This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Note to future self about moving the regexp flag bits around.
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
663f364b 4 * 2005, 2006, 2007, 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
bbe252da 69typedef struct regexp {
f8149455
YO
70 /* what engine created this regexp? */
71 const struct regexp_engine* engine;
28d8d7f4 72 struct regexp* mother_re; /* what re is this a lightweight copy of? */
f8149455
YO
73
74 /* Information about the match that the perl core uses to manage things */
bbe252da 75 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO
76 I32 minlen; /* mininum possible length of string to match */
77 I32 minlenret; /* mininum possible length of $& */
58e23c8d 78 U32 gofs; /* chars left of pos that we search from */
f8149455 79 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 80 in the final match, used for optimisations */
f8149455 81 U32 nparens; /* number of capture buffers */
bbe252da 82
f8149455
YO
83 /* private engine specific data */
84 U32 intflags; /* Engine Specific Internal flags */
85 void *pprivate; /* Data private to the regex engine which
86 created this object. */
87
88 /* Data about the last/current match. These are modified during matching*/
89 U32 lastparen; /* last open paren matched */
90 U32 lastcloseparen; /* last close paren matched */
f0ab9afb
NC
91 regexp_paren_pair *swap; /* Swap copy of *offs */
92 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
28d8d7f4 93
bbe252da
YO
94 char *subbeg; /* saved or original string
95 so \digit works forever. */
bbe252da 96 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
1b5c067c 97 I32 sublen; /* Length of string pointed by subbeg */
f8149455
YO
98
99
100 /* Information about the match that isn't often used */
d7b33318 101 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
bcdf7404
YO
102 char *wrapped; /* wrapped version of the pattern */
103 I32 wraplen; /* length of wrapped */
2c96c0d6
NC
104 unsigned pre_prefix:4; /* offset from wrapped to the start of precomp */
105 unsigned seen_evals:28; /* number of eval groups in the pattern - for security checks */
f8149455
YO
106 HV *paren_names; /* Optional hash of paren names */
107
108 /* Refcount of this regexp */
109 I32 refcnt; /* Refcount of this regexp */
f8fc2ecf
YO
110} regexp;
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 {
3ab4a224 123 REGEXP* (*comp) (pTHX_ const SV * const pattern, const 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
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
bbe252da
YO
218/* Flags stored in regexp->extflags
219 * These are used by code external to the regexp engine
e357fc67
YO
220 *
221 * Note that flags starting with RXf_PMf_ have exact equivalents
222 * stored in op_pmflags and which are defined in op.h, they are defined
223 * numerically here only for clarity.
cb5027f2
YO
224 *
225 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
226 * so that regnodes.h is updated with the changes.
227 *
bbe252da
YO
228 */
229
230/* Anchor and GPOS related stuff */
231#define RXf_ANCH_BOL 0x00000001
232#define RXf_ANCH_MBOL 0x00000002
233#define RXf_ANCH_SBOL 0x00000004
234#define RXf_ANCH_GPOS 0x00000008
235#define RXf_GPOS_SEEN 0x00000010
236#define RXf_GPOS_FLOAT 0x00000020
e357fc67 237/* two bits here */
bbe252da
YO
238#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
239#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
e357fc67
YO
240#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
241
242/* Flags indicating special patterns */
c737faaf 243#define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */
e357fc67
YO
244#define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
245#define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
640f820d 246#define RXf_NULL 0x40000000 /* Pattern is // */
e357fc67 247
a1cf41cd
NC
248/* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME
249 * If you change these you need to change the equivalent flags in op.h, and
250 * vice versa. */
e357fc67
YO
251#define RXf_PMf_LOCALE 0x00000800 /* use locale */
252#define RXf_PMf_MULTILINE 0x00001000 /* /m */
253#define RXf_PMf_SINGLELINE 0x00002000 /* /s */
254#define RXf_PMf_FOLD 0x00004000 /* /i */
255#define RXf_PMf_EXTENDED 0x00008000 /* /x */
f7819f85 256#define RXf_PMf_KEEPCOPY 0x00010000 /* /p */
e357fc67 257/* these flags are transfered from the PMOP->op_pmflags member during compilation */
cde0cee5
YO
258#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
259#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
260
a20207d7
YO
261#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
262 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
263 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
264 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
265 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 266
bcdf7404
YO
267/* chars and strings used as regex pattern modifiers
268 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO
269 *
270 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
271 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
272 * for its own uses. So 'k' is out as well.
