (ReANY(prog)->minlen)
496#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
497#define RX_GOFS(prog) (ReANY(prog)->gofs)
498#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
499#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
500#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
03c83e26
DM
501/* last match was zero-length */
502#define RX_ZERO_LEN(prog) \
99a90e59
FC
503 (RX_OFFS(prog)[0].start + (SSize_t)RX_GOFS(prog) \
504 == RX_OFFS(prog)[0].end)
220fc49f 505
be8e71aa
YO
506#endif /* PLUGGABLE_RE_EXTENSION */
507
486ec47a 508/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 509
db2c6cb3 510#ifdef PERL_ANY_COW
ed252734 511#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
512 STMT_START {if (RX_SAVED_COPY(rx)) { \
513 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
514 } \
515 if (RX_MATCH_COPIED(rx)) { \
07bc277f 516 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
517 RX_MATCH_COPIED_off(rx); \
518 }} STMT_END
519#else
520#define RX_MATCH_COPY_FREE(rx) \
521 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 522 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
523 RX_MATCH_COPIED_off(rx); \
524 }} STMT_END
525#endif
526
07bc277f
NC
527#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
528#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
529#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
530#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f 531#define RX_MATCH_UTF8_set(prog, t) ((t) \
0603fe5c
DM
532 ? RX_MATCH_UTF8_on(prog) \
533 : RX_MATCH_UTF8_off(prog))
efd26800
NC
534
535/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 536#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 537
a340edde
DM
538
539/* bits in flags arg of Perl_regexec_flags() */
540
541#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
542#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
543#define REXEC_SCREAM 0x04 /* currently unused. */
544#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
545#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
546 no need to copy string again */
547
548 /* under REXEC_COPY_STR, it's ok for the
549 engine (modulo PL_sawamperand etc)
550 to skip copying: ... */
551#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
552#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
553#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
554 the start pos (so s/.\G// would fail
555 on second iteration */
c277df42 556
7fe44985 557#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
558# define ReREFCNT_inc(re) \
559 ({ \
560 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
561 REGEXP *const _rerefcnt_inc = (re); \
562 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
563 SvREFCNT_inc(_rerefcnt_inc); \
564 _rerefcnt_inc; \
288b8c02
NC
565 })
566# define ReREFCNT_dec(re) \
567 ({ \
568 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
569 REGEXP *const _rerefcnt_dec = (re); \
570 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
571 })
572#else
573# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 574# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 575#endif
8d919b0a 576#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
577
578/* FIXME for plugins. */
cf93c79d
IZ
579
580#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
581#define FBMcf_TAIL_DOLLARM 2
582#define FBMcf_TAIL_Z 4
583#define FBMcf_TAIL_z 8
584#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
585
586#define FBMrf_MULTILINE 1
cad2e5aa 587
331b2dcc
DM
588struct regmatch_state;
589struct regmatch_slab;
8adc0f72 590
bf2039a9
DM
591/* like regmatch_info_aux, but contains extra fields only needed if the
592 * pattern contains (?{}). If used, is snuck into the second slot in the
593 * regmatch_state stack at the start of execution */
8adc0f72
DM
594
595typedef struct {
596 regexp *rex;
597 PMOP *curpm; /* saved PL_curpm */
598#ifdef PERL_ANY_COW
599 SV *saved_copy; /* saved saved_copy field from rex */
600#endif
601 char *subbeg; /* saved subbeg field from rex */
602 STRLEN sublen; /* saved sublen field from rex */
603 STRLEN suboffset; /* saved suboffset field from rex */
604 STRLEN subcoffset; /* saved subcoffset field from rex */
605 MAGIC *pos_magic; /* pos() magic attached to $_ */
ea3daa5d 606 SSize_t pos; /* the original value of pos() in pos_magic */
25fdce4a 607 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
608} regmatch_info_aux_eval;
609
610
611/* fields that logically live in regmatch_info, but which need cleaning
612 * up on croak(), and so are instead are snuck into the first slot in
613 * the regmatch_state stack at the start of execution */
614
615typedef struct {
616 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
617 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
618 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 619 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
620} regmatch_info_aux;
621
8adc0f72 622
3b0527fe 623/* some basic information about the current match that is created by
bf2039a9
DM
624 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
625 * It is allocated as a local var on the stack, so nothing should be
626 * stored in it that needs preserving or clearing up on croak().
