This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add Regexp::Keep \K functionality to regex engine as well as add \v and \V, cleanup...
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
699a97de 4 * 2005, 2006 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;
bbe252da 34
785a26d5
YO
35struct reg_substr_datum {
36 I32 min_offset;
37 I32 max_offset;
38 SV *substr; /* non-utf8 variant */
39 SV *utf8_substr; /* utf8 variant */
40 I32 end_shift;
41};
42struct reg_substr_data {
43 struct reg_substr_datum data[3]; /* Actual array */
44};
f9f4320a 45
f8c7b90f 46#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
47#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
48#else
49#define SV_SAVED_COPY
ed252734 50#endif
f8149455
YO
51/* this is ordered such that the most commonly used
52 fields are at the start of the struct */
bbe252da 53typedef struct regexp {
f8149455
YO
54 /* what engine created this regexp? */
55 const struct regexp_engine* engine;
56
57 /* Information about the match that the perl core uses to manage things */
bbe252da 58 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO
59 I32 minlen; /* mininum possible length of string to match */
60 I32 minlenret; /* mininum possible length of $& */
58e23c8d 61 U32 gofs; /* chars left of pos that we search from */
f8149455 62 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 63 in the final match, used for optimisations */
f8149455 64 U32 nparens; /* number of capture buffers */
bbe252da 65
f8149455
YO
66 /* private engine specific data */
67 U32 intflags; /* Engine Specific Internal flags */
68 void *pprivate; /* Data private to the regex engine which
69 created this object. */
70
71 /* Data about the last/current match. These are modified during matching*/
72 U32 lastparen; /* last open paren matched */
73 U32 lastcloseparen; /* last close paren matched */
74 I32 *startp; /* Array of offsets from start of string (@-) */
75 I32 *endp; /* Array of offsets from start of string (@+) */
bbe252da
YO
76 char *subbeg; /* saved or original string
77 so \digit works forever. */
78 I32 sublen; /* Length of string pointed by subbeg */
bbe252da 79 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
f8149455
YO
80
81
82 /* Information about the match that isn't often used */
83 char *precomp; /* pre-compilation regular expression */
84 I32 prelen; /* length of precomp */
85 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
86 HV *paren_names; /* Optional hash of paren names */
87
88 /* Refcount of this regexp */
89 I32 refcnt; /* Refcount of this regexp */
f8fc2ecf
YO
90} regexp;
91
785a26d5 92/* used for high speed searches */
f9f4320a
YO
93typedef struct re_scream_pos_data_s
94{
95 char **scream_olds; /* match pos */
96 I32 *scream_pos; /* Internal iterator of scream. */
97} re_scream_pos_data;
98
785a26d5
YO
99/* regexp_engine structure. This is the dispatch table for regexes.
100 * Any regex engine implementation must be able to build one of these.
101 */
f9f4320a 102typedef struct regexp_engine {
f2f78491
YO
103 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
104 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
f9f4320a
YO
105 char* strbeg, I32 minend, SV* screamer,
106 void* data, U32 flags);
f2f78491
YO
107 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
108 char *strend, U32 flags,
109 struct re_scream_pos_data_s *data);
110 SV* (*checkstr) (pTHX_ regexp *prog);
111 void (*free) (pTHX_ struct regexp* r);
112#ifdef USE_ITHREADS
f8149455 113 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
f9f4320a
YO
114#endif
115} regexp_engine;
116
bbe252da
YO
117/* Flags stored in regexp->extflags
118 * These are used by code external to the regexp engine
e357fc67
YO
119 *
120 * Note that flags starting with RXf_PMf_ have exact equivalents
121 * stored in op_pmflags and which are defined in op.h, they are defined
122 * numerically here only for clarity.
