This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
op_reg_common.h: Continue refactoring
[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
c277df42
IZ
21struct regnode {
22 U8 flags;
23 U8 type;
24 U16 next_off;
25};
26
27typedef struct regnode regnode;
28
cad2e5aa 29struct reg_substr_data;
2779dcf1 30
0ee3c8d0
JH
31struct reg_data;
32
f9f4320a 33struct regexp_engine;
28d8d7f4 34struct regexp;
bbe252da 35
785a26d5
YO
36struct reg_substr_datum {
37 I32 min_offset;
38 I32 max_offset;
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
41 I32 end_shift;
42};
43struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
45};
f9f4320a 46
f8c7b90f 47#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
48#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
49#else
50#define SV_SAVED_COPY
ed252734 51#endif
28d8d7f4 52
f0ab9afb
NC
53typedef struct regexp_paren_pair {
54 I32 start;
55 I32 end;
56} regexp_paren_pair;
28d8d7f4 57
882227b7
AB
58/*
59 The regexp/REGEXP struct, see L<perlreapi> for further documentation
60 on the individual fields. The struct is ordered so that the most
61 commonly used fields are placed at the start.
62
63 Any patch that adds items to this struct will need to include
64 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
65 (C<pregfree()>). This involves freeing or cloning items in the
66 regexp's data array based on the data item's type.
67*/
68
08e44740
NC
69#define _REGEXP_COMMON \
70 /* what engine created this regexp? */ \
71 const struct regexp_engine* engine; \
72 REGEXP *mother_re; /* what re is this a lightweight copy of? */ \
89f6f287 73 HV *paren_names; /* Optional hash of paren names */ \
08e44740
NC
74 /* Information about the match that the perl core uses to */ \
75 /* manage things */ \
76 U32 extflags; /* Flags used both externally and internally */ \
77 I32 minlen; /* mininum possible length of string to match */\
78 I32 minlenret; /* mininum possible length of $& */ \
79 U32 gofs; /* chars left of pos that we search from */ \
80 /* substring data about strings that must appear in the */ \
81 /* final match, used for optimisations */ \
82 struct reg_substr_data *substrs; \
83 U32 nparens; /* number of capture buffers */ \
84 /* private engine specific data */ \
85 U32 intflags; /* Engine Specific Internal flags */ \
86 void *pprivate; /* Data private to the regex engine which */ \
87 /* created this object. */ \
88 /* Data about the last/current match. These are modified */ \
89 /* during matching */ \
90 U32 lastparen; /* last open paren matched */ \
91 U32 lastcloseparen; /* last close paren matched */ \
e9105d30 92 regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \
08e44740
NC
93 /* Array of offsets for (@-) and (@+) */ \
94 regexp_paren_pair *offs; \
95 /* saved or original string so \digit works forever. */ \
96 char *subbeg; \
97 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\
98 I32 sublen; /* Length of string pointed by subbeg */ \
99 /* Information about the match that isn't often used */ \
100 /* offset from wrapped to the start of precomp */ \
654eccd5 101 PERL_BITFIELD32 pre_prefix:4; \
08e44740 102 /* number of eval groups in the pattern - for security checks */\
654eccd5 103 PERL_BITFIELD32 seen_evals:28
08e44740 104
bbe252da 105typedef struct regexp {
288b8c02 106 _XPV_HEAD;
08e44740 107 _REGEXP_COMMON;
f8fc2ecf
YO
108} regexp;
109
89f6f287 110#define RXp_PAREN_NAMES(rx) ((rx)->paren_names)
5daac39c 111
785a26d5 112/* used for high speed searches */
f9f4320a
YO
113typedef struct re_scream_pos_data_s
114{
115 char **scream_olds; /* match pos */
116 I32 *scream_pos; /* Internal iterator of scream. */
117} re_scream_pos_data;
118
785a26d5
YO
119/* regexp_engine structure. This is the dispatch table for regexes.
120 * Any regex engine implementation must be able to build one of these.
