This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
https://perl5.git.perl.org / perl5.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Protect me (and my fellow muppets) from screens of "syntax errors" if
[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 {
288b8c02
NC		 70 _XPV_HEAD;
71 _XPVMG_HEAD;
f8149455
YO		 72 /* what engine created this regexp? */
73 const struct regexp_engine* engine;
288b8c02 74 REGEXP *mother_re; /* what re is this a lightweight copy of? */
f8149455
YO		 75
76 /* Information about the match that the perl core uses to manage things */
bbe252da 77 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO		 78 I32 minlen; /* mininum possible length of string to match */
79 I32 minlenret; /* mininum possible length of $& */
58e23c8d 80 U32 gofs; /* chars left of pos that we search from */
f8149455 81 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 82 in the final match, used for optimisations */
f8149455 83 U32 nparens; /* number of capture buffers */
bbe252da 84
f8149455
YO		 85 /* private engine specific data */
86 U32 intflags; /* Engine Specific Internal flags */
87 void *pprivate; /* Data private to the regex engine which
88 created this object. */
89
90 /* Data about the last/current match. These are modified during matching*/
91 U32 lastparen; /* last open paren matched */
92 U32 lastcloseparen; /* last close paren matched */
f0ab9afb
NC		 93 regexp_paren_pair *swap; /* Swap copy of *offs */
94 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
28d8d7f4 95
bbe252da
YO		 96 char *subbeg; /* saved or original string
97 so \digit works forever. */
bbe252da 98 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
1b5c067c 99 I32 sublen; /* Length of string pointed by subbeg */
f8149455
YO		 100
101
102 /* Information about the match that isn't often used */
d7b33318 103 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
bcdf7404
YO		 104 char *wrapped; /* wrapped version of the pattern */
105 I32 wraplen; /* length of wrapped */
2c96c0d6
NC		 106 unsigned pre_prefix:4; /* offset from wrapped to the start of precomp */
107 unsigned seen_evals:28; /* number of eval groups in the pattern - for security checks */
f8149455 108 HV *paren_names; /* Optional hash of paren names */
f8fc2ecf
YO		 109} regexp;
110
785a26d5 111/* used for high speed searches */
f9f4320a
YO		 112typedef struct re_scream_pos_data_s
113{
114 char **scream_olds; /* match pos */
115 I32 *scream_pos; /* Internal iterator of scream. */
116} re_scream_pos_data;
117
785a26d5
YO		 118/* regexp_engine structure. This is the dispatch table for regexes.
119 * Any regex engine implementation must be able to build one of these.
120 */
f9f4320a 121typedef struct regexp_engine {
3ab4a224 122 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
49d7dfbc
AB		 123 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
124 char* strbeg, I32 minend, SV* screamer,
125 void* data, U32 flags);
126 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a
YO		 127 char *strend, const U32 flags,
128 re_scream_pos_data *data);
49d7dfbc
AB		 129 SV* (*checkstr) (pTHX_ REGEXP * const rx);
130 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 131 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 132 SV * const sv);
2fdbfb4d
AB		 133 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
134 SV const * const value);
135 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
136 const I32 paren);
192b9cd1
AB		 137 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
138 SV * const value, const U32 flags);
139 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
140 const U32 flags);
49d7dfbc 141 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 142#ifdef USE_ITHREADS
49d7dfbc 143 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 144#endif
f9f4320a
YO		 145} regexp_engine;
146
192b9cd1
AB		 147/*
148 These are passed to the numbered capture variable callbacks as the
149 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
150 $2 etc.
151*/
f1b875a0
YO		 152#define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */
153#define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */
154#define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */
192b9cd1
AB		 155
156/*
157 Flags that are passed to the named_buff and named_buff_iter
158 callbacks above. Those routines are called from universal.c via the
159 Tie::Hash::NamedCapture interface for %+ and %- and the re::
160 functions in the same file.