627 * For that, see the aux_info and aux_info_eval members of the
628 * regmatch_state union. */
3b0527fe
DM
629
630typedef struct {
28d03b21 631 REGEXP *prog; /* the regex being executed */
9d9163fb 632 const char * strbeg; /* real start of string */
28d03b21
DM
633 char *strend; /* one byte beyond last char of match string */
634 char *till; /* matches shorter than this fail (see minlen arg) */
635 SV *sv; /* the SV string currently being matched */
636 char *ganch; /* position of \G anchor */
637 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
638 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
639 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
640 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
641 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 642 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 643 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
644 bool is_utf8_pat; /* regex is utf8 */
645 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 646 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
647} regmatch_info;
648
5d9a96ca
DM
649
650/* structures for holding and saving the state maintained by regmatch() */
651
d56b3014 652#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 653#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 654#endif
6bda09f9 655
5d9a96ca
DM
656typedef I32 CHECKPOINT;
657
a0374537 658typedef struct regmatch_state {
40a82448 659 int resume_state; /* where to jump to on return */
24d3c4a9 660 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
661
662 union {
77cb431f 663
bf2039a9
DM
664 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
665 * the nest. They aren't saved backtrack state; rather they
666 * represent one or two extra chunks of data that need allocating
667 * at the start of a match. These fields would logically live in
668 * the regmatch_info struct, except that is allocated on the
669 * C stack, and these fields are all things that require cleanup
670 * after a croak(), when the stack is lost.
671 * As a convenience, we just use the first 1 or 2 regmatch_state
672 * slots to store this info, as we will be allocating a slab of
673 * these anyway. Otherwise we'd have to malloc and then free them,
674 * or allocate them on the save stack (where they will get
675 * realloced if the save stack grows).
676 * info_aux contains the extra fields that are always needed;
677 * info_aux_eval contains extra fields that only needed if
678 * the pattern contains code blocks
679 * We split them into two separate structs to avoid increasing
680 * the size of the union.
681 */
682
683 regmatch_info_aux info_aux;
684
685 regmatch_info_aux_eval info_aux_eval;
686
77cb431f
DM
687 /* this is a fake union member that matches the first element
688 * of each member that needs to store positive backtrack
689 * information */
690 struct {
691 struct regmatch_state *prev_yes_state;
692 } yes;
693
fae667d5
YO
694 /* branchlike members */
695 /* this is a fake union member that matches the first elements
696 * of each member that needs to behave like a branch */
697 struct {
5d458dd8
YO
698 /* this first element must match u.yes */
699 struct regmatch_state *prev_yes_state;
fae667d5 700 U32 lastparen;
f6033a9d 701 U32 lastcloseparen;
fae667d5
YO
702 CHECKPOINT cp;
703
704 } branchlike;
705
706 struct {
707 /* the first elements must match u.branchlike */
708 struct regmatch_state *prev_yes_state;
709 U32 lastparen;
f6033a9d 710 U32 lastcloseparen;
fae667d5
YO
711 CHECKPOINT cp;
712
713 regnode *next_branch; /* next branch node */
714 } branch;
715
716 struct {
717 /* the first elements must match u.branchlike */
718 struct regmatch_state *prev_yes_state;
719 U32 lastparen;
f6033a9d 720 U32 lastcloseparen;
fae667d5
YO
721 CHECKPOINT cp;
722
2e64971a 723 U32 accepted; /* how many accepting states left */
6603fe3e 724 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 725 U16 *jump; /* positive offsets from me */
7f69552c 726 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
727 U8 *firstpos;/* pos in string of first trie match */
728 U32 firstchars;/* len in chars of firstpos from start */
729 U16 nextword;/* next word to try */
730 U16 topword; /* longest accepted word */
e822a8b4
