This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
simplify regcomp.c by using vars to avoid repeated macros
[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; \
b82ad30c 133 CV *qr_anoncv /* the anon sub wrapped round qr/(?{..})/ */
08e44740 134
bbe252da 135typedef struct regexp {
288b8c02 136 _XPV_HEAD;
08e44740 137 _REGEXP_COMMON;
f8fc2ecf
YO
138} regexp;
139
89f6f287 140#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 141
785a26d5 142/* used for high speed searches */
f9f4320a
YO
143typedef struct re_scream_pos_data_s
144{
145 char **scream_olds; /* match pos */
146 I32 *scream_pos; /* Internal iterator of scream. */
147} re_scream_pos_data;
148
785a26d5
YO
149/* regexp_engine structure. This is the dispatch table for regexes.
150 * Any regex engine implementation must be able to build one of these.
151 */
f9f4320a 152typedef struct regexp_engine {
1593ad57 153 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
154 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
155 char* strbeg, I32 minend, SV* screamer,
156 void* data, U32 flags);
157 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
158 char *strend, const U32 flags,
159 re_scream_pos_data *data);
49d7dfbc
AB
160 SV* (*checkstr) (pTHX_ REGEXP * const rx);
161 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 162 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 163 SV * const sv);
2fdbfb4d
AB
164 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
165 SV const * const value);
166 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
167 const I32 paren);
192b9cd1
AB
168 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
169 SV * const value, const U32 flags);
170 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
171 const U32 flags);
49d7dfbc 172 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 173#ifdef USE_ITHREADS
49d7dfbc 174 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 175#endif
3c13cae6
DM
176 REGEXP* (*op_comp) (pTHX_ SV ** const patternp, int pat_count,
177 OP *expr, const struct regexp_engine* eng,
178 REGEXP *VOL old_re,
76ac488f 179 bool *is_bare_re, U32 orig_rx_flags, U32 pm_flags);
f9f4320a
YO
180} regexp_engine;
181
192b9cd1
AB
182/*
183 These are passed to the numbered capture variable callbacks as the
184 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
185 $2 etc.
186*/
2c7b5d76
DM
187#define RX_BUFF_IDX_CARET_PREMATCH -5 /* ${^PREMATCH} */
188#define RX_BUFF_IDX_CARET_POSTMATCH -4 /* ${^POSTMATCH} */
189#define RX_BUFF_IDX_CARET_FULLMATCH -3 /* ${^MATCH} */
190#define RX_BUFF_IDX_PREMATCH -2 /* $` */
191#define RX_BUFF_IDX_POSTMATCH -1 /* $' */
192#define RX_BUFF_IDX_FULLMATCH 0 /* $& */
192b9cd1
AB
193
194/*
195 Flags that are passed to the named_buff and named_buff_iter
196 callbacks above. Those routines are called from universal.c via the
197 Tie::Hash::NamedCapture interface for %+ and %- and the re::
198 functions in the same file.
199*/
200
201/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
202#define RXapif_FETCH 0x0001
203#define RXapif_STORE 0x0002
204#define RXapif_DELETE 0x0004
205#define RXapif_CLEAR 0x0008
206#define RXapif_EXISTS 0x0010
207#define RXapif_SCALAR 0x0020
208#define RXapif_FIRSTKEY 0x0040
209#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
210
211/* Whether %+ or %- is being operated on */
f1b875a0
YO
212#define RXapif_ONE 0x0100 /* %+ */
213#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
214
215/* Whether this is being called from a re:: function */
f1b875a0
YO
216#define RXapif_REGNAME 0x0400
217#define RXapif_REGNAMES 0x0800
7948fc08 218#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 219
f7e71195
AB
220/*
221=head1 REGEXP Functions
222
223=for apidoc Am|REGEXP *|SvRX|SV *sv
224
225Convenience macro to get the REGEXP from a SV. This is approximately
226equivalent to the following snippet:
227
228 if (SvMAGICAL(sv))
229 mg_get(sv);
1af910ea
DL
230 if (SvROK(sv))
231 sv = MUTABLE_SV(SvRV(sv));
232 if (SvTYPE(sv) == SVt_REGEXP)
233 return (REGEXP*) sv;
f7e71195
AB
234
235NULL will be returned if a REGEXP* is not found.