bcdf7404
YO
273 */
274#define EXEC_PAT_MOD 'e'
87e95b7f 275#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO
276#define ONCE_PAT_MOD 'o'
277#define GLOBAL_PAT_MOD 'g'
278#define CONTINUE_PAT_MOD 'c'
279#define MULTILINE_PAT_MOD 'm'
280#define SINGLE_PAT_MOD 's'
281#define IGNORE_PAT_MOD 'i'
282#define XTENDED_PAT_MOD 'x'
283
284#define ONCE_PAT_MODS "o"
87e95b7f 285#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO
286#define EXEC_PAT_MODS "e"
287#define LOOP_PAT_MODS "gc"
288
289#define STD_PAT_MODS "msix"
290
291#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
292
293#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
294#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
295#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
296#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
297
cb5027f2
YO
298/*
299 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
300 * so that regnodes.h is updated with the changes.
301 *
302 */
bcdf7404 303
bbe252da 304/* What we have seen */
bbe252da
YO
305#define RXf_LOOKBEHIND_SEEN 0x00020000
306#define RXf_EVAL_SEEN 0x00040000
307#define RXf_CANY_SEEN 0x00080000
308
309/* Special */
310#define RXf_NOSCAN 0x00100000
311#define RXf_CHECK_ALL 0x00200000
312
313/* UTF8 related */
314#define RXf_UTF8 0x00400000
995b21a2 315#define RXf_MATCH_UTF8 0x00800000
bbe252da
YO
316
317/* Intuit related */
318#define RXf_USE_INTUIT_NOML 0x01000000
319#define RXf_USE_INTUIT_ML 0x02000000
320#define RXf_INTUIT_TAIL 0x04000000
0ac6acae
AB
321
322/*
323 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
324 be used by regex engines to check whether they should set
325 RXf_SKIPWHITE
326*/
327#define RXf_SPLIT 0x08000000
328
bbe252da
YO
329#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
330
331/* Copy and tainted info */
332#define RXf_COPY_DONE 0x10000000
333#define RXf_TAINTED_SEEN 0x20000000
c737faaf
YO
334#define RXf_TAINTED 0x80000000 /* this pattern is tainted */
335
cb5027f2
YO
336/*
337 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
338 * so that regnodes.h is updated with the changes.
339 *
340 */
bbe252da 341
bbe252da
YO
342#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
343#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
344#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
345#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751
GS
346#define RX_MATCH_TAINTED_set(prog, t) ((t) \
347 ? RX_MATCH_TAINTED_on(prog) \
348 : RX_MATCH_TAINTED_off(prog))
c277df42 349
bbe252da
YO
350#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
351#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
352#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d
IZ
353#define RX_MATCH_COPIED_set(prog,t) ((t) \
354 ? RX_MATCH_COPIED_on(prog) \
355 : RX_MATCH_COPIED_off(prog))
f9f4320a 356
220fc49f 357/* For source compatibility. We used to store these explicitly. */
2c96c0d6 358#define RX_PRECOMP(prog) ((prog)->wrapped + (prog)->pre_prefix)
51aa2004
NC
359/* FIXME? Are we hardcoding too much here and constraining plugin extension
360 writers? Specifically, the value 1 assumes that the wrapped version always
361 has exactly one character at the end, a ')'. Will that always be true? */
362#define RX_PRELEN(prog) ((prog)->wraplen - (prog)->pre_prefix - 1)
866c78d1
NC
363#define RX_WRAPPED(prog) ((prog)->wrapped)
364#define RX_WRAPLEN(prog) ((prog)->wraplen)
365
220fc49f 366
be8e71aa
YO
367#endif /* PLUGGABLE_RE_EXTENSION */
368
369/* Stuff that needs to be included in the plugable extension goes below here */
370
f8c7b90f 371#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
372#define RX_MATCH_COPY_FREE(rx) \
373 STMT_START {if (rx->saved_copy) { \
374 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
375 } \
376 if (RX_MATCH_COPIED(rx)) { \
377 Safefree(rx->subbeg); \
378 RX_MATCH_COPIED_off(rx); \
379 }} STMT_END
380#else
381#define RX_MATCH_COPY_FREE(rx) \
382 STMT_START {if (RX_MATCH_COPIED(rx)) { \
383 Safefree(rx->subbeg); \
384 RX_MATCH_COPIED_off(rx); \
385 }} STMT_END
386#endif
387
995b21a2
YO
388#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
389#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
390#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
391#define RX_MATCH_UTF8_set(prog, t) ((t) \
392 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
393 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
394
cad2e5aa
JH
395#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
396#define REXEC_CHECKED 0x02 /* check_substr already checked. */
397#define REXEC_SCREAM 0x04 /* use scream table. */
398#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 399#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 400
155aba94 401#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 402#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