bbe252da
YO
123 */
124
125/* Anchor and GPOS related stuff */
126#define RXf_ANCH_BOL 0x00000001
127#define RXf_ANCH_MBOL 0x00000002
128#define RXf_ANCH_SBOL 0x00000004
129#define RXf_ANCH_GPOS 0x00000008
130#define RXf_GPOS_SEEN 0x00000010
131#define RXf_GPOS_FLOAT 0x00000020
e357fc67 132/* two bits here */
bbe252da
YO
133#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
134#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
e357fc67
YO
135#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
136
137/* Flags indicating special patterns */
138#define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
139#define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
140
141/* 0xF800 of extflags is used by (RXf_)PMf_COMPILETIME */
142#define RXf_PMf_LOCALE 0x00000800 /* use locale */
143#define RXf_PMf_MULTILINE 0x00001000 /* /m */
144#define RXf_PMf_SINGLELINE 0x00002000 /* /s */
145#define RXf_PMf_FOLD 0x00004000 /* /i */
146#define RXf_PMf_EXTENDED 0x00008000 /* /x */
147/* these flags are transfered from the PMOP->op_pmflags member during compilation */
bbe252da
YO
148#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
149
150/* What we have seen */
151/* one bit here */
152#define RXf_LOOKBEHIND_SEEN 0x00020000
153#define RXf_EVAL_SEEN 0x00040000
154#define RXf_CANY_SEEN 0x00080000
155
156/* Special */
157#define RXf_NOSCAN 0x00100000
158#define RXf_CHECK_ALL 0x00200000
159
160/* UTF8 related */
161#define RXf_UTF8 0x00400000
162#define RXf_MATCH_UTF8 0x00800000
163
164/* Intuit related */
165#define RXf_USE_INTUIT_NOML 0x01000000
166#define RXf_USE_INTUIT_ML 0x02000000
167#define RXf_INTUIT_TAIL 0x04000000
168/* one bit here */
169#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
170
171/* Copy and tainted info */
172#define RXf_COPY_DONE 0x10000000
173#define RXf_TAINTED_SEEN 0x20000000
174/* two bits here */
175
176
177#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
178#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
179#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
180#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751
GS
181#define RX_MATCH_TAINTED_set(prog, t) ((t) \
182 ? RX_MATCH_TAINTED_on(prog) \
183 : RX_MATCH_TAINTED_off(prog))
c277df42 184
bbe252da
YO
185#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
186#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
187#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d
IZ
188#define RX_MATCH_COPIED_set(prog,t) ((t) \
189 ? RX_MATCH_COPIED_on(prog) \
190 : RX_MATCH_COPIED_off(prog))
f9f4320a 191
be8e71aa
YO
192#endif /* PLUGGABLE_RE_EXTENSION */
193
194/* Stuff that needs to be included in the plugable extension goes below here */
195
f8c7b90f 196#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
197#define RX_MATCH_COPY_FREE(rx) \
198 STMT_START {if (rx->saved_copy) { \
199 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
200 } \
201 if (RX_MATCH_COPIED(rx)) { \
202 Safefree(rx->subbeg); \
203 RX_MATCH_COPIED_off(rx); \
204 }} STMT_END
205#else
206#define RX_MATCH_COPY_FREE(rx) \
207 STMT_START {if (RX_MATCH_COPIED(rx)) { \
208 Safefree(rx->subbeg); \
209 RX_MATCH_COPIED_off(rx); \
210 }} STMT_END
211#endif
212
bbe252da
YO
213#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
214#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
215#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
216#define RX_MATCH_UTF8_set(prog, t) ((t) \
217 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
218 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
219
cad2e5aa
JH
220#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
221#define REXEC_CHECKED 0x02 /* check_substr already checked. */
222#define REXEC_SCREAM 0x04 /* use scream table. */
223#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 224#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 225
155aba94 226#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 227#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
228
229#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
230#define FBMcf_TAIL_DOLLARM 2
231#define FBMcf_TAIL_Z 4
232#define FBMcf_TAIL_z 8
233#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
234
235#define FBMrf_MULTILINE 1
cad2e5aa 236
5d9a96ca
DM
237/* an accepting state/position*/