236
237=for apidoc Am|bool|SvRXOK|SV* sv
238
1af910ea
DL
239Returns a boolean indicating whether the SV (or the one it references)
240is a REGEXP.
f7e71195
AB
241
242If you want to do something with the REGEXP* later use SvRX instead
243and check for NULL.
244
245=cut
246*/
247
248#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
249#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
250
251
7948fc08 252/* Flags stored in regexp->extflags
bbe252da 253 * These are used by code external to the regexp engine
e357fc67 254 *
5b126c84
KW
255 * Note that the flags whose names start with RXf_PMf_ are defined in
256 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2 257 *
eb2624c9
FC
258 * NOTE: if you modify any RXf flags you should run regen.pl or
259 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2 260 *
bbe252da
YO
261 */
262
5b126c84
KW
263#include "op_reg_common.h"
264
cde0cee5 265#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 266
a20207d7
YO
267#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
268 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
269 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
270 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
271 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 272
94b03d7d 273/* Note, includes charset ones, assumes 0 is the default for them */
fb85c044 274#define STD_PMMOD_FLAGS_CLEAR(pmfl) \
a62b1201 275 *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_CHARSET)
fb85c044 276
bcdf7404 277/* chars and strings used as regex pattern modifiers
486ec47a 278 * Singular is a 'c'har, plural is a "string"
87e95b7f
YO
279 *
280 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
281 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
282 * for its own uses. So 'k' is out as well.
bcdf7404 283 */
85508812 284#define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */
bcdf7404 285#define EXEC_PAT_MOD 'e'
87e95b7f 286#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
287#define ONCE_PAT_MOD 'o'
288#define GLOBAL_PAT_MOD 'g'
289#define CONTINUE_PAT_MOD 'c'
290#define MULTILINE_PAT_MOD 'm'
291#define SINGLE_PAT_MOD 's'
292#define IGNORE_PAT_MOD 'i'
293#define XTENDED_PAT_MOD 'x'
4f4d7508 294#define NONDESTRUCT_PAT_MOD 'r'
9de15fec
KW
295#define LOCALE_PAT_MOD 'l'
296#define UNICODE_PAT_MOD 'u'
50e91148 297#define DEPENDS_PAT_MOD 'd'
cfaf538b 298#define ASCII_RESTRICT_PAT_MOD 'a'
bcdf7404
YO
299
300#define ONCE_PAT_MODS "o"
87e95b7f 301#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
302#define EXEC_PAT_MODS "e"
303#define LOOP_PAT_MODS "gc"
4f4d7508 304#define NONDESTRUCT_PAT_MODS "r"
9de15fec
KW
305#define LOCALE_PAT_MODS "l"
306#define UNICODE_PAT_MODS "u"
50e91148 307#define DEPENDS_PAT_MODS "d"
cfaf538b 308#define ASCII_RESTRICT_PAT_MODS "a"
df7a8460 309#define ASCII_MORE_RESTRICT_PAT_MODS "aa"
bcdf7404 310
78fecfc8 311/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
312 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
313 * character is bit +1, etc. */
bcdf7404
YO
314#define STD_PAT_MODS "msix"
315
94b03d7d
KW
316#define CHARSET_PAT_MODS ASCII_RESTRICT_PAT_MODS DEPENDS_PAT_MODS LOCALE_PAT_MODS UNICODE_PAT_MODS
317
78fecfc8 318/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
319 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
320 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
321#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
322
323#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
94b03d7d 324#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS CHARSET_PAT_MODS
bcdf7404 325#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 326#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 327
cb5027f2 328/*
eb2624c9
FC
329 * NOTE: if you modify any RXf flags you should run regen.pl or
330 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
331 *
332 */
bcdf7404 333
e795e964
KW
334/* Leave some space, so future bit allocations can go either in the shared or