403
404#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
405#define FBMcf_TAIL_DOLLARM 2
406#define FBMcf_TAIL_Z 4
407#define FBMcf_TAIL_z 8
408#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
409
410#define FBMrf_MULTILINE 1
cad2e5aa 411
5d9a96ca
DM
412/* an accepting state/position*/
413struct _reg_trie_accepted {
414 U8 *endpos;
415 U16 wordnum;
416};
417typedef struct _reg_trie_accepted reg_trie_accepted;
418
3b0527fe
DM
419/* some basic information about the current match that is created by
420 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
421
422typedef struct {
423 regexp *prog;
424 char *bol;
425 char *till;
426 SV *sv;
427 char *ganch;
24b23f37 428 char *cutpoint;
3b0527fe
DM
429} regmatch_info;
430
5d9a96ca
DM
431
432/* structures for holding and saving the state maintained by regmatch() */
433
d56b3014 434#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 435#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 436#endif
6bda09f9 437
5d9a96ca
DM
438typedef I32 CHECKPOINT;
439
a0374537 440typedef struct regmatch_state {
40a82448 441 int resume_state; /* where to jump to on return */
24d3c4a9 442 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
443
444 union {
77cb431f
DM
445
446 /* this is a fake union member that matches the first element
447 * of each member that needs to store positive backtrack
448 * information */
449 struct {
450 struct regmatch_state *prev_yes_state;
451 } yes;
452
fae667d5
YO
453 /* branchlike members */
454 /* this is a fake union member that matches the first elements
455 * of each member that needs to behave like a branch */
456 struct {
5d458dd8
YO
457 /* this first element must match u.yes */
458 struct regmatch_state *prev_yes_state;
fae667d5
YO
459 U32 lastparen;
460 CHECKPOINT cp;
461
462 } branchlike;
463
464 struct {
465 /* the first elements must match u.branchlike */
466 struct regmatch_state *prev_yes_state;
467 U32 lastparen;
468 CHECKPOINT cp;
469
470 regnode *next_branch; /* next branch node */
471 } branch;
472
473 struct {
474 /* the first elements must match u.branchlike */
475 struct regmatch_state *prev_yes_state;
476 U32 lastparen;
477 CHECKPOINT cp;
478
479 reg_trie_accepted *accept_buff; /* accepting states we have seen */
166ba7cd 480 U32 accepted; /* how many accepting states we have seen */
7f69552c 481 U16 *jump; /* positive offsets from me */
166ba7cd 482 regnode *B; /* node following the trie */
7f69552c 483 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
484 } trie;
485
fae667d5
YO
486 /* special types - these members are used to store state for special
487 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 488 struct {
77cb431f
DM
489 /* this first element must match u.yes */
490 struct regmatch_state *prev_yes_state;
faec1544
DM
491 struct regmatch_state *prev_eval;
492 struct regmatch_state *prev_curlyx;
aa283a38 493 regexp *prev_rex;
faec1544
DM
494 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
495 flip when transitioning between
496 inner and outer rexen */
aa283a38
DM
497 CHECKPOINT cp; /* remember current savestack indexes */
498 CHECKPOINT lastcp;
6bda09f9 499 U32 close_paren; /* which close bracket is our end */
1b5c067c 500 regnode *B; /* the node following us */
e822a8b4
DM
501 } eval;
502
503 struct {
c476f425 504 /* this first element must match u.yes */
262b90c4 505 struct regmatch_state *prev_yes_state;
fae667d5
YO
506 I32 wanted;
507 I32 logical; /* saved copy of 'logical' var */
508 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
509 } ifmatch; /* and SUSPEND/UNLESSM */
510
511 struct {
512 /* this first element must match u.yes */
513 struct regmatch_state *prev_yes_state;
514 struct regmatch_state *prev_mark;
515 SV* mark_name;
516 char *mark_loc;
517 } mark;
518
519 struct {
520 int val;
521 } keeper;
522
523 /* quantifiers - these members are used for storing state for
524 for the regops used to implement quantifiers */
525 struct {
526 /* this first element must match u.yes */
527 struct regmatch_state *prev_yes_state;
c476f425 528 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 529 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425
DM
530 CHECKPOINT cp; /* remember current savestack index */
531 bool minmod;
a0374537 532 int parenfloor;/* how far back to strip paren data */
c476f425
DM
533 int min; /* the minimal number of A's to match */
534 int max; /* the maximal number of A's to match */
c476f425
DM
535
536 /* these two are modified by WHILEM */
537 int count; /* how many instances of A we've matched */
538 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
539 } curlyx;
540
541 struct {
c476f425 542 /* this first element must match u.yes */
262b90c4 543 struct regmatch_state *prev_yes_state;
