This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
The new regexp compilation function must be added to the engine structure.
[perl5.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2
JH
3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
4 * 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
YO
17#ifndef PLUGGABLE_RE_EXTENSION
18/* we don't want to include this stuff if we are inside Nicholas'
19 * pluggable regex engine code */
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
c74340f9
YO
35typedef struct regexp_paren_ofs {
36 I32 *startp;
37 I32 *endp;
38} regexp_paren_ofs;
f9f4320a 39
f8c7b90f 40#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da
YO
41#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
42#else
43#define SV_SAVED_COPY
ed252734 44#endif
f8149455
YO
45/* this is ordered such that the most commonly used
46 fields are at the start of the struct */
bbe252da 47typedef struct regexp {
f8149455
YO
48 /* what engine created this regexp? */
49 const struct regexp_engine* engine;
50
51 /* Information about the match that the perl core uses to manage things */
bbe252da 52 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO
53 I32 minlen; /* mininum possible length of string to match */
54 I32 minlenret; /* mininum possible length of $& */
58e23c8d 55 U32 gofs; /* chars left of pos that we search from */
f8149455 56 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 57 in the final match, used for optimisations */
f8149455 58 U32 nparens; /* number of capture buffers */
bbe252da 59
f8149455
YO
60 /* private engine specific data */
61 U32 intflags; /* Engine Specific Internal flags */
62 void *pprivate; /* Data private to the regex engine which
63 created this object. */
64
65 /* Data about the last/current match. These are modified during matching*/
66 U32 lastparen; /* last open paren matched */
67 U32 lastcloseparen; /* last close paren matched */
68 I32 *startp; /* Array of offsets from start of string (@-) */
69 I32 *endp; /* Array of offsets from start of string (@+) */
bbe252da
YO
70 char *subbeg; /* saved or original string
71 so \digit works forever. */
72 I32 sublen; /* Length of string pointed by subbeg */
bbe252da 73 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
f8149455
YO
74
75
76 /* Information about the match that isn't often used */
77 char *precomp; /* pre-compilation regular expression */
78 I32 prelen; /* length of precomp */
79 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
80 HV *paren_names; /* Optional hash of paren names */
81
82 /* Refcount of this regexp */
83 I32 refcnt; /* Refcount of this regexp */
f8fc2ecf
YO
84} regexp;
85
86
87typedef struct regexp_internal {
bbe252da
YO
88 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
89 U32 *offsets; /* offset annotations 20001228 MJD
90 data about mapping the program to the
91 string*/
92 regnode *regstclass; /* Optional startclass as identified or constructed
93 by the optimiser */
94 struct reg_data *data; /* Additional miscellaneous data used by the program.
95 Used to make it easier to clone and free arbitrary
96 data that the regops need. Often the ARG field of
97 a regop is an index into this structure */
c277df42 98 regnode program[1]; /* Unwarranted chumminess with compiler. */
f8fc2ecf 99} regexp_internal;
378cc40b 100
f8fc2ecf
YO
101#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
102#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
103#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
f9f4320a
YO
104
105typedef struct re_scream_pos_data_s
106{
107 char **scream_olds; /* match pos */
108 I32 *scream_pos; /* Internal iterator of scream. */
109} re_scream_pos_data;
110
111typedef struct regexp_engine {
f2f78491 112 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
3c6f95d1 113 regexp* (*compile) (pTHX_ char *exp, char *xend, PMOP *pm);
f2f78491 114 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
f9f4320a
YO
115 char* strbeg, I32 minend, SV* screamer,
116 void* data, U32 flags);
f2f78491
YO
117 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
118 char *strend, U32 flags,
119 struct re_scream_pos_data_s *data);
120 SV* (*checkstr) (pTHX_ regexp *prog);
121 void (*free) (pTHX_ struct regexp* r);
122#ifdef USE_ITHREADS