335 * unshared area without affecting binary compatibility */
df7a8460 336#define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+1)
f4a4cecb 337
bf28aba1
KW
338/* Manually decorate this function with gcc-style attributes just to
339 * avoid having to restructure the header files and their called order,
340 * as proto.h would have to be included before this file, and isn't */
341
a62b1201
KW
342PERL_STATIC_INLINE const char *
343get_regex_charset_name(const U32 flags, STRLEN* const lenp)
344 __attribute__warn_unused_result__;
345
df7a8460 346#define MAX_CHARSET_NAME_LENGTH 2
a62b1201
KW
347
348PERL_STATIC_INLINE const char *
349get_regex_charset_name(const U32 flags, STRLEN* const lenp)
350{
351 /* Returns a string that corresponds to the name of the regex character set
352 * given by 'flags', and *lenp is set the length of that string, which
353 * cannot exceed MAX_CHARSET_NAME_LENGTH characters */
354
355 *lenp = 1;
356 switch (get_regex_charset(flags)) {
50e91148 357 case REGEX_DEPENDS_CHARSET: return DEPENDS_PAT_MODS;
a62b1201
KW
358 case REGEX_LOCALE_CHARSET: return LOCALE_PAT_MODS;
359 case REGEX_UNICODE_CHARSET: return UNICODE_PAT_MODS;
cfaf538b 360 case REGEX_ASCII_RESTRICTED_CHARSET: return ASCII_RESTRICT_PAT_MODS;
df7a8460 361 case REGEX_ASCII_MORE_RESTRICTED_CHARSET:
07075d19
KW
362 *lenp = 2;
363 return ASCII_MORE_RESTRICT_PAT_MODS;
0984e55e
JC
364 default:
365 return "?"; /* Unknown */
a62b1201 366 }
a62b1201
KW
367}
368
cccd1425
FC
369/*
370 Two flags no longer used.
371 RXf_SPLIT used to be set in Perl_pmruntime if op_flags & OPf_SPECIAL,
372 i.e., split. It was used by the regex engine to check whether it should
373 set RXf_SKIPWHITE. Regexp plugins on CPAN also have done the same thing
374 historically, so we leave these flags defined.
375*/
376#ifndef PERL_CORE
377# define RXf_SPLIT 0
378# define RXf_SKIPWHITE 0
379#endif
380
381
52d81aa8 382/* Anchor and GPOS related stuff */
1ad4ec54
KW
383#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0))
384#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1))
385#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2))
386#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3))
387#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4))
388#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5))
52d81aa8
NC
389/* two bits here */
390#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
391#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
392#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
393
bbe252da 394/* What we have seen */
1ad4ec54
KW
395#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6))
396#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7))
397#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8))
bbe252da
YO
398
399/* Special */
1ad4ec54
KW
400#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9))
401#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10))
bbe252da
YO
402
403/* UTF8 related */
e795e964 404#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
405
406/* Intuit related */
e795e964
KW
407#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12))
408#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13))
409#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14))
bbe252da
YO
410#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
411
2ed8c61f
FC
412#define RXf_MODIFIES_VARS (1<<(RXf_BASE_SHIFT+15))
413
bbe252da 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 /^/ */
e795e964 425#define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */
ca835a20 426#define RXf_NULL (1U<<(RXf_BASE_SHIFT+22)) /* Pattern is // */
e795e964 427#if RXf_BASE_SHIFT+22 > 31
1850c8f9
KW
428# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
429#endif
c737faaf 430
cb5027f2 431/*
eb2624c9
FC
432 * NOTE: if you modify any RXf flags you should run regen.pl or
433 * regen/regcomp.pl so that regnodes.h is updated with the changes.