c476f425
DM
544 struct regmatch_state *save_curlyx;
545 CHECKPOINT cp; /* remember current savestack indexes */
546 CHECKPOINT lastcp;
547 char *save_lastloc; /* previous curlyx.lastloc */
548 I32 cache_offset;
549 I32 cache_mask;
e822a8b4
DM
550 } whilem;
551
552 struct {
5d458dd8
YO
553 /* this first element must match u.yes */
554 struct regmatch_state *prev_yes_state;
e822a8b4 555 I32 c1, c2; /* case fold search */
40a82448
DM
556 CHECKPOINT cp;
557 I32 alen; /* length of first-matched A string */
558 I32 count;
0cadcf80 559 bool minmod;
40a82448
DM
560 regnode *A, *B; /* the nodes corresponding to /A*B/ */
561 regnode *me; /* the curlym node */
e822a8b4
DM
562 } curlym;
563
564 struct {
3b6647e0 565 U32 paren;
c255a977 566 CHECKPOINT cp;
e822a8b4 567 I32 c1, c2; /* case fold search */
c255a977
DM
568 char *maxpos; /* highest possible point in string to match */
569 char *oldloc; /* the previous locinput */
e822a8b4 570 int count;
c255a977
DM
571 int min, max; /* {m,n} */
572 regnode *A, *B; /* the nodes corresponding to /A*B/ */
573 } curly; /* and CURLYN/PLUS/STAR */
dad79028 574
d8319b27 575 } u;
5d9a96ca
DM
576} regmatch_state;
577
578/* how many regmatch_state structs to allocate as a single slab.
579 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
580 * pointers, plus 1 for any mythical malloc overhead. */
581
582#define PERL_REGMATCH_SLAB_SLOTS \
583 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
584
585typedef struct regmatch_slab {
586 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
587 struct regmatch_slab *prev, *next;
588} regmatch_slab;
1ade1aa1 589
46ab3289
NC
590#define PL_reg_flags PL_reg_state.re_state_reg_flags
591#define PL_bostr PL_reg_state.re_state_bostr
592#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 593#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 594#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC
595#define PL_reglastparen PL_reg_state.re_state_reglastparen
596#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
597#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
598#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
599#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
600#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
601#define PL_reg_magic PL_reg_state.re_state_reg_magic
602#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
603#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
604#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
605#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
606#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
607#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
608#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
609#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
610#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
611#define PL_regsize PL_reg_state.re_state_regsize
612#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
613#define PL_nrs PL_reg_state.re_state_nrs
614
1ade1aa1
NC
615struct re_save_state {
616 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC
617 U32 re_state_reg_start_tmpl; /* from regexec.c */
618 I32 re_state_reg_eval_set; /* from regexec.c */
619 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC
620 char *re_state_bostr;
621 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 622 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 623 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC
624 U32 *re_state_reglastparen; /* Similarly for lastparen. */
625 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 626 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 627 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC
628 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
629 PMOP *re_state_reg_curpm; /* from regexec.c */
630 char *re_state_reg_oldsaved; /* old saved substr during match */
631 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC
632 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
633 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC
634 I32 re_state_reg_maxiter; /* max wait until caching pos */
635 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 636 U32 re_state_regsize; /* from regexec.c */
1b5c067c 637 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC
638 char *re_state_reg_starttry; /* from regexec.c */
639#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 640 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
641#endif
642};
643
644#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
645 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 646
1ade1aa1
NC
647/*
648 * Local variables:
649 * c-indentation-style: bsd
650 * c-basic-offset: 4
651 * indent-tabs-mode: t
652 * End:
653 *
654 * ex: set ts=8 sts=4 sw=4 noet:
655 */