f8149455 123 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
f9f4320a
YO
124#endif
125} regexp_engine;
126
bbe252da
YO
127/*
128 * Flags stored in regexp->intflags
129 * These are used only internally to the regexp engine
130 */
131#define PREGf_SKIP 0x00000001
132#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
133#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
134#define PREGf_VERBARG_SEEN 0x00000008
135#define PREGf_CUTGROUP_SEEN 0x00000010
136
137
138/* Flags stored in regexp->extflags
139 * These are used by code external to the regexp engine
140 */
141
142/* Anchor and GPOS related stuff */
143#define RXf_ANCH_BOL 0x00000001
144#define RXf_ANCH_MBOL 0x00000002
145#define RXf_ANCH_SBOL 0x00000004
146#define RXf_ANCH_GPOS 0x00000008
147#define RXf_GPOS_SEEN 0x00000010
148#define RXf_GPOS_FLOAT 0x00000020
149/* five bits here */
150#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
151#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
152#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
153/*
154 * 0xF800 of extflags is used by PMf_COMPILETIME
155 * These are the regex equivelent of the PMf_xyz stuff defined
156 * in op.h
157 */
158#define RXf_PMf_LOCALE 0x00000800
159#define RXf_PMf_MULTILINE 0x00001000
160#define RXf_PMf_SINGLELINE 0x00002000
161#define RXf_PMf_FOLD 0x00004000
162#define RXf_PMf_EXTENDED 0x00008000
163#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
164
165/* What we have seen */
166/* one bit here */
167#define RXf_LOOKBEHIND_SEEN 0x00020000
168#define RXf_EVAL_SEEN 0x00040000
169#define RXf_CANY_SEEN 0x00080000
170
171/* Special */
172#define RXf_NOSCAN 0x00100000
173#define RXf_CHECK_ALL 0x00200000
174
175/* UTF8 related */
176#define RXf_UTF8 0x00400000
177#define RXf_MATCH_UTF8 0x00800000
178
179/* Intuit related */
180#define RXf_USE_INTUIT_NOML 0x01000000
181#define RXf_USE_INTUIT_ML 0x02000000
182#define RXf_INTUIT_TAIL 0x04000000
183/* one bit here */
184#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
185
186/* Copy and tainted info */
187#define RXf_COPY_DONE 0x10000000
188#define RXf_TAINTED_SEEN 0x20000000
189/* two bits here */
190
191
192#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
193#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
194#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
195#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751
GS
196#define RX_MATCH_TAINTED_set(prog, t) ((t) \
197 ? RX_MATCH_TAINTED_on(prog) \
198 : RX_MATCH_TAINTED_off(prog))
c277df42 199
bbe252da
YO
200#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
201#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
202#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d
IZ
203#define RX_MATCH_COPIED_set(prog,t) ((t) \
204 ? RX_MATCH_COPIED_on(prog) \
205 : RX_MATCH_COPIED_off(prog))
f9f4320a 206
be8e71aa
YO
207#endif /* PLUGGABLE_RE_EXTENSION */
208
209/* Stuff that needs to be included in the plugable extension goes below here */
210
f8c7b90f 211#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
212#define RX_MATCH_COPY_FREE(rx) \
213 STMT_START {if (rx->saved_copy) { \
214 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
215 } \
216 if (RX_MATCH_COPIED(rx)) { \
217 Safefree(rx->subbeg); \
218 RX_MATCH_COPIED_off(rx); \
219 }} STMT_END
220#else
221#define RX_MATCH_COPY_FREE(rx) \
222 STMT_START {if (RX_MATCH_COPIED(rx)) { \
223 Safefree(rx->subbeg); \
224 RX_MATCH_COPIED_off(rx); \
225 }} STMT_END
226#endif
227
bbe252da
YO
228#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
229#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
230#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
231#define RX_MATCH_UTF8_set(prog, t) ((t) \
232 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
233 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
234
cad2e5aa
JH
235#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
236#define REXEC_CHECKED 0x02 /* check_substr already checked. */
237#define REXEC_SCREAM 0x04 /* use scream table. */
238#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 239#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 240
155aba94 241#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 242#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
243
244#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