cb5027f2
YO
434 *
435 */
bbe252da 436
284167a5
SM
437#if NO_TAINT_SUPPORT
438# define RX_ISTAINTED(prog) 0
439# define RX_TAINT_on(prog) NOOP
272d35c9
DM
440# define RXp_MATCH_TAINTED(prog) 0
441# define RX_MATCH_TAINTED(prog) 0
442# define RXp_MATCH_TAINTED_on(prog) NOOP
443# define RX_MATCH_TAINTED_on(prog) NOOP
444# define RX_MATCH_TAINTED_off(prog) NOOP
284167a5
SM
445#else
446# define RX_ISTAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED)
447# define RX_TAINT_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED)
272d35c9
DM
448# define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
449# define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
450# define RXp_MATCH_TAINTED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
451# define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
452# define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
284167a5
SM
453#endif
454
bbe252da 455#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
72311751
GS
456#define RX_MATCH_TAINTED_set(prog, t) ((t) \
457 ? RX_MATCH_TAINTED_on(prog) \
458 : RX_MATCH_TAINTED_off(prog))
c277df42 459
07bc277f
NC
460#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
461#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
462#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
463#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
464#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
465#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
466#define RX_MATCH_COPIED_set(prog,t) ((t) \
467 ? RX_MATCH_COPIED_on(prog) \
468 : RX_MATCH_COPIED_off(prog))
f9f4320a 469
07bc277f 470#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 471
07bc277f 472/* For source compatibility. We used to store these explicitly. */
8d919b0a
FC
473#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ReANY(prog)->pre_prefix)
474#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ReANY(prog)->pre_prefix)
5509d87a
NC
475/* FIXME? Are we hardcoding too much here and constraining plugin extension
476 writers? Specifically, the value 1 assumes that the wrapped version always
477 has exactly one character at the end, a ')'. Will that always be true? */
8d919b0a
FC
478#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ReANY(prog)->pre_prefix - 1)
479#define RX_WRAPPED(prog) ReANY(prog)->xpv_len_u.xpvlenu_pv
480#define RX_WRAPPED_const(prog) ((const char *)RX_WRAPPED(prog))
9d17798d 481#define RX_WRAPLEN(prog) SvCUR(prog)
8d919b0a 482#define RX_CHECK_SUBSTR(prog) (ReANY(prog)->check_substr)
288b8c02 483#define RX_REFCNT(prog) SvREFCNT(prog)
8d919b0a
FC
484#define RX_EXTFLAGS(prog) RXp_EXTFLAGS(ReANY(prog))
485#define RX_ENGINE(prog) (ReANY(prog)->engine)
486#define RX_SUBBEG(prog) (ReANY(prog)->subbeg)
487#define RX_SUBOFFSET(prog) (ReANY(prog)->suboffset)
488#define RX_SUBCOFFSET(prog) (ReANY(prog)->subcoffset)
489#define RX_OFFS(prog) (ReANY(prog)->offs)
490#define RX_NPARENS(prog) (ReANY(prog)->nparens)
491#define RX_SUBLEN(prog) (ReANY(prog)->sublen)
492#define RX_MINLEN(prog) (ReANY(prog)->minlen)
493#define RX_MINLENRET(prog) (ReANY(prog)->minlenret)
494#define RX_GOFS(prog) (ReANY(prog)->gofs)
495#define RX_LASTPAREN(prog) (ReANY(prog)->lastparen)
496#define RX_LASTCLOSEPAREN(prog) (ReANY(prog)->lastcloseparen)
497#define RX_SAVED_COPY(prog) (ReANY(prog)->saved_copy)
220fc49f 498
be8e71aa
YO
499#endif /* PLUGGABLE_RE_EXTENSION */
500
486ec47a 501/* Stuff that needs to be included in the pluggable extension goes below here */
be8e71aa 502
db2c6cb3 503#ifdef PERL_ANY_COW
ed252734 504#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
505 STMT_START {if (RX_SAVED_COPY(rx)) { \
506 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
507 } \
508 if (RX_MATCH_COPIED(rx)) { \
07bc277f 509 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
510 RX_MATCH_COPIED_off(rx); \
511 }} STMT_END
512#else
