This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Stop minlen regexp optimisation from rejecting long strings
[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 */ \
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 */\
127 I32 sublen; /* Length of string pointed by subbeg */ \
6502e081
DM
128 I32 suboffset; /* byte offset of subbeg from logical start of str */ \
129 I32 subcoffset; /* suboffset equiv, but in chars (for @-/@+) */ \
08e44740
NC
130 /* Information about the match that isn't often used */ \
131 /* offset from wrapped to the start of precomp */ \
654eccd5 132 PERL_BITFIELD32 pre_prefix:4; \
dbc200c5
YO
133 /* original flags used to compile the pattern, may differ */ \
134 /* from extflags in various ways */ \
135 PERL_BITFIELD32 compflags:9; \
b82ad30c 136 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 137
bbe252da 138typedef struct regexp {
288b8c02 139 _XPV_HEAD;
08e44740 140 _REGEXP_COMMON;
f8fc2ecf
YO
141} regexp;
142
89f6f287 143#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 144
785a26d5 145/* used for high speed searches */
f9f4320a
YO
146typedef struct re_scream_pos_data_s
147{
148 char **scream_olds; /* match pos */
149 I32 *scream_pos; /* Internal iterator of scream. */
150} re_scream_pos_data;
151
785a26d5
YO
152/* regexp_engine structure. This is the dispatch table for regexes.
153 * Any regex engine implementation must be able to build one of these.
154 */
f9f4320a 155typedef struct regexp_engine {
1593ad57 156 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc 157 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
a340edde 158 char* strbeg, I32 minend, SV* sv,
49d7dfbc 159 void* data, U32 flags);
52a21eb3
DM
160 char* (*intuit) (pTHX_
161 REGEXP * const rx,
162 SV *sv,
163 const char * const strbeg,
164 char *strpos,
165 char *strend,
166 const U32 flags,
9f61653a 167 re_scream_pos_data *data);
49d7dfbc
AB
168 SV* (*checkstr) (pTHX_ REGEXP * const rx);
169 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 170 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 171 SV * const sv);
2fdbfb4d
AB
172 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
173 SV const * const value);
174 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
175 const I32 paren);
192b9cd1
AB
176 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
177 SV * const value, const U32 flags);
178 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
179 const U32 flags);
49d7dfbc 180 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 181#ifdef USE_ITHREADS
49d7dfbc 182 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 183#endif
3c13cae6
DM
184 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
185 OP *expr, const struct regexp_engine* eng,
186 REGEXP *VOL old_re,
76ac488f 187 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
188} regexp_engine;
189
192b9cd1
AB
190/*
191 These are passed to the numbered capture variable callbacks as the
192 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
193 $2 etc.
194*/
2c7b5d76
DM
195#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
196#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
197#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
198#define RX_BUFF_IDX_PREMATCH -2 /* $` */
199#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
200#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
201
202/*
203 Flags that are passed to the named_buff and named_buff_iter
204 callbacks above. Those routines are called from universal.c via the
205 Tie::Hash::NamedCapture interface for %+ and %- and the re::
206 functions in the same file.
207*/
208
209/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
210#define RXapif_FETCH 0x0001
211#define RXapif_STORE 0x0002
212#define RXapif_DELETE 0x0004
213#define RXapif_CLEAR 0x0008
214#define RXapif_EXISTS 0x0010
215#define RXapif_SCALAR 0x0020
216#define RXapif_FIRSTKEY 0x0040
217#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
218
219/* Whether %+ or %- is being operated on */
f1b875a0
YO
220#define RXapif_ONE 0x0100 /* %+ */
221#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
222
223/* Whether this is being called from a re:: function */
f1b875a0
YO
224#define RXapif_REGNAME 0x0400
225#define RXapif_REGNAMES 0x0800
7948fc08 226#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 227
f7e71195
AB
228/*
229=head1 REGEXP Functions
230
231=for apidoc Am|REGEXP *|SvRX|SV *sv
232
233Convenience macro to get the REGEXP from a SV. This is approximately
234equivalent to the following snippet:
235
236 if (SvMAGICAL(sv))
237 mg_get(sv);
1af910ea
DL
238 if (SvROK(sv))
239 sv = MUTABLE_SV(SvRV(sv));
240 if (SvTYPE(sv) == SVt_REGEXP)
241 return (REGEXP*) sv;
f7e71195
AB
242
243NULL will be returned if a REGEXP* is not found.