161*/
162
163/* The Tie::Hash::NamedCapture operation this is part of, if any */
f1b875a0
YO		 164#define RXapif_FETCH 0x0001
165#define RXapif_STORE 0x0002
166#define RXapif_DELETE 0x0004
167#define RXapif_CLEAR 0x0008
168#define RXapif_EXISTS 0x0010
169#define RXapif_SCALAR 0x0020
170#define RXapif_FIRSTKEY 0x0040
171#define RXapif_NEXTKEY 0x0080
192b9cd1
AB		 172
173/* Whether %+ or %- is being operated on */
f1b875a0
YO		 174#define RXapif_ONE 0x0100 /* %+ */
175#define RXapif_ALL 0x0200 /* %- */
192b9cd1
AB		 176
177/* Whether this is being called from a re:: function */
f1b875a0
YO		 178#define RXapif_REGNAME 0x0400
179#define RXapif_REGNAMES 0x0800
180#define RXapif_REGNAMES_COUNT 0x1000
192b9cd1 181
f7e71195
AB		 182/*
183=head1 REGEXP Functions
184
185=for apidoc Am|REGEXP *|SvRX|SV *sv
186
187Convenience macro to get the REGEXP from a SV. This is approximately
188equivalent to the following snippet:
189
190 if (SvMAGICAL(sv))
191 mg_get(sv);
192 if (SvROK(sv) &&
193 (tmpsv = (SV*)SvRV(sv)) &&
194 SvTYPE(tmpsv) == SVt_PVMG &&
195 (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr)))
196 {
197 return (REGEXP *)tmpmg->mg_obj;
198 }
199
200NULL will be returned if a REGEXP* is not found.
201
202=for apidoc Am|bool|SvRXOK|SV* sv
203
204Returns a boolean indicating whether the SV contains qr magic
205(PERL_MAGIC_qr).
206
207If you want to do something with the REGEXP* later use SvRX instead
208and check for NULL.
209
210=cut
211*/
212
213#define SvRX(sv) (Perl_get_re_arg(aTHX_ sv))
214#define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE)
215
216
bbe252da
YO		 217/* Flags stored in regexp->extflags
218 * These are used by code external to the regexp engine
e357fc67
YO		 219 *
220 * Note that flags starting with RXf_PMf_ have exact equivalents
221 * stored in op_pmflags and which are defined in op.h, they are defined
222 * numerically here only for clarity.
cb5027f2
YO		 223 *
224 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
225 * so that regnodes.h is updated with the changes.
226 *
bbe252da
YO		 227 */
228
52d81aa8 229/* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME
a1cf41cd
NC		 230 * If you change these you need to change the equivalent flags in op.h, and
231 * vice versa. */
52d81aa8
NC		 232#define RXf_PMf_MULTILINE 0x00000001 /* /m */
233#define RXf_PMf_SINGLELINE 0x00000002 /* /s */
234#define RXf_PMf_FOLD 0x00000004 /* /i */
235#define RXf_PMf_EXTENDED 0x00000008 /* /x */
236#define RXf_PMf_KEEPCOPY 0x00000010 /* /p */
237#define RXf_PMf_LOCALE 0x00000020 /* use locale */
e357fc67 238/* these flags are transfered from the PMOP->op_pmflags member during compilation */
52d81aa8 239#define RXf_PMf_STD_PMMOD_SHIFT 0
cde0cee5
YO		 240#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
241#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
242
a20207d7
YO		 243#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
244 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
245 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
246 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
247 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 248
bcdf7404
YO		 249/* chars and strings used as regex pattern modifiers
250 * Singlular is a 'c'har, plural is a "string"
87e95b7f
YO		 251 *
252 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
253 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
254 * for its own uses. So 'k' is out as well.