513#define RX_MATCH_COPY_FREE(rx) \
514 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 515 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
516 RX_MATCH_COPIED_off(rx); \
517 }} STMT_END
518#endif
519
07bc277f
NC
520#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
521#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
522#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
523#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
524#define RX_MATCH_UTF8_set(prog, t) ((t) \
525 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
526 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
527
528/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 529#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 530
cad2e5aa
JH
531#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
532#define REXEC_CHECKED 0x02 /* check_substr already checked. */
533#define REXEC_SCREAM 0x04 /* use scream table. */
534#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 535#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
6502e081
DM
536 /* under REXEC_COPY_STR, it's ok for the
537 * engine (modulo PL_sawamperand etc)
538 * to skip copying ... */
539#define REXEC_COPY_SKIP_PRE 0x20 /* ...the $` part of the string, or */
540#define REXEC_COPY_SKIP_POST 0x40 /* ...the $' part of the string */
c277df42 541
7fe44985 542#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
543# define ReREFCNT_inc(re) \
544 ({ \
545 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
546 REGEXP *const _rerefcnt_inc = (re); \
547 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
548 SvREFCNT_inc(_rerefcnt_inc); \
549 _rerefcnt_inc; \
288b8c02
NC
550 })
551# define ReREFCNT_dec(re) \
552 ({ \
553 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
554 REGEXP *const _rerefcnt_dec = (re); \
555 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
556 })
557#else
558# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 559# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02 560#endif
8d919b0a 561#define ReANY(re) S_ReANY((const REGEXP *)(re))
288b8c02
NC
562
563/* FIXME for plugins. */
cf93c79d
IZ
564
565#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
566#define FBMcf_TAIL_DOLLARM 2
567#define FBMcf_TAIL_Z 4
568#define FBMcf_TAIL_z 8
569#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
570
571#define FBMrf_MULTILINE 1
cad2e5aa 572
3b0527fe
DM
573/* some basic information about the current match that is created by
574 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
575
576typedef struct {
288b8c02 577 REGEXP *prog;
3b0527fe
DM
578 char *bol;
579 char *till;
580 SV *sv;
581 char *ganch;
24b23f37 582 char *cutpoint;
984e6dd1 583 bool is_utf8_pat;
39819bd9 584 bool warned; /* we have issued a recursion warning; no need for more */
3b0527fe
DM
585} regmatch_info;
586
5d9a96ca
DM
587
588/* structures for holding and saving the state maintained by regmatch() */
589
d56b3014 590#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 591#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 592#endif
6bda09f9 593
5d9a96ca
DM
594typedef I32 CHECKPOINT;
595
a0374537 596typedef struct regmatch_state {
40a82448 597 int resume_state; /* where to jump to on return */
24d3c4a9 598 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
599
600 union {
77cb431f
DM
601
602 /* this is a fake union member that matches the first element
603 * of each member that needs to store positive backtrack
604 * information */
605 struct {
606 struct regmatch_state *prev_yes_state;
607 } yes;
608
fae667d5
YO
609 /* branchlike members */
610 /* this is a fake union member that matches the first elements
611 * of each member that needs to behave like a branch */
612 struct {
5d458dd8
YO
613 /* this first element must match u.yes */
614 struct regmatch_state *prev_yes_state;
fae667d5 615 U32 lastparen;
f6033a9d 616 U32 lastcloseparen;