244
245=for apidoc Am|bool|SvRXOK|SV* sv
246
1af910ea
DL
247Returns a boolean indicating whether the SV (or the one it references)
248is a REGEXP.
f7e71195
AB
249
250If you want to do something with the REGEXP* later use SvRX instead
251and check for NULL.
252
253=cut
254*/
255
256#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
257#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
258
259
7948fc08 260/* Flags stored in regexp->extflags
bbe252da 261 * These are used by code external to the regexp engine
e357fc67 262 *
5b126c84
KW
263 * Note that the flags whose names start with RXf_PMf_ are defined in
264 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 265 *
eb2624c9
FC
266 * NOTE: if you modify any RXf flags you should run regen.pl or
267 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 268 *
bbe252da
YO
269 */
270
5b126c84
KW
271#include "op_reg_common.h"
272
cde0cee5 273#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 274
a20207d7
YO
275#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
276 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
277 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
278 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
279 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 280
94b03d7d 281/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 282#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 283 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 284
bcdf7404 285/* chars and strings used as regex pattern modifiers
486ec47a 286 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
287 *
288 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
289 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
290 * for its own uses. So 'k' is out as well.
bcdf7404 291 */
85508812 292#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 293#define EXEC_PAT_MOD 'e'
87e95b7f 294#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
295#define ONCE_PAT_MOD 'o'
296#define GLOBAL_PAT_MOD 'g'
297#define CONTINUE_PAT_MOD 'c'
298#define MULTILINE_PAT_MOD 'm'
299#define SINGLE_PAT_MOD 's'
300#define IGNORE_PAT_MOD 'i'
301#define XTENDED_PAT_MOD 'x'
4f4d7508 302#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
303#define LOCALE_PAT_MOD 'l'
304#define UNICODE_PAT_MOD 'u'
50e91148 305#define DEPENDS_PAT_MOD 'd'
cfaf538b 306#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
307
308#define ONCE_PAT_MODS "o"
87e95b7f 309#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
310#define EXEC_PAT_MODS "e"
311#define LOOP_PAT_MODS "gc"
4f4d7508 312#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
313#define LOCALE_PAT_MODS "l"
314#define UNICODE_PAT_MODS "u"
50e91148 315#define DEPENDS_PAT_MODS "d"
cfaf538b 316#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 317#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 318
78fecfc8 319/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
320 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
321 * character is bit +1, etc. */
bcdf7404
YO
322#define STD_PAT_MODS "msix"
323
94b03d7d
KW
324#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
325
78fecfc8 326/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
327 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
328 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
329#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
330
331#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 332#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 333#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 334#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 335
cb5027f2 336/*
eb2624c9
FC
337 * NOTE: if you modify any RXf flags you should run regen.pl or
338 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
339 *
340 */
bcdf7404 341
e795e964
KW
342/* Leave some space, so future bit allocations can go either in the shared or
343 * unshared area without affecting binary compatibility */
dbc200c5
YO
344#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT)
345
346/*
347 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
348 be used by regex engines to check whether they should set
349 RXf_SKIPWHITE
350*/
351#define RXf_SPLIT (1<<(RXf_BASE_SHIFT-1))
352#if RXf_SPLIT != RXf_PMf_SPLIT
353# error "RXf_SPLIT does not match RXf_PMf_SPLIT"
354#endif
f4a4cecb 355
bf28aba1
KW
356/* Manually decorate this function with gcc-style attributes just to
357 * avoid having to restructure the header files and their called order,
358 * as proto.h would have to be included before this file, and isn't */
359
a62b1201
KW
360PERL_STATIC_INLINE const char *
361get_regex_charset_name(const U32 flags, STRLEN* const lenp)
362 __attribute__warn_unused_result__;
363
df7a8460 364#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
365
366PERL_STATIC_INLINE const char *
367get_regex_charset_name(const U32 flags, STRLEN* const lenp)
368{
369 /* Returns a string that corresponds to the name of the regex character set
370 * given by 'flags', and *lenp is set the length of that string, which
371 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
372
373 *lenp = 1;
374 switch (get_regex_charset(flags)) {
50e91148 375 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
376 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
377 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 378 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 379 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