bcdf7404
YO		 255 */
256#define EXEC_PAT_MOD 'e'
87e95b7f 257#define KEEPCOPY_PAT_MOD 'p'
bcdf7404
YO		 258#define ONCE_PAT_MOD 'o'
259#define GLOBAL_PAT_MOD 'g'
260#define CONTINUE_PAT_MOD 'c'
261#define MULTILINE_PAT_MOD 'm'
262#define SINGLE_PAT_MOD 's'
263#define IGNORE_PAT_MOD 'i'
264#define XTENDED_PAT_MOD 'x'
265
266#define ONCE_PAT_MODS "o"
87e95b7f 267#define KEEPCOPY_PAT_MODS "p"
bcdf7404
YO		 268#define EXEC_PAT_MODS "e"
269#define LOOP_PAT_MODS "gc"
270
271#define STD_PAT_MODS "msix"
272
273#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
274
275#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
276#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
277#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
278#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
279
cb5027f2
YO		 280/*
281 * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl
282 * so that regnodes.h is updated with the changes.
283 *
284 */
bcdf7404 285
52d81aa8
NC		 286/* Anchor and GPOS related stuff */
287#define RXf_ANCH_BOL 0x00000100
288#define RXf_ANCH_MBOL 0x00000200
289#define RXf_ANCH_SBOL 0x00000400
290#define RXf_ANCH_GPOS 0x00000800
291#define RXf_GPOS_SEEN 0x00001000
292#define RXf_GPOS_FLOAT 0x00002000
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 */
52d81aa8
NC		 299#define RXf_LOOKBEHIND_SEEN 0x00004000
300#define RXf_EVAL_SEEN 0x00008000
301#define RXf_CANY_SEEN 0x00010000
bbe252da
YO		 302
303/* Special */
52d81aa8
NC		 304#define RXf_NOSCAN 0x00020000
305#define RXf_CHECK_ALL 0x00040000
bbe252da
YO		 306
307/* UTF8 related */
52d81aa8
NC		 308#define RXf_UTF8 0x00080000
309#define RXf_MATCH_UTF8 0x00100000
bbe252da
YO		 310
311/* Intuit related */
52d81aa8
NC		 312#define RXf_USE_INTUIT_NOML 0x00200000
313#define RXf_USE_INTUIT_ML 0x00400000
314#define RXf_INTUIT_TAIL 0x00800000
0ac6acae
AB		 315
316/*
317 Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will
318 be used by regex engines to check whether they should set
319 RXf_SKIPWHITE
320*/
52d81aa8 321#define RXf_SPLIT 0x01000000
0ac6acae 322
bbe252da
YO		 323#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
324
325/* Copy and tainted info */
52d81aa8
NC		 326#define RXf_COPY_DONE 0x02000000
327#define RXf_TAINTED_SEEN 0x04000000
328#define RXf_TAINTED 0x08000000 /* this pattern is tainted */
329
330/* Flags indicating special patterns */
331#define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */
332#define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */
333#define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */
334#define RXf_NULL 0x80000000 /* Pattern is // */
c737faaf 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 342#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
07bc277f
NC		 343#define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
344#define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN)
345#define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN)
346#define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN)
72311751
GS		 347#define RX_MATCH_TAINTED_set(prog, t) ((t) \
348 ? RX_MATCH_TAINTED_on(prog) \
349 : RX_MATCH_TAINTED_off(prog))
c277df42 350
07bc277f
NC		 351#define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE)
352#define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE)
353#define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE)
354#define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE)
355#define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
356#define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE)
cf93c79d
IZ		 357#define RX_MATCH_COPIED_set(prog,t) ((t) \
358 ? RX_MATCH_COPIED_on(prog) \
359 : RX_MATCH_COPIED_off(prog))
f9f4320a 360
07bc277f 361#define RXp_PRECOMP(rx) ((rx)->wrapped + (rx)->pre_prefix)
51aa2004
NC		 362/* FIXME? Are we hardcoding too much here and constraining plugin extension
363 writers? Specifically, the value 1 assumes that the wrapped version always
364 has exactly one character at the end, a ')'. Will that always be true? */
07bc277f
NC		 365#define RXp_PRELEN(rx) ((rx)->wraplen - (rx)->pre_prefix - 1)
366#define RXp_WRAPPED(rx) ((rx)->wrapped)
367#define RXp_WRAPLEN(rx) ((rx)->wraplen)
368#define RXp_EXTFLAGS(rx) ((rx)->extflags)
866c78d1 369
07bc277f 370/* For source compatibility. We used to store these explicitly. */
288b8c02