238struct _reg_trie_accepted {
239 U8 *endpos;
240 U16 wordnum;
241};
242typedef struct _reg_trie_accepted reg_trie_accepted;
243
3b0527fe
DM
244/* some basic information about the current match that is created by
245 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
246
247typedef struct {
248 regexp *prog;
249 char *bol;
250 char *till;
251 SV *sv;
252 char *ganch;
24b23f37 253 char *cutpoint;
3b0527fe
DM
254} regmatch_info;
255
5d9a96ca
DM
256
257/* structures for holding and saving the state maintained by regmatch() */
258
6bda09f9
YO
259#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
260
5d9a96ca
DM
261typedef I32 CHECKPOINT;
262
a0374537 263typedef struct regmatch_state {
40a82448 264 int resume_state; /* where to jump to on return */
24d3c4a9 265 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
266
267 union {
77cb431f
DM
268
269 /* this is a fake union member that matches the first element
270 * of each member that needs to store positive backtrack
271 * information */
272 struct {
273 struct regmatch_state *prev_yes_state;
274 } yes;
275
e822a8b4 276 struct {
5d458dd8
YO
277 /* this first element must match u.yes */
278 struct regmatch_state *prev_yes_state;
e822a8b4 279 reg_trie_accepted *accept_buff;
166ba7cd 280 U32 accepted; /* how many accepting states we have seen */
7f69552c 281 U16 *jump; /* positive offsets from me */
166ba7cd 282 regnode *B; /* node following the trie */
7f69552c 283 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
284 } trie;
285
286 struct {
77cb431f
DM
287 /* this first element must match u.yes */
288 struct regmatch_state *prev_yes_state;
faec1544
DM
289 struct regmatch_state *prev_eval;
290 struct regmatch_state *prev_curlyx;
aa283a38 291 regexp *prev_rex;
faec1544
DM
292 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
293 flip when transitioning between
294 inner and outer rexen */
aa283a38
DM
295 CHECKPOINT cp; /* remember current savestack indexes */
296 CHECKPOINT lastcp;
40a82448 297 regnode *B; /* the node following us */
6bda09f9 298 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
299 } eval;
300
301 struct {
c476f425 302 /* this first element must match u.yes */
262b90c4 303 struct regmatch_state *prev_yes_state;
c476f425
DM
304 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
305 CHECKPOINT cp; /* remember current savestack index */
306 bool minmod;
a0374537 307 int parenfloor;/* how far back to strip paren data */
c476f425
DM
308 int min; /* the minimal number of A's to match */
309 int max; /* the maximal number of A's to match */
310 regnode *A, *B; /* the nodes corresponding to /A*B/ */
311
312 /* these two are modified by WHILEM */
313 int count; /* how many instances of A we've matched */
314 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
315 } curlyx;
316
317 struct {
c476f425 318 /* this first element must match u.yes */
262b90c4 319 struct regmatch_state *prev_yes_state;
c476f425
DM
320 struct regmatch_state *save_curlyx;
321 CHECKPOINT cp; /* remember current savestack indexes */
322 CHECKPOINT lastcp;
323 char *save_lastloc; /* previous curlyx.lastloc */
324 I32 cache_offset;
325 I32 cache_mask;
e822a8b4
DM
326 } whilem;
327
328 struct {
5d458dd8
YO
329 /* this first element must match u.yes */
330 struct regmatch_state *prev_yes_state;
3b6647e0 331 U32 lastparen;
40a82448
DM
332 regnode *next_branch; /* next branch node */
333 CHECKPOINT cp;
334 } branch;
335
336 struct {
dad79028
DM
337 /* this first element must match u.yes */
338 struct regmatch_state *prev_yes_state;
e822a8b4 339 I32 c1, c2; /* case fold search */
40a82448
DM
340 CHECKPOINT cp;
341 I32 alen; /* length of first-matched A string */
342 I32 count;
0cadcf80 343 bool minmod;
40a82448
DM
344 regnode *A, *B; /* the nodes corresponding to /A*B/ */
345 regnode *me; /* the curlym node */
e822a8b4
DM
346 } curlym;
347
348 struct {
3b6647e0 349 U32 paren;
c255a977 350 CHECKPOINT cp;
e822a8b4 351 I32 c1, c2; /* case fold search */
c255a977
DM
352 char *maxpos; /* highest possible point in string to match */
353 char *oldloc; /* the previous locinput */
e822a8b4 354 int count;
c255a977
DM
355 int min, max; /* {m,n} */
356 regnode *A, *B; /* the nodes corresponding to /A*B/ */
357 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
358