380 *lenp = 2;
381 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
382 default:
383 return "?"; /* Unknown */
a62b1201 384 }
a62b1201
KW
385}
386
52d81aa8 387/* Anchor and GPOS related stuff */
1ad4ec54
KW
388#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
389#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
390#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
391#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
392#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
393#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
394/* two bits here */
395#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
396#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
397#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
398
bbe252da 399/* What we have seen */
dbc200c5 400#define RXf_NO_INPLACE_SUBST (1<<(RXf_BASE_SHIFT+6))
1ad4ec54
KW
401#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
402#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
403
404/* Special */
1ad4ec54
KW
405#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
406#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
407
408/* UTF8 related */
0603fe5c 409#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) /* $1 etc are utf8 */
bbe252da
YO
410
411/* Intuit related */
e795e964
KW
412#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
413#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
414#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
415#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
416
417/* Copy and tainted info */
e795e964 418#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16))
ef07e810
DM
419
420/* during execution: pattern temporarily tainted by executing locale ops;
421 * post-execution: $1 et al are tainted */
e795e964 422#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17))
ef07e810
DM
423/* this pattern was tainted during compilation */
424#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18))
52d81aa8
NC
425
426/* Flags indicating special patterns */
e795e964 427#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */
dbc200c5 428#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split " " */
e795e964 429#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 430#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 431#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
432# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
433#endif
c737faaf 434
cb5027f2 435/*
eb2624c9
FC
436 * NOTE: if you modify any RXf flags you should run regen.pl or
437 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
438 *
439 */
bbe252da 440
284167a5
SM
441#if NO_TAINT_SUPPORT
442# define RX_ISTAINTED(prog) 0
443# define RX_TAINT_on(prog) NOOP
272d35c9
DM
444# define RXp_MATCH_TAINTED(prog) 0
445# define RX_MATCH_TAINTED(prog) 0
446# define RXp_MATCH_TAINTED_on(prog) NOOP
447# define RX_MATCH_TAINTED_on(prog) NOOP
448# define RX_MATCH_TAINTED_off(prog) NOOP
284167a5
SM
449#else
450# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
451# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
272d35c9
DM
452# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
453# define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
454# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
455# define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
456# define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
284167a5
SM
457#endif
458
bbe252da 459#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
72311751
GS
460#define RX_MATCH_TAINTED_set(prog, t) ((t) \
461 ? RX_MATCH_TAINTED_on(prog) \
462 : RX_MATCH_TAINTED_off(prog))
c277df42 463
07bc277f
NC
464#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
465#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
466#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
467#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
468#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
469#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
470#define RX_MATCH_COPIED_set(prog,t) ((t) \
471 ? RX_MATCH_COPIED_on(prog) \
472 : RX_MATCH_COPIED_off(prog))
f9f4320a 473
07bc277f 474#define RXp_EXTFLAGS(rx) ((rx)->extflags)
dbc200c5 475#define RXp_COMPFLAGS(rx) ((rx)->compflags)
866c78d1 476
07bc277f 477/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
478#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
479#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
480/* FIXME? Are we hardcoding too much here and constraining plugin extension
481 writers? Specifically, the value 1 assumes that the wrapped version always
482 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
483#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
484#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
485#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 486#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 487#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 488#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a 489#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
dbc200c5 490#define RX_COMPFLAGS(prog) RXp_COMPFLAGS(ReANY(prog))
8d919b0a
FC
491#define RX_ENGINE(prog) (ReANY(prog)->engine)
492#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
493#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
494#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
495#define RX_OFFS(prog) (ReANY(prog)->offs)
496#define RX_NPARENS(prog) (ReANY(prog)->nparens)