121 */
f9f4320a 122typedef struct regexp_engine {
1593ad57 123 REGEXP* (*comp) (pTHX_ SV * const pattern, U32 flags);
49d7dfbc
AB
124 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
125 char* strbeg, I32 minend, SV* screamer,
126 void* data, U32 flags);
127 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO
128 char *strend, const U32 flags,
129 re_scream_pos_data *data);
49d7dfbc
AB
130 SV* (*checkstr) (pTHX_ REGEXP * const rx);
131 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 132 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 133 SV * const sv);
2fdbfb4d
AB
134 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
135 SV const * const value);
136 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
137 const I32 paren);
192b9cd1
AB
138 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
139 SV * const value, const U32 flags);
140 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
141 const U32 flags);
49d7dfbc 142 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 143#ifdef USE_ITHREADS
49d7dfbc 144 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 145#endif
f9f4320a
YO
146} regexp_engine;
147
192b9cd1
AB
148/*
149 These are passed to the numbered capture variable callbacks as the
150 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
151 $2 etc.
152*/
f1b875a0
YO
153#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
154#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
155#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB
156
157/*
158 Flags that are passed to the named_buff and named_buff_iter
159 callbacks above. Those routines are called from universal.c via the
160 Tie::Hash::NamedCapture interface for %+ and %- and the re::
161 functions in the same file.
162*/
163
164/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO
165#define RXapif_FETCH 0x0001
166#define RXapif_STORE 0x0002
167#define RXapif_DELETE 0x0004
168#define RXapif_CLEAR 0x0008
169#define RXapif_EXISTS 0x0010
170#define RXapif_SCALAR 0x0020
171#define RXapif_FIRSTKEY 0x0040
172#define RXapif_NEXTKEY 0x0080
192b9cd1
AB
173
174/* Whether %+ or %- is being operated on */
f1b875a0
YO
175#define RXapif_ONE 0x0100 /* %+ */
176#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB
177
178/* Whether this is being called from a re:: function */
f1b875a0
YO
179#define RXapif_REGNAME 0x0400
180#define RXapif_REGNAMES 0x0800
7948fc08 181#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 182
f7e71195
AB
183/*
184=head1 REGEXP Functions
185
186=for apidoc Am|REGEXP *|SvRX|SV *sv
187
188Convenience macro to get the REGEXP from a SV. This is approximately
189equivalent to the following snippet:
190
191 if (SvMAGICAL(sv))
192 mg_get(sv);
193 if (SvROK(sv) &&
194 (tmpsv = (SV*)SvRV(sv)) &&
195 SvTYPE(tmpsv) == SVt_PVMG &&
196 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
197 {
198 return (REGEXP *)tmpmg->mg_obj;
199 }
200
201NULL will be returned if a REGEXP* is not found.
202
203=for apidoc Am|bool|SvRXOK|SV* sv
204
205Returns a boolean indicating whether the SV contains qr magic
206(PERL_MAGIC_qr).
207
208If you want to do something with the REGEXP* later use SvRX instead
209and check for NULL.
210
211=cut
212*/
213
214#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
215#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
216
217
7948fc08 218/* Flags stored in regexp->extflags
bbe252da 219 * These are used by code external to the regexp engine
e357fc67 220 *
5b126c84
KW
221 * Note that the flags whose names start with RXf_PMf_ are defined in
222 * op_reg_common.h, being copied from the parallel flags of op_pmflags
cb5027f2
YO
223 *
224 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 225 * so that regnodes.h is updated with the changes.
cb5027f2 226 *
bbe252da
YO
227 */
228
5b126c84
KW
229#include "op_reg_common.h"
230
cde0cee5 231#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
cde0cee5 232
a20207d7
YO
233#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
234 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
235 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
236 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
237 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 238
bcdf7404
YO
239/* chars and strings used as regex pattern modifiers
240 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
241 *
242 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
243 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
244 * for its own uses. So 'k' is out as well.