NC		 371#define RX_PRECOMP(prog) RXp_PRECOMP((struct regexp *)SvANY(prog))
372#define RX_PRELEN(prog) RXp_PRELEN((struct regexp *)SvANY(prog))
373#define RX_WRAPPED(prog) RXp_WRAPPED((struct regexp *)SvANY(prog))
374#define RX_WRAPLEN(prog) RXp_WRAPLEN((struct regexp *)SvANY(prog))
375#define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr)
376#define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog))
377#define RX_REFCNT(prog) SvREFCNT(prog)
378#define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine)
379#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
380#define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs)
381#define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens)
382#define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen)
383#define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg)
384#define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen)
385#define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret)
386#define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs)
387#define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen)
388#define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen)
389#define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals)
bdd9a1b1 390#define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy)
220fc49f 391
be8e71aa
YO		 392#endif /* PLUGGABLE_RE_EXTENSION */
393
394/* Stuff that needs to be included in the plugable extension goes below here */
395
f8c7b90f 396#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 397#define RX_MATCH_COPY_FREE(rx) \
bdd9a1b1
NC		 398 STMT_START {if (RX_SAVED_COPY(rx)) { \
399 SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \
ed252734
NC		 400 } \
401 if (RX_MATCH_COPIED(rx)) { \
07bc277f 402 Safefree(RX_SUBBEG(rx)); \
ed252734
NC		 403 RX_MATCH_COPIED_off(rx); \
404 }} STMT_END
405#else
406#define RX_MATCH_COPY_FREE(rx) \
407 STMT_START {if (RX_MATCH_COPIED(rx)) { \
07bc277f 408 Safefree(RX_SUBBEG(rx)); \
ed252734
NC		 409 RX_MATCH_COPIED_off(rx); \
410 }} STMT_END
411#endif
412
07bc277f
NC		 413#define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8)
414#define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8)
415#define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8)
416#define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS		 417#define RX_MATCH_UTF8_set(prog, t) ((t) \
418 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
419 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
420
cad2e5aa
JH		 421#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
422#define REXEC_CHECKED 0x02 /* check_substr already checked. */
423#define REXEC_SCREAM 0x04 /* use scream table. */
424#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 425#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 426
288b8c02
NC		 427#if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC)
428# define ReREFCNT_inc(re) \
429 ({ \
430 /* This is here to generate a casting warning if incorrect. */ \
431 REGEXP *const zwapp = (re); \
432 SvREFCNT_inc(zwapp); \
433 })
434# define ReREFCNT_dec(re) \
435 ({ \
436 /* This is here to generate a casting warning if incorrect. */ \
437 REGEXP *const boff = (re); \
438 SvREFCNT_dec(boff); \
439 })
440#else
441# define ReREFCNT_dec(re) SvREFCNT_dec(re)
442# define ReREFCNT_inc(re) SvREFCNT_inc(re)
443#endif
444
445/* FIXME for plugins. */
cf93c79d
IZ		 446
447#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH		 448#define FBMcf_TAIL_DOLLARM 2
449#define FBMcf_TAIL_Z 4
450#define FBMcf_TAIL_z 8
451#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ		 452
453#define FBMrf_MULTILINE 1
cad2e5aa 454
5d9a96ca
DM		 455/* an accepting state/position*/
456struct _reg_trie_accepted {
457 U8 *endpos;
458 U16 wordnum;
459};
460typedef struct _reg_trie_accepted reg_trie_accepted;
461
3b0527fe
DM		 462/* some basic information about the current match that is created by
463 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
464
465typedef struct {
288b8c02 466 REGEXP *prog;
3b0527fe
DM		 467 char *bol;
468 char *till;
469 SV *sv;
470 char *ganch;
24b23f37 471 char *cutpoint;
3b0527fe
DM		 472} regmatch_info;
473
5d9a96ca
DM		 474
475/* structures for holding and saving the state maintained by regmatch() */
476
d56b3014 477#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 478#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 479#endif