497#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
498#define RX_MINLEN(prog) (ReANY(prog)->minlen)
499#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
500#define RX_GOFS(prog) (ReANY(prog)->gofs)
501#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
502#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
503#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
03c83e26
DM
504/* last match was zero-length */
505#define RX_ZERO_LEN(prog) \
506 (RX_OFFS(prog)[0].start + RX_GOFS(prog) == (UV)RX_OFFS(prog)[0].end)
220fc49f 507
be8e71aa
YO
508#endif /* PLUGGABLE_RE_EXTENSION */
509
486ec47a 510/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 511
db2c6cb3 512#ifdef PERL_ANY_COW
ed252734 513#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
514 STMT_START {if (RX_SAVED_COPY(rx)) { \
515 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
516 } \
517 if (RX_MATCH_COPIED(rx)) { \
07bc277f 518 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
519 RX_MATCH_COPIED_off(rx); \
520 }} STMT_END
521#else
522#define RX_MATCH_COPY_FREE(rx) \
523 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 524 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
525 RX_MATCH_COPIED_off(rx); \
526 }} STMT_END
527#endif
528
07bc277f
NC
529#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
530#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
531#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
532#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f 533#define RX_MATCH_UTF8_set(prog, t) ((t) \
0603fe5c
DM
534 ? RX_MATCH_UTF8_on(prog) \
535 : RX_MATCH_UTF8_off(prog))
efd26800
NC
536
537/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 538#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 539
a340edde
DM
540
541/* bits in flags arg of Perl_regexec_flags() */
542
543#define REXEC_COPY_STR 0x01 /* Need to copy the string for captures. */
544#define REXEC_CHECKED 0x02 /* re_intuit_start() already called. */
545#define REXEC_SCREAM 0x04 /* currently unused. */
546#define REXEC_IGNOREPOS 0x08 /* use stringarg, not pos(), for \G match */
547#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g:
548 no need to copy string again */
549
550 /* under REXEC_COPY_STR, it's ok for the
551 engine (modulo PL_sawamperand etc)
552 to skip copying: ... */
553#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
554#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
d5e7783a
DM
555#define REXEC_FAIL_ON_UNDERFLOW 0x80 /* fail the match if $& would start before
556 the start pos (so s/.\G// would fail
557 on second iteration */
c277df42 558
7fe44985 559#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
560# define ReREFCNT_inc(re) \
561 ({ \
562 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
563 REGEXP *const _rerefcnt_inc = (re); \
564 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
565 SvREFCNT_inc(_rerefcnt_inc); \
566 _rerefcnt_inc; \
288b8c02
NC
567 })
568# define ReREFCNT_dec(re) \
569 ({ \
570 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
571 REGEXP *const _rerefcnt_dec = (re); \
572 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
573 })
574#else
575# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 576# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 577#endif
8d919b0a 578#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
579
580/* FIXME for plugins. */
cf93c79d
IZ
581
582#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
583#define FBMcf_TAIL_DOLLARM 2
584#define FBMcf_TAIL_Z 4
585#define FBMcf_TAIL_z 8
586#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
587
588#define FBMrf_MULTILINE 1
cad2e5aa 589
331b2dcc
DM
590struct regmatch_state;
591struct regmatch_slab;
8adc0f72 592
bf2039a9
DM
593/* like regmatch_info_aux, but contains extra fields only needed if the
594 * pattern contains (?{}). If used, is snuck into the second slot in the
595 * regmatch_state stack at the start of execution */
8adc0f72
DM
596
597typedef struct {
598 regexp *rex;
599 PMOP *curpm; /* saved PL_curpm */
600#ifdef PERL_ANY_COW
601 SV *saved_copy; /* saved saved_copy field from rex */
602#endif
603 char *subbeg; /* saved subbeg field from rex */
604 STRLEN sublen; /* saved sublen field from rex */
605 STRLEN suboffset; /* saved suboffset field from rex */
606 STRLEN subcoffset; /* saved subcoffset field from rex */
607 MAGIC *pos_magic; /* pos() magic attached to $_ */
608 I32 pos; /* the original value of pos() in pos_magic */
25fdce4a 609 U8 pos_flags; /* flags to be restored; currently only MGf_BYTES*/
bf2039a9
DM
610} regmatch_info_aux_eval;
611
612
613/* fields that logically live in regmatch_info, but which need cleaning
614 * up on croak(), and so are instead are snuck into the first slot in
615 * the regmatch_state stack at the start of execution */
616
617typedef struct {
618 regmatch_info_aux_eval *info_aux_eval;
331b2dcc
DM
619 struct regmatch_state *old_regmatch_state; /* saved PL_regmatch_state */
620 struct regmatch_slab *old_regmatch_slab; /* saved PL_regmatch_slab */
2ac8ff4b 621 char *poscache; /* S-L cache of fail positions of WHILEMs */
bf2039a9
DM
622} regmatch_info_aux;
623
8adc0f72 624
3b0527fe 625/* some basic information about the current match that is created by
bf2039a9
DM
626 * Perl_regexec_flags and then passed to regtry(), regmatch() etc.
627 * It is allocated as a local var on the stack, so nothing should be
628 * stored in it that needs preserving or clearing up on croak().