245#define FBMcf_TAIL_DOLLARM 2
246#define FBMcf_TAIL_Z 4
247#define FBMcf_TAIL_z 8
248#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
249
250#define FBMrf_MULTILINE 1
cad2e5aa 251
5d9a96ca
DM
252/* an accepting state/position*/
253struct _reg_trie_accepted {
254 U8 *endpos;
255 U16 wordnum;
256};
257typedef struct _reg_trie_accepted reg_trie_accepted;
258
3b0527fe
DM
259/* some basic information about the current match that is created by
260 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
261
262typedef struct {
263 regexp *prog;
264 char *bol;
265 char *till;
266 SV *sv;
267 char *ganch;
24b23f37 268 char *cutpoint;
3b0527fe
DM
269} regmatch_info;
270
5d9a96ca
DM
271
272/* structures for holding and saving the state maintained by regmatch() */
273
6bda09f9
YO
274#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
275
5d9a96ca
DM
276typedef I32 CHECKPOINT;
277
a0374537 278typedef struct regmatch_state {
40a82448 279 int resume_state; /* where to jump to on return */
24d3c4a9 280 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
281
282 union {
77cb431f
DM
283
284 /* this is a fake union member that matches the first element
285 * of each member that needs to store positive backtrack
286 * information */
287 struct {
288 struct regmatch_state *prev_yes_state;
289 } yes;
290
e822a8b4 291 struct {
5d458dd8
YO
292 /* this first element must match u.yes */
293 struct regmatch_state *prev_yes_state;
e822a8b4 294 reg_trie_accepted *accept_buff;
166ba7cd 295 U32 accepted; /* how many accepting states we have seen */
7f69552c 296 U16 *jump; /* positive offsets from me */
166ba7cd 297 regnode *B; /* node following the trie */
7f69552c 298 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
299 } trie;
300
301 struct {
77cb431f
DM
302 /* this first element must match u.yes */
303 struct regmatch_state *prev_yes_state;
faec1544
DM
304 struct regmatch_state *prev_eval;
305 struct regmatch_state *prev_curlyx;
aa283a38 306 regexp *prev_rex;
faec1544
DM
307 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
308 flip when transitioning between
309 inner and outer rexen */
aa283a38
DM
310 CHECKPOINT cp; /* remember current savestack indexes */
311 CHECKPOINT lastcp;
40a82448 312 regnode *B; /* the node following us */
6bda09f9 313 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
314 } eval;
315
316 struct {
c476f425 317 /* this first element must match u.yes */
262b90c4 318 struct regmatch_state *prev_yes_state;
c476f425
DM
319 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
320 CHECKPOINT cp; /* remember current savestack index */
321 bool minmod;
a0374537 322 int parenfloor;/* how far back to strip paren data */
c476f425
DM
323 int min; /* the minimal number of A's to match */
324 int max; /* the maximal number of A's to match */
325 regnode *A, *B; /* the nodes corresponding to /A*B/ */
326
327 /* these two are modified by WHILEM */
328 int count; /* how many instances of A we've matched */
329 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
330 } curlyx;
331
332 struct {
c476f425 333 /* this first element must match u.yes */
262b90c4 334 struct regmatch_state *prev_yes_state;
c476f425
DM
335 struct regmatch_state *save_curlyx;
336 CHECKPOINT cp; /* remember current savestack indexes */
337 CHECKPOINT lastcp;
338 char *save_lastloc; /* previous curlyx.lastloc */
339 I32 cache_offset;
340 I32 cache_mask;
e822a8b4
DM
341 } whilem;
342
343 struct {
5d458dd8
YO
344 /* this first element must match u.yes */
345 struct regmatch_state *prev_yes_state;
3b6647e0 346 U32 lastparen;
40a82448
DM
347 regnode *next_branch; /* next branch node */
348 CHECKPOINT cp;
349 } branch;
350
351 struct {
dad79028
DM
352 /* this first element must match u.yes */
353 struct regmatch_state *prev_yes_state;
e822a8b4 354 I32 c1, c2; /* case fold search */
40a82448
DM
355 CHECKPOINT cp;
356 I32 alen; /* length of first-matched A string */
357 I32 count;
0cadcf80 358 bool minmod;
40a82448
DM
359 regnode *A, *B; /* the nodes corresponding to /A*B/ */
360 regnode *me; /* the curlym node */
e822a8b4
DM
361 } curlym;
362
363 struct {
3b6647e0 364 U32 paren;
c255a977 365 CHECKPOINT cp;
e822a8b4 366 I32 c1, c2; /* case fold search */
c255a977
DM
367 char *maxpos; /* highest possible point in string to match */