bcdf7404
YO
245 */
246#define EXEC_PAT_MOD 'e'
87e95b7f 247#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
248#define ONCE_PAT_MOD 'o'
249#define GLOBAL_PAT_MOD 'g'
250#define CONTINUE_PAT_MOD 'c'
251#define MULTILINE_PAT_MOD 'm'
252#define SINGLE_PAT_MOD 's'
253#define IGNORE_PAT_MOD 'i'
254#define XTENDED_PAT_MOD 'x'
4f4d7508 255#define NONDESTRUCT_PAT_MOD 'r'
bcdf7404
YO
256
257#define ONCE_PAT_MODS "o"
87e95b7f 258#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
259#define EXEC_PAT_MODS "e"
260#define LOOP_PAT_MODS "gc"
4f4d7508 261#define NONDESTRUCT_PAT_MODS "r"
bcdf7404 262
78fecfc8 263/* This string is expected by regcomp.c to be ordered so that the first
c18d5d15
KW
264 * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next
265 * character is bit +1, etc. */
bcdf7404
YO
266#define STD_PAT_MODS "msix"
267
78fecfc8 268/* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered
c18d5d15
KW
269 * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of
270 * extflags; the next character is in bit +1, etc. */
bcdf7404
YO
271#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
272
273#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
274#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
275#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
4f4d7508 276#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS
bcdf7404 277
cb5027f2
YO
278/*
279 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 280 * so that regnodes.h is updated with the changes.
cb5027f2
YO
281 *
282 */
bcdf7404 283
3214c85f 284#define RXf_BASE_SHIFT ((_RXf_PMf_SHIFT_NEXT)-1)
f4a4cecb 285
52d81aa8 286/* Anchor and GPOS related stuff */
f4a4cecb
KW
287#define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+3))
288#define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+4))
289#define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+5))
290#define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+6))
291#define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+7))
292#define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+8))
52d81aa8
NC
293/* two bits here */
294#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
295#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
296#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
297
bbe252da 298/* What we have seen */
f4a4cecb
KW
299#define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+9))
300#define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+10))
301#define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+11))
bbe252da
YO
302
303/* Special */
f4a4cecb
KW
304#define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+12))
305#define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+13))
bbe252da
YO
306
307/* UTF8 related */
f4a4cecb 308#define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+15))
bbe252da
YO
309
310/* Intuit related */
f4a4cecb
KW
311#define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+16))
312#define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+17))
313#define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+18))
0ac6acae
AB
314
315/*
316 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
317 be used by regex engines to check whether they should set
318 RXf_SKIPWHITE
319*/
f4a4cecb 320#define RXf_SPLIT (1<<(RXf_BASE_SHIFT+19))
0ac6acae 321
bbe252da
YO
322#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
323
324/* Copy and tainted info */
f4a4cecb
KW
325#define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+20))
326#define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+21))
327#define RXf_TAINTED (1<<(RXf_BASE_SHIFT+22)) /* this pattern is tainted */
52d81aa8
NC
328
329/* Flags indicating special patterns */
f4a4cecb
KW
330#define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+23)) /* Pattern is /^/ */
331#define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+24)) /* Pattern is for a split / / */
332#define RXf_WHITE (1<<(RXf_BASE_SHIFT+25)) /* Pattern is /\s+/ */
333#define RXf_NULL (1<<(RXf_BASE_SHIFT+26)) /* Pattern is // */
334#if RXf_BASE_SHIFT+26 > 31
1850c8f9
KW
335# error Too many RXf_PMf bits used. See regnodes.h for any spare in middle
336#endif
c737faaf 337
cb5027f2
YO
338/*
339 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
7948fc08 340 * so that regnodes.h is updated with the changes.