629 * For that, see the aux_info and aux_info_eval members of the
630 * regmatch_state union. */
3b0527fe
DM
631
632typedef struct {
28d03b21 633 REGEXP *prog; /* the regex being executed */
9d9163fb 634 const char * strbeg; /* real start of string */
28d03b21
DM
635 char *strend; /* one byte beyond last char of match string */
636 char *till; /* matches shorter than this fail (see minlen arg) */
637 SV *sv; /* the SV string currently being matched */
638 char *ganch; /* position of \G anchor */
639 char *cutpoint; /* (*COMMIT) position (if any) */
bf2039a9
DM
640 regmatch_info_aux *info_aux; /* extra fields that need cleanup */
641 regmatch_info_aux_eval *info_aux_eval; /* extra saved state for (?{}) */
1cb48e53
DM
642 I32 poscache_maxiter; /* how many whilems todo before S-L cache kicks in */
643 I32 poscache_iter; /* current countdown from _maxiter to zero */
2ac8ff4b 644 STRLEN poscache_size; /* size of regmatch_info_aux.poscache */
02d5137b 645 bool intuit; /* re_intuit_start() is the top-level caller */
ba44c216
DM
646 bool is_utf8_pat; /* regex is utf8 */
647 bool is_utf8_target; /* string being matched is utf8 */
39819bd9 648 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
649} regmatch_info;
650
5d9a96ca
DM
651
652/* structures for holding and saving the state maintained by regmatch() */
653
d56b3014 654#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 655#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 656#endif
6bda09f9 657
5d9a96ca
DM
658typedef I32 CHECKPOINT;
659
a0374537 660typedef struct regmatch_state {
40a82448 661 int resume_state; /* where to jump to on return */
24d3c4a9 662 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
663
664 union {
77cb431f 665
bf2039a9
DM
666 /* the 'info_aux' and 'info_aux_eval' union members are cuckoos in
667 * the nest. They aren't saved backtrack state; rather they
668 * represent one or two extra chunks of data that need allocating
669 * at the start of a match. These fields would logically live in
670 * the regmatch_info struct, except that is allocated on the
671 * C stack, and these fields are all things that require cleanup
672 * after a croak(), when the stack is lost.
673 * As a convenience, we just use the first 1 or 2 regmatch_state
674 * slots to store this info, as we will be allocating a slab of
675 * these anyway. Otherwise we'd have to malloc and then free them,
676 * or allocate them on the save stack (where they will get
677 * realloced if the save stack grows).
678 * info_aux contains the extra fields that are always needed;
679 * info_aux_eval contains extra fields that only needed if
680 * the pattern contains code blocks
681 * We split them into two separate structs to avoid increasing
682 * the size of the union.
683 */
684
685 regmatch_info_aux info_aux;
686
687 regmatch_info_aux_eval info_aux_eval;
688
77cb431f
DM
689 /* this is a fake union member that matches the first element
690 * of each member that needs to store positive backtrack
691 * information */
692 struct {
693 struct regmatch_state *prev_yes_state;
694 } yes;
695
fae667d5
YO
696 /* branchlike members */
697 /* this is a fake union member that matches the first elements
698 * of each member that needs to behave like a branch */
699 struct {
5d458dd8
YO
700 /* this first element must match u.yes */
701 struct regmatch_state *prev_yes_state;
fae667d5 702 U32 lastparen;
f6033a9d 703 U32 lastcloseparen;
fae667d5
YO
704 CHECKPOINT cp;
705
706 } branchlike;
707
708 struct {
709 /* the first elements must match u.branchlike */
710 struct regmatch_state *prev_yes_state;
711 U32 lastparen;
f6033a9d 712 U32 lastcloseparen;
fae667d5
YO
713 CHECKPOINT cp;
714
715 regnode *next_branch; /* next branch node */
716 } branch;
717
718 struct {
719 /* the first elements must match u.branchlike */
720 struct regmatch_state *prev_yes_state;
721 U32 lastparen;
f6033a9d 722 U32 lastcloseparen;
fae667d5
YO
723 CHECKPOINT cp;
724
2e64971a 725 U32 accepted; /* how many accepting states left */
6603fe3e 726 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 727 U16 *jump; /* positive offsets from me */
7f69552c 728 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
729 U8 *firstpos;/* pos in string of first trie match */
730 U32 firstchars;/* len in chars of firstpos from start */