fae667d5
YO
617 CHECKPOINT cp;
618
619 } branchlike;
620
621 struct {
622 /* the first elements must match u.branchlike */
623 struct regmatch_state *prev_yes_state;
624 U32 lastparen;
f6033a9d 625 U32 lastcloseparen;
fae667d5
YO
626 CHECKPOINT cp;
627
628 regnode *next_branch; /* next branch node */
629 } branch;
630
631 struct {
632 /* the first elements must match u.branchlike */
633 struct regmatch_state *prev_yes_state;
634 U32 lastparen;
f6033a9d 635 U32 lastcloseparen;
fae667d5
YO
636 CHECKPOINT cp;
637
2e64971a 638 U32 accepted; /* how many accepting states left */
6603fe3e 639 bool longfold;/* saw a fold with a 1->n char mapping */
7f69552c 640 U16 *jump; /* positive offsets from me */
7f69552c 641 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
642 U8 *firstpos;/* pos in string of first trie match */
643 U32 firstchars;/* len in chars of firstpos from start */
644 U16 nextword;/* next word to try */
645 U16 topword; /* longest accepted word */
e822a8b4
DM
646 } trie;
647
fae667d5
YO
648 /* special types - these members are used to store state for special
649 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 650 struct {
77cb431f
DM
651 /* this first element must match u.yes */
652 struct regmatch_state *prev_yes_state;
faec1544
DM
653 struct regmatch_state *prev_eval;
654 struct regmatch_state *prev_curlyx;
288b8c02 655 REGEXP *prev_rex;
984e6dd1 656 bool saved_utf8_pat; /* saved copy of is_utf8_pat */
aa283a38
DM
657 CHECKPOINT cp; /* remember current savestack indexes */
658 CHECKPOINT lastcp;
6bda09f9 659 U32 close_paren; /* which close bracket is our end */
1b5c067c 660 regnode *B; /* the node following us */
e822a8b4
DM
661 } eval;
662
663 struct {
c476f425 664 /* this first element must match u.yes */
262b90c4 665 struct regmatch_state *prev_yes_state;
fae667d5
YO
666 I32 wanted;
667 I32 logical; /* saved copy of 'logical' var */
668 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
669 } ifmatch; /* and SUSPEND/UNLESSM */
670
671 struct {
672 /* this first element must match u.yes */
673 struct regmatch_state *prev_yes_state;
674 struct regmatch_state *prev_mark;
675 SV* mark_name;
676 char *mark_loc;
677 } mark;
678
679 struct {
680 int val;
681 } keeper;
682
683 /* quantifiers - these members are used for storing state for
684 for the regops used to implement quantifiers */
685 struct {
686 /* this first element must match u.yes */
687 struct regmatch_state *prev_yes_state;
c476f425 688 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
689 regnode *me; /* the CURLYX node */
690 regnode *B; /* the B node in /A*B/ */
c476f425
DM
691 CHECKPOINT cp; /* remember current savestack index */
692 bool minmod;
a0374537 693 int parenfloor;/* how far back to strip paren data */
c476f425
DM
694
695 /* these two are modified by WHILEM */
696 int count; /* how many instances of A we've matched */
697 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
698 } curlyx;
699
700 struct {
c476f425 701 /* this first element must match u.yes */
262b90c4 702 struct regmatch_state *prev_yes_state;
c476f425
DM
703 struct regmatch_state *save_curlyx;
704 CHECKPOINT cp; /* remember current savestack indexes */
705 CHECKPOINT lastcp;
706 char *save_lastloc; /* previous curlyx.lastloc */
707 I32 cache_offset;
708 I32 cache_mask;
e822a8b4
DM
709 } whilem;
710
711 struct {
5d458dd8
YO
712 /* this first element must match u.yes */
713 struct regmatch_state *prev_yes_state;
79a2a0e8 714 int c1, c2; /* case fold search */
40a82448 715 CHECKPOINT cp;
f6033a9d
DM
716 U32 lastparen;
717 U32 lastcloseparen;
40a82448
DM
718 I32 alen; /* length of first-matched A string */
719 I32 count;
0cadcf80 720 bool minmod;
40a82448
DM
721 regnode *A, *B; /* the nodes corresponding to /A*B/ */
722 regnode *me; /* the curlym node */
79a2a0e8
KW
723 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