6bda09f9 480
5d9a96ca
DM		 481typedef I32 CHECKPOINT;
482
a0374537 483typedef struct regmatch_state {
40a82448 484 int resume_state; /* where to jump to on return */
24d3c4a9 485 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM		 486
487 union {
77cb431f
DM		 488
489 /* this is a fake union member that matches the first element
490 * of each member that needs to store positive backtrack
491 * information */
492 struct {
493 struct regmatch_state *prev_yes_state;
494 } yes;
495
fae667d5
YO		 496 /* branchlike members */
497 /* this is a fake union member that matches the first elements
498 * of each member that needs to behave like a branch */
499 struct {
5d458dd8
YO		 500 /* this first element must match u.yes */
501 struct regmatch_state *prev_yes_state;
fae667d5
YO		 502 U32 lastparen;
503 CHECKPOINT cp;
504
505 } branchlike;
506
507 struct {
508 /* the first elements must match u.branchlike */
509 struct regmatch_state *prev_yes_state;
510 U32 lastparen;
511 CHECKPOINT cp;
512
513 regnode *next_branch; /* next branch node */
514 } branch;
515
516 struct {
517 /* the first elements must match u.branchlike */
518 struct regmatch_state *prev_yes_state;
519 U32 lastparen;
520 CHECKPOINT cp;
521
522 reg_trie_accepted *accept_buff; /* accepting states we have seen */
166ba7cd 523 U32 accepted; /* how many accepting states we have seen */
7f69552c 524 U16 *jump; /* positive offsets from me */
166ba7cd 525 regnode *B; /* node following the trie */
7f69552c 526 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM		 527 } trie;
528
fae667d5
YO		 529 /* special types - these members are used to store state for special
530 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 531 struct {
77cb431f
DM		 532 /* this first element must match u.yes */
533 struct regmatch_state *prev_yes_state;
faec1544
DM		 534 struct regmatch_state *prev_eval;
535 struct regmatch_state *prev_curlyx;
288b8c02 536 REGEXP *prev_rex;
faec1544
DM		 537 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
538 flip when transitioning between
539 inner and outer rexen */
aa283a38
DM		 540 CHECKPOINT cp; /* remember current savestack indexes */
541 CHECKPOINT lastcp;
6bda09f9 542 U32 close_paren; /* which close bracket is our end */
1b5c067c 543 regnode *B; /* the node following us */
e822a8b4
DM		 544 } eval;
545
546 struct {
c476f425 547 /* this first element must match u.yes */
262b90c4 548 struct regmatch_state *prev_yes_state;
fae667d5
YO		 549 I32 wanted;
550 I32 logical; /* saved copy of 'logical' var */
551 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
552 } ifmatch; /* and SUSPEND/UNLESSM */
553
554 struct {
555 /* this first element must match u.yes */
556 struct regmatch_state *prev_yes_state;
557 struct regmatch_state *prev_mark;
558 SV* mark_name;
559 char *mark_loc;
560 } mark;
561
562 struct {
563 int val;
564 } keeper;
565
566 /* quantifiers - these members are used for storing state for
567 for the regops used to implement quantifiers */
568 struct {
569 /* this first element must match u.yes */
570 struct regmatch_state *prev_yes_state;
c476f425 571 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 572 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425
DM		 573 CHECKPOINT cp; /* remember current savestack index */
574 bool minmod;
a0374537 575 int parenfloor;/* how far back to strip paren data */
c476f425
DM		 576 int min; /* the minimal number of A's to match */
577 int max; /* the maximal number of A's to match */
c476f425
DM		 578
579 /* these two are modified by WHILEM */
580 int count; /* how many instances of A we've matched */
581 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM		 582 } curlyx;
583
584 struct {
c476f425 585 /* this first element must match u.yes */
262b90c4 586 struct regmatch_state *prev_yes_state;
c476f425
DM		 587 struct regmatch_state *save_curlyx;
588 CHECKPOINT cp; /* remember current savestack indexes */
589 CHECKPOINT lastcp;
590 char *save_lastloc; /* previous curlyx.lastloc */
591 I32 cache_offset;
592 I32 cache_mask;
e822a8b4
DM		 593 } whilem;
594
595 struct {
5d458dd8
YO		 596 /* this first element must match u.yes */
597 struct regmatch_state *prev_yes_state;
e822a8b4 598 I32 c1, c2; /* case fold search */
40a82448
DM		 599 CHECKPOINT cp;
600 I32 alen; /* length of first-matched A string */
601 I32 count;
0cadcf80 602 bool minmod;