731 U16 nextword;/* next word to try */
732 U16 topword; /* longest accepted word */
e822a8b4
DM
733 } trie;
734
fae667d5
YO
735 /* special types - these members are used to store state for special
736 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 737 struct {
77cb431f
DM
738 /* this first element must match u.yes */
739 struct regmatch_state *prev_yes_state;
faec1544
DM
740 struct regmatch_state *prev_eval;
741 struct regmatch_state *prev_curlyx;
288b8c02 742 REGEXP *prev_rex;
aa283a38
DM
743 CHECKPOINT cp; /* remember current savestack indexes */
744 CHECKPOINT lastcp;
6bda09f9 745 U32 close_paren; /* which close bracket is our end */
1b5c067c 746 regnode *B; /* the node following us */
e822a8b4
DM
747 } eval;
748
749 struct {
c476f425 750 /* this first element must match u.yes */
262b90c4 751 struct regmatch_state *prev_yes_state;
fae667d5
YO
752 I32 wanted;
753 I32 logical; /* saved copy of 'logical' var */
754 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
755 } ifmatch; /* and SUSPEND/UNLESSM */
756
757 struct {
758 /* this first element must match u.yes */
759 struct regmatch_state *prev_yes_state;
760 struct regmatch_state *prev_mark;
761 SV* mark_name;
762 char *mark_loc;
763 } mark;
764
765 struct {
766 int val;
767 } keeper;
768
769 /* quantifiers - these members are used for storing state for
770 for the regops used to implement quantifiers */
771 struct {
772 /* this first element must match u.yes */
773 struct regmatch_state *prev_yes_state;
c476f425 774 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
775 regnode *me; /* the CURLYX node */
776 regnode *B; /* the B node in /A*B/ */
c476f425
DM
777 CHECKPOINT cp; /* remember current savestack index */
778 bool minmod;
a0374537 779 int parenfloor;/* how far back to strip paren data */
c476f425
DM
780
781 /* these two are modified by WHILEM */
782 int count; /* how many instances of A we've matched */
783 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
784 } curlyx;
785
786 struct {
c476f425 787 /* this first element must match u.yes */
262b90c4 788 struct regmatch_state *prev_yes_state;
c476f425
DM
789 struct regmatch_state *save_curlyx;
790 CHECKPOINT cp; /* remember current savestack indexes */
791 CHECKPOINT lastcp;
792 char *save_lastloc; /* previous curlyx.lastloc */
793 I32 cache_offset;
794 I32 cache_mask;
e822a8b4
DM
795 } whilem;
796
797 struct {
5d458dd8
YO
798 /* this first element must match u.yes */
799 struct regmatch_state *prev_yes_state;
79a2a0e8 800 int c1, c2; /* case fold search */
40a82448 801 CHECKPOINT cp;
f6033a9d
DM
802 U32 lastparen;
803 U32 lastcloseparen;
40a82448
DM
804 I32 alen; /* length of first-matched A string */
805 I32 count;
0cadcf80 806 bool minmod;
40a82448
DM
807 regnode *A, *B; /* the nodes corresponding to /A*B/ */
808 regnode *me; /* the curlym node */
79a2a0e8
KW
809 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
810 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
811 } curlym;
812
813 struct {
3b6647e0 814 U32 paren;
c255a977 815 CHECKPOINT cp;
f6033a9d
DM
816 U32 lastparen;
817 U32 lastcloseparen;
79a2a0e8 818 int c1, c2; /* case fold search */
c255a977
DM
819 char *maxpos; /* highest possible point in string to match */
820 char *oldloc; /* the previous locinput */
e822a8b4 821 int count;
c255a977
DM
822 int min, max; /* {m,n} */
823 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
824 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
825 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 826 } curly; /* and CURLYN/PLUS/STAR */
dad79028 827
d8319b27 828 } u;
5d9a96ca
DM
829} regmatch_state;
830
831/* how many regmatch_state structs to allocate as a single slab.
832 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
833 * pointers, plus 1 for any mythical malloc overhead. */
834
835#define PERL_REGMATCH_SLAB_SLOTS \
836 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
837
838typedef struct regmatch_slab {
839 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
840 struct regmatch_slab *prev, *next;
841} regmatch_slab;
1ade1aa1 842
46ab3289 843
cde0cee5 844
1ade1aa1
NC
845/*
846 * Local variables:
847 * c-indentation-style: bsd
848 * c-basic-offset: 4
14d04a33 849 * indent-tabs-mode: nil
1ade1aa1
NC
850 * End:
851 *
14d04a33 852 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 853 */