cb5027f2
YO
341 *
342 */
bbe252da 343
bbe252da 344#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC
345#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
346#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
347#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
348#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS
349#define RX_MATCH_TAINTED_set(prog, t) ((t) \
350 ? RX_MATCH_TAINTED_on(prog) \
351 : RX_MATCH_TAINTED_off(prog))
c277df42 352
07bc277f
NC
353#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
354#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
355#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
356#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
357#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
358#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ
359#define RX_MATCH_COPIED_set(prog,t) ((t) \
360 ? RX_MATCH_COPIED_on(prog) \
361 : RX_MATCH_COPIED_off(prog))
f9f4320a 362
07bc277f 363#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 364
07bc277f 365/* For source compatibility. We used to store these explicitly. */
f7c278bf 366#define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
d2c6dc5e 367#define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix)
5509d87a
NC
368/* FIXME? Are we hardcoding too much here and constraining plugin extension
369 writers? Specifically, the value 1 assumes that the wrapped version always
370 has exactly one character at the end, a ')'. Will that always be true? */
371#define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1)
f7c278bf 372#define RX_WRAPPED(prog) SvPVX(prog)
d2c6dc5e 373#define RX_WRAPPED_const(prog) SvPVX_const(prog)
9d17798d 374#define RX_WRAPLEN(prog) SvCUR(prog)
288b8c02 375#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
288b8c02 376#define RX_REFCNT(prog) SvREFCNT(prog)
7fe44985 377#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
4450b630
NC
378# define RX_EXTFLAGS(prog) \
379 (*({ \
7948fc08
RU
380 const REGEXP *const _rx_extflags = (prog); \
381 assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \
382 &RXp_EXTFLAGS(SvANY(_rx_extflags)); \
4450b630
NC
383 }))
384# define RX_ENGINE(prog) \
385 (*({ \
7948fc08
RU
386 const REGEXP *const _rx_engine = (prog); \
387 assert(SvTYPE(_rx_engine) == SVt_REGEXP); \
388 &SvANY(_rx_engine)->engine; \
4450b630
NC
389 }))
390# define RX_SUBBEG(prog) \
391 (*({ \
7948fc08
RU
392 const REGEXP *const _rx_subbeg = (prog); \
393 assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \
394 &SvANY(_rx_subbeg)->subbeg; \
4450b630 395 }))
200ff815 396# define RX_OFFS(prog) \
4450b630 397 (*({ \
7948fc08
RU
398 const REGEXP *const _rx_offs = (prog); \
399 assert(SvTYPE(_rx_offs) == SVt_REGEXP); \
400 &SvANY(_rx_offs)->offs; \
4450b630
NC
401 }))
402# define RX_NPARENS(prog) \
403 (*({ \
7948fc08
RU
404 const REGEXP *const _rx_nparens = (prog); \
405 assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \
406 &SvANY(_rx_nparens)->nparens; \
4450b630
NC
407 }))
408#else
409# define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
410# define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
411# define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
412# define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
413# define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
414#endif
288b8c02 415#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
288b8c02
NC
416#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
417#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
418#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
419#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
420#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
421#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 422#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 423
be8e71aa
YO
424#endif /* PLUGGABLE_RE_EXTENSION */
425
426/* Stuff that needs to be included in the plugable extension goes below here */
427
f8c7b90f 428#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 429#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC
430 STMT_START {if (RX_SAVED_COPY(rx)) { \
431 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC
432 } \
433 if (RX_MATCH_COPIED(rx)) { \
07bc277f 434 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
435 RX_MATCH_COPIED_off(rx); \
436 }} STMT_END
437#else
438#define RX_MATCH_COPY_FREE(rx) \
439 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 440 Safefree(RX_SUBBEG(rx)); \
ed252734
NC
441 RX_MATCH_COPIED_off(rx); \
442 }} STMT_END
443#endif
444
07bc277f
NC
445#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
446#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
447#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
448#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
449#define RX_MATCH_UTF8_set(prog, t) ((t) \
450 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
451 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
efd26800
NC
452
453/* Whether the pattern stored at RX_WRAPPED is in UTF-8 */
8f6ae13c 454#define RX_UTF8(prog) SvUTF8(prog)
7948fc08 455
cad2e5aa
JH
456#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
457#define REXEC_CHECKED 0x02 /* check_substr already checked. */