359 struct {
360 /* this first element must match u.yes */
361 struct regmatch_state *prev_yes_state;
362 I32 wanted;
24d3c4a9 363 I32 logical; /* saved copy of 'logical' var */
40a82448 364 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 365 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
366
367 struct {
368 /* this first element must match u.yes */
369 struct regmatch_state *prev_yes_state;
370 struct regmatch_state *prev_mark;
371 SV* mark_name;
372 char *mark_loc;
373 } mark;
ee9b8eae
YO
374
375 struct {
376 int val;
377 } keeper;
d8319b27 378 } u;
5d9a96ca
DM
379} regmatch_state;
380
381/* how many regmatch_state structs to allocate as a single slab.
382 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
383 * pointers, plus 1 for any mythical malloc overhead. */
384
385#define PERL_REGMATCH_SLAB_SLOTS \
386 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
387
388typedef struct regmatch_slab {
389 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
390 struct regmatch_slab *prev, *next;
391} regmatch_slab;
1ade1aa1 392
46ab3289
NC
393#define PL_reg_flags PL_reg_state.re_state_reg_flags
394#define PL_bostr PL_reg_state.re_state_bostr
395#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
396#define PL_regeol PL_reg_state.re_state_regeol
397#define PL_regstartp PL_reg_state.re_state_regstartp
398#define PL_regendp PL_reg_state.re_state_regendp
399#define PL_reglastparen PL_reg_state.re_state_reglastparen
400#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
401#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
402#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
403#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
404#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
405#define PL_reg_magic PL_reg_state.re_state_reg_magic
406#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
407#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
408#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
409#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
410#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
411#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
412#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
413#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
414#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
415#define PL_regsize PL_reg_state.re_state_regsize
416#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
417#define PL_nrs PL_reg_state.re_state_nrs
418
1ade1aa1
NC
419struct re_save_state {
420 U32 re_state_reg_flags; /* from regexec.c */
421 char *re_state_bostr;
422 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
423 char *re_state_regeol; /* End of input, for $ check. */
424 I32 *re_state_regstartp; /* Pointer to startp array. */
425 I32 *re_state_regendp; /* Ditto for endp. */
426 U32 *re_state_reglastparen; /* Similarly for lastparen. */
427 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
428 char **re_state_reg_start_tmp; /* from regexec.c */
429 U32 re_state_reg_start_tmpl; /* from regexec.c */
430 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
431 bool re_state_reg_match_utf8; /* from regexec.c */
432 MAGIC *re_state_reg_magic; /* from regexec.c */
433 I32 re_state_reg_oldpos; /* from regexec.c */
434 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
435 PMOP *re_state_reg_curpm; /* from regexec.c */
436 char *re_state_reg_oldsaved; /* old saved substr during match */
437 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
438 I32 re_state_reg_maxiter; /* max wait until caching pos */
439 I32 re_state_reg_leftiter; /* wait until caching pos */
440 char *re_state_reg_poscache; /* cache of pos of WHILEM */
441 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 442 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
443 char *re_state_reg_starttry; /* from regexec.c */
444#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 445 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
446#endif
447};
448
449#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
450 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
451/*
452 * Local variables:
453 * c-indentation-style: bsd
454 * c-basic-offset: 4
455 * indent-tabs-mode: t
456 * End:
457 *
458 * ex: set ts=8 sts=4 sw=4 noet:
459 */