724 U8 c2_utf8[UTF8_MAXBYTES+1];
e822a8b4
DM
725 } curlym;
726
727 struct {
3b6647e0 728 U32 paren;
c255a977 729 CHECKPOINT cp;
f6033a9d
DM
730 U32 lastparen;
731 U32 lastcloseparen;
79a2a0e8 732 int c1, c2; /* case fold search */
c255a977
DM
733 char *maxpos; /* highest possible point in string to match */
734 char *oldloc; /* the previous locinput */
e822a8b4 735 int count;
c255a977
DM
736 int min, max; /* {m,n} */
737 regnode *A, *B; /* the nodes corresponding to /A*B/ */
79a2a0e8
KW
738 U8 c1_utf8[UTF8_MAXBYTES+1]; /* */
739 U8 c2_utf8[UTF8_MAXBYTES+1];
c255a977 740 } curly; /* and CURLYN/PLUS/STAR */
dad79028 741
d8319b27 742 } u;
5d9a96ca
DM
743} regmatch_state;
744
745/* how many regmatch_state structs to allocate as a single slab.
746 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
747 * pointers, plus 1 for any mythical malloc overhead. */
748
749#define PERL_REGMATCH_SLAB_SLOTS \
750 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
751
752typedef struct regmatch_slab {
753 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
754 struct regmatch_slab *prev, *next;
755} regmatch_slab;
1ade1aa1 756
46ab3289 757#define PL_bostr PL_reg_state.re_state_bostr
46ab3289 758#define PL_regeol PL_reg_state.re_state_regeol
46ab3289
NC
759#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
760#define PL_reg_magic PL_reg_state.re_state_reg_magic
761#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
762#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
763#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
764#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
765#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
6502e081
DM
766#define PL_reg_oldsavedoffset PL_reg_state.re_state_reg_oldsavedoffset
767#define PL_reg_oldsavedcoffset PL_reg_state.re_state_reg_oldsavedcoffset
46ab3289
NC
768#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
769#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
770#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
771#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
46ab3289
NC
772#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
773#define PL_nrs PL_reg_state.re_state_nrs
774
1ade1aa1 775struct re_save_state {
ed301438 776 bool re_state_eval_setup_done; /* from regexec.c */
1b5c067c 777 bool re_state_reg_match_utf8; /* from regexec.c */
d24ca0c5 778 bool re_reparsing; /* runtime (?{}) fed back into parser */
a31421fb
NC
779 /* Space for U8 */
780 I32 re_state_reg_oldpos; /* from regexec.c */
781 I32 re_state_reg_maxiter; /* max wait until caching pos */
782 I32 re_state_reg_leftiter; /* wait until caching pos */
1ade1aa1 783 char *re_state_bostr;
1ade1aa1 784 char *re_state_regeol; /* End of input, for $ check. */
1ade1aa1 785 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
786 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
787 PMOP *re_state_reg_curpm; /* from regexec.c */
788 char *re_state_reg_oldsaved; /* old saved substr during match */
789 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
6502e081
DM
790 STRLEN re_state_reg_oldsavedoffset; /* old offset of saved substr during match */
791 STRLEN re_state_reg_oldsavedcoffset;/* old coffset of saved substr during match */
1b5c067c 792 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
1b5c067c 793 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1 794 char *re_state_reg_starttry; /* from regexec.c */
db2c6cb3 795#ifdef PERL_ANY_COW
46ab3289 796 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
797#endif
798};
799
800#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
801 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 802
1ade1aa1
NC
803/*
804 * Local variables:
805 * c-indentation-style: bsd
806 * c-basic-offset: 4
14d04a33 807 * indent-tabs-mode: nil
1ade1aa1
NC
808 * End:
809 *
14d04a33 810 * ex: set ts=8 sts=4 sw=4 et:
1ade1aa1 811 */