458#define REXEC_SCREAM 0x04 /* use scream table. */
459#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 460#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 461
7fe44985 462#if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN)
288b8c02
NC
463# define ReREFCNT_inc(re) \
464 ({ \
465 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
466 REGEXP *const _rerefcnt_inc = (re); \
467 assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \
468 SvREFCNT_inc(_rerefcnt_inc); \
469 _rerefcnt_inc; \
288b8c02
NC
470 })
471# define ReREFCNT_dec(re) \
472 ({ \
473 /* This is here to generate a casting warning if incorrect. */ \
7948fc08
RU
474 REGEXP *const _rerefcnt_dec = (re); \
475 SvREFCNT_dec(_rerefcnt_dec); \
288b8c02
NC
476 })
477#else
478# define ReREFCNT_dec(re) SvREFCNT_dec(re)
b7e0bd9e 479# define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re))
288b8c02
NC
480#endif
481
482/* FIXME for plugins. */
cf93c79d
IZ
483
484#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
485#define FBMcf_TAIL_DOLLARM 2
486#define FBMcf_TAIL_Z 4
487#define FBMcf_TAIL_z 8
488#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
489
490#define FBMrf_MULTILINE 1
cad2e5aa 491
3b0527fe
DM
492/* some basic information about the current match that is created by
493 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
494
495typedef struct {
288b8c02 496 REGEXP *prog;
3b0527fe
DM
497 char *bol;
498 char *till;
499 SV *sv;
500 char *ganch;
24b23f37 501 char *cutpoint;
3b0527fe
DM
502} regmatch_info;
503
5d9a96ca
DM
504
505/* structures for holding and saving the state maintained by regmatch() */
506
d56b3014 507#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 508#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 509#endif
6bda09f9 510
5d9a96ca
DM
511typedef I32 CHECKPOINT;
512
a0374537 513typedef struct regmatch_state {
40a82448 514 int resume_state; /* where to jump to on return */
24d3c4a9 515 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
516
517 union {
77cb431f
DM
518
519 /* this is a fake union member that matches the first element
520 * of each member that needs to store positive backtrack
521 * information */
522 struct {
523 struct regmatch_state *prev_yes_state;
524 } yes;
525
fae667d5
YO
526 /* branchlike members */
527 /* this is a fake union member that matches the first elements
528 * of each member that needs to behave like a branch */
529 struct {
5d458dd8
YO
530 /* this first element must match u.yes */
531 struct regmatch_state *prev_yes_state;
fae667d5
YO
532 U32 lastparen;
533 CHECKPOINT cp;
534
535 } branchlike;
536
537 struct {
538 /* the first elements must match u.branchlike */
539 struct regmatch_state *prev_yes_state;
540 U32 lastparen;
541 CHECKPOINT cp;
542
543 regnode *next_branch; /* next branch node */
544 } branch;
545
546 struct {
547 /* the first elements must match u.branchlike */
548 struct regmatch_state *prev_yes_state;
549 U32 lastparen;
550 CHECKPOINT cp;
551
2e64971a 552 U32 accepted; /* how many accepting states left */
7f69552c 553 U16 *jump; /* positive offsets from me */
166ba7cd 554 regnode *B; /* node following the trie */
7f69552c 555 regnode *me; /* Which node am I - needed for jump tries*/
2e64971a
DM
556 U8 *firstpos;/* pos in string of first trie match */
557 U32 firstchars;/* len in chars of firstpos from start */
558 U16 nextword;/* next word to try */
559 U16 topword; /* longest accepted word */
560 bool longfold;/* saw a fold with a 1->n char mapping */
e822a8b4
DM
561 } trie;
562
fae667d5
YO
563 /* special types - these members are used to store state for special
564 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 565 struct {
77cb431f
DM
566 /* this first element must match u.yes */
567 struct regmatch_state *prev_yes_state;
faec1544
DM
568 struct regmatch_state *prev_eval;
569 struct regmatch_state *prev_curlyx;
288b8c02 570 REGEXP *prev_rex;
faec1544
DM
571 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
572 flip when transitioning between
573 inner and outer rexen */
aa283a38
DM
574 CHECKPOINT cp; /* remember current savestack indexes */
575 CHECKPOINT lastcp;
6bda09f9 576 U32 close_paren; /* which close bracket is our end */
1b5c067c 577 regnode *B; /* the node following us */
e822a8b4
DM
578 } eval;
579
580 struct {
c476f425 581 /* this first element must match u.yes */
262b90c4 582 struct regmatch_state *prev_yes_state;
fae667d5
YO
583 I32 wanted;
584 I32 logical; /* saved copy of 'logical' var */
585 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
586 } ifmatch; /* and SUSPEND/UNLESSM */
587
588 struct {
589 /* this first element must match u.yes */
590 struct regmatch_state *prev_yes_state;
591 struct regmatch_state *prev_mark;
592 SV* mark_name;
593 char *mark_loc;
594 } mark;
595
596 struct {
597 int val;
598 } keeper;
599
600 /* quantifiers - these members are used for storing state for
601 for the regops used to implement quantifiers */
602 struct {
603 /* this first element must match u.yes */
604 struct regmatch_state *prev_yes_state;
c476f425 605 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
d02d6d97
DM
606 regnode *me; /* the CURLYX node */
607 regnode *B; /* the B node in /A*B/ */
c476f425
DM