DM
731 } trie;
732
fae667d5
YO
733 /* special types - these members are used to store state for special
734 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 735 struct {
77cb431f
DM
736 /* this first element must match u.yes */
737 struct regmatch_state *prev_yes_state;
faec1544
DM
738 struct regmatch_state *prev_eval;
739 struct regmatch_state *prev_curlyx;
288b8c02 740 REGEXP *prev_rex;
aa283a38
DM
741 CHECKPOINT cp; /* remember current savestack indexes */
742 CHECKPOINT lastcp;
6bda09f9 743 U32 close_paren; /* which close bracket is our end */
1b5c067c 744 regnode *B; /* the node following us */
e822a8b4
DM
745 } eval;
746
747 struct {
c476f425 748 /* this first element must match u.yes */
262b90c4 749 struct regmatch_state *prev_yes_state;
fae667d5
YO
750 I32 wanted;
751 I32 logical; /* saved copy of 'logical' var */
752 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
753 } ifmatch; /* and SUSPEND/UNLESSM */
754
755 struct {
756 /* this first element must match u.yes */
757 struct regmatch_state *prev_yes_state;
758 struct regmatch_state *prev_mark;
759 SV* mark_name;
760 char *mark_loc;
761 } mark;
762
763 struct {
764 int val;
765 } keeper;
766
767 /* quantifiers - these members are used for storing state for
768 for the regops used to implement quantifiers */
769 struct {
770 /* this first element must match u.yes */
771 struct regmatch_state *prev_yes_state;
c476f425 772 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
773 regnode *me; /* the CURLYX node */
774 regnode *B; /* the B node in /A*B/ */
c476f425
DM
775 CHECKPOINT cp; /* remember current savestack index */
776 bool minmod;
a0374537 777 int parenfloor;/* how far back to strip paren data */
c476f425
DM
778
779 /* these two are modified by WHILEM */
780 int count; /* how many instances of A we've matched */
781 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
782 } curlyx;
783
784 struct {
c476f425 785 /* this first element must match u.yes */
262b90c4 786 struct regmatch_state *prev_yes_state;
c476f425
DM
787 struct regmatch_state *save_curlyx;
788 CHECKPOINT cp; /* remember current savestack indexes */
789 CHECKPOINT lastcp;
790 char *save_lastloc; /* previous curlyx.lastloc */
791 I32 cache_offset;
792 I32 cache_mask;
e822a8b4
DM
793 } whilem;
794
795 struct {
5d458dd8
YO
796 /* this first element must match u.yes */
797 struct regmatch_state *prev_yes_state;
79a2a0e8 798 int c1, c2; /* case fold search */
40a82448 799 CHECKPOINT cp;
f6033a9d
DM
800 U32 lastparen;
801 U32 lastcloseparen;
40a82448
DM
802 I32 alen; /* length of first-matched A string */
803 I32 count;
0cadcf80 804 bool minmod;
40a82448
DM
805 regnode *A, *B; /* the nodes corresponding to /A*B/ */
806 regnode *me; /* the curlym node */
79a2a0e8
KW
807 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
808 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
809 } curlym;
810
811 struct {
3b6647e0 812 U32 paren;
c255a977 813 CHECKPOINT cp;
f6033a9d
DM
814 U32 lastparen;
815 U32 lastcloseparen;
79a2a0e8 816 int c1, c2; /* case fold search */
c255a977
DM
817 char *maxpos; /* highest possible point in string to match */
818 char *oldloc; /* the previous locinput */
e822a8b4 819 int count;
c255a977
DM
820 int min, max; /* {m,n} */
821 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
822 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
823 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 824 } curly; /* and CURLYN/PLUS/STAR */
dad79028 825
d8319b27 826 } u;
5d9a96ca
DM
827} regmatch_state;
828
829/* how many regmatch_state structs to allocate as a single slab.
830 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
831 * pointers, plus 1 for any mythical malloc overhead. */
832
833#define PERL_REGMATCH_SLAB_SLOTS \
834 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
835
836typedef struct regmatch_slab {
837 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
838 struct regmatch_slab *prev, *next;
839} regmatch_slab;
1ade1aa1 840
46ab3289 841
cde0cee5 842
1ade1aa1
NC
843/*
844 * Local variables:
845 * c-indentation-style: bsd
846 * c-basic-offset: 4
14d04a33 847 * indent-tabs-mode: nil
1ade1aa1
NC
848 * End:
849 *
14d04a33 850 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 851 */