40a82448
DM		 603 regnode *A, *B; /* the nodes corresponding to /A*B/ */
604 regnode *me; /* the curlym node */
e822a8b4
DM		 605 } curlym;
606
607 struct {
3b6647e0 608 U32 paren;
c255a977 609 CHECKPOINT cp;
e822a8b4 610 I32 c1, c2; /* case fold search */
c255a977
DM		 611 char *maxpos; /* highest possible point in string to match */
612 char *oldloc; /* the previous locinput */
e822a8b4 613 int count;
c255a977
DM		 614 int min, max; /* {m,n} */
615 regnode *A, *B; /* the nodes corresponding to /A*B/ */
616 } curly; /* and CURLYN/PLUS/STAR */
dad79028 617
d8319b27 618 } u;
5d9a96ca
DM		 619} regmatch_state;
620
621/* how many regmatch_state structs to allocate as a single slab.
622 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
623 * pointers, plus 1 for any mythical malloc overhead. */
624
625#define PERL_REGMATCH_SLAB_SLOTS \
626 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
627
628typedef struct regmatch_slab {
629 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
630 struct regmatch_slab *prev, *next;
631} regmatch_slab;
1ade1aa1 632
46ab3289
NC		 633#define PL_reg_flags PL_reg_state.re_state_reg_flags
634#define PL_bostr PL_reg_state.re_state_bostr
635#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 636#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 637#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289
NC		 638#define PL_reglastparen PL_reg_state.re_state_reglastparen
639#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC		 640#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
641#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
642#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC		 643#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
644#define PL_reg_magic PL_reg_state.re_state_reg_magic
645#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
646#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
647#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
648#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
649#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
650#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
651#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
652#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
653#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
654#define PL_regsize PL_reg_state.re_state_regsize
655#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
656#define PL_nrs PL_reg_state.re_state_nrs
657
1ade1aa1
NC		 658struct re_save_state {
659 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c
NC		 660 U32 re_state_reg_start_tmpl; /* from regexec.c */
661 I32 re_state_reg_eval_set; /* from regexec.c */
662 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1
NC		 663 char *re_state_bostr;
664 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 665 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 666 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1
NC		 667 U32 *re_state_reglastparen; /* Similarly for lastparen. */
668 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 669 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 670 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1
NC		 671 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
672 PMOP *re_state_reg_curpm; /* from regexec.c */
673 char *re_state_reg_oldsaved; /* old saved substr during match */
674 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c
NC		 675 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
676 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1
NC		 677 I32 re_state_reg_maxiter; /* max wait until caching pos */
678 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 679 U32 re_state_regsize; /* from regexec.c */
1b5c067c 680 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1
NC		 681 char *re_state_reg_starttry; /* from regexec.c */
682#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 683 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		 684#endif
685};
686
687#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
688 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 689
1ade1aa1
NC		 690/*
691 * Local variables:
692 * c-indentation-style: bsd
693 * c-basic-offset: 4
694 * indent-tabs-mode: t
695 * End:
696 *
697 * ex: set ts=8 sts=4 sw=4 noet:
698 */
Unnamed repository; edit this file 'description' to name the repository.