608 CHECKPOINT cp; /* remember current savestack index */
609 bool minmod;
a0374537 610 int parenfloor;/* how far back to strip paren data */
c476f425
DM
611
612 /* these two are modified by WHILEM */
613 int count; /* how many instances of A we've matched */
614 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
615 } curlyx;
616
617 struct {
c476f425 618 /* this first element must match u.yes */
262b90c4 619 struct regmatch_state *prev_yes_state;
c476f425
DM
620 struct regmatch_state *save_curlyx;
621 CHECKPOINT cp; /* remember current savestack indexes */
622 CHECKPOINT lastcp;
623 char *save_lastloc; /* previous curlyx.lastloc */
624 I32 cache_offset;
625 I32 cache_mask;
e822a8b4
DM
626 } whilem;
627
628 struct {
5d458dd8
YO
629 /* this first element must match u.yes */
630 struct regmatch_state *prev_yes_state;
e822a8b4 631 I32 c1, c2; /* case fold search */
40a82448
DM
632 CHECKPOINT cp;
633 I32 alen; /* length of first-matched A string */
634 I32 count;
0cadcf80 635 bool minmod;
40a82448
DM
636 regnode *A, *B; /* the nodes corresponding to /A*B/ */
637 regnode *me; /* the curlym node */
e822a8b4
DM
638 } curlym;
639
640 struct {
3b6647e0 641 U32 paren;
c255a977 642 CHECKPOINT cp;
e822a8b4 643 I32 c1, c2; /* case fold search */
c255a977
DM
644 char *maxpos; /* highest possible point in string to match */
645 char *oldloc; /* the previous locinput */
e822a8b4 646 int count;
c255a977
DM
647 int min, max; /* {m,n} */
648 regnode *A, *B; /* the nodes corresponding to /A*B/ */
649 } curly; /* and CURLYN/PLUS/STAR */
dad79028 650
d8319b27 651 } u;
5d9a96ca
DM
652} regmatch_state;
653
654/* how many regmatch_state structs to allocate as a single slab.
655 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
656 * pointers, plus 1 for any mythical malloc overhead. */
657
658#define PERL_REGMATCH_SLAB_SLOTS \
659 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
660
661typedef struct regmatch_slab {
662 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
663 struct regmatch_slab *prev, *next;
664} regmatch_slab;
1ade1aa1 665
46ab3289
NC
666#define PL_reg_flags PL_reg_state.re_state_reg_flags
667#define PL_bostr PL_reg_state.re_state_bostr
668#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 669#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 670#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
671#define PL_reglastparen PL_reg_state.re_state_reglastparen
672#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
673#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
674#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
675#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
676#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
677#define PL_reg_magic PL_reg_state.re_state_reg_magic
678#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
679#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
680#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
681#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
682#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
683#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
684#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
685#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
686#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
687#define PL_regsize PL_reg_state.re_state_regsize
688#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
689#define PL_nrs PL_reg_state.re_state_nrs
690
1ade1aa1
NC
691struct re_save_state {
692 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
693 U32 re_state_reg_start_tmpl; /* from regexec.c */
694 I32 re_state_reg_eval_set; /* from regexec.c */
695 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
696 char *re_state_bostr;
697 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 698 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 699 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
700 U32 *re_state_reglastparen; /* Similarly for lastparen. */
701 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 702 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 703 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
704 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
705 PMOP *re_state_reg_curpm; /* from regexec.c */
706 char *re_state_reg_oldsaved; /* old saved substr during match */
707 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
708 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
709 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
710 I32 re_state_reg_maxiter; /* max wait until caching pos */
711 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 712 U32 re_state_regsize; /* from regexec.c */
1b5c067c 713 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
714 char *re_state_reg_starttry; /* from regexec.c */
715#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 716 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
717#endif
718};
719
720#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
721 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 722
1ade1aa1
NC
723/*
724 * Local variables:
725 * c-indentation-style: bsd
726 * c-basic-offset: 4
727 * indent-tabs-mode: t
728 * End:
729 *
730 * ex: set ts=8 sts=4 sw=4 noet:
731 */