368 char *oldloc; /* the previous locinput */
e822a8b4 369 int count;
c255a977
DM
370 int min, max; /* {m,n} */
371 regnode *A, *B; /* the nodes corresponding to /A*B/ */
372 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
373
374 struct {
375 /* this first element must match u.yes */
376 struct regmatch_state *prev_yes_state;
377 I32 wanted;
24d3c4a9 378 I32 logical; /* saved copy of 'logical' var */
40a82448 379 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 380 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
381
382 struct {
383 /* this first element must match u.yes */
384 struct regmatch_state *prev_yes_state;
385 struct regmatch_state *prev_mark;
386 SV* mark_name;
387 char *mark_loc;
388 } mark;
d8319b27 389 } u;
5d9a96ca
DM
390} regmatch_state;
391
392/* how many regmatch_state structs to allocate as a single slab.
393 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
394 * pointers, plus 1 for any mythical malloc overhead. */
395
396#define PERL_REGMATCH_SLAB_SLOTS \
397 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
398
399typedef struct regmatch_slab {
400 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
401 struct regmatch_slab *prev, *next;
402} regmatch_slab;
1ade1aa1 403
46ab3289
NC
404#define PL_reg_flags PL_reg_state.re_state_reg_flags
405#define PL_bostr PL_reg_state.re_state_bostr
406#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
407#define PL_regeol PL_reg_state.re_state_regeol
408#define PL_regstartp PL_reg_state.re_state_regstartp
409#define PL_regendp PL_reg_state.re_state_regendp
410#define PL_reglastparen PL_reg_state.re_state_reglastparen
411#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
412#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
413#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
414#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
415#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
416#define PL_reg_magic PL_reg_state.re_state_reg_magic
417#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
418#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
419#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
420#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
421#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
422#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
423#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
424#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
425#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
426#define PL_regsize PL_reg_state.re_state_regsize
427#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
428#define PL_nrs PL_reg_state.re_state_nrs
429
1ade1aa1
NC
430struct re_save_state {
431 U32 re_state_reg_flags; /* from regexec.c */
432 char *re_state_bostr;
433 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
434 char *re_state_regeol; /* End of input, for $ check. */
435 I32 *re_state_regstartp; /* Pointer to startp array. */
436 I32 *re_state_regendp; /* Ditto for endp. */
437 U32 *re_state_reglastparen; /* Similarly for lastparen. */
438 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
439 char **re_state_reg_start_tmp; /* from regexec.c */
440 U32 re_state_reg_start_tmpl; /* from regexec.c */
441 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
442 bool re_state_reg_match_utf8; /* from regexec.c */
443 MAGIC *re_state_reg_magic; /* from regexec.c */
444 I32 re_state_reg_oldpos; /* from regexec.c */
445 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
446 PMOP *re_state_reg_curpm; /* from regexec.c */
447 char *re_state_reg_oldsaved; /* old saved substr during match */
448 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
449 I32 re_state_reg_maxiter; /* max wait until caching pos */
450 I32 re_state_reg_leftiter; /* wait until caching pos */
451 char *re_state_reg_poscache; /* cache of pos of WHILEM */
452 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 453 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
454 char *re_state_reg_starttry; /* from regexec.c */
455#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 456 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
457#endif
458};
459
460#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
461 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
462/*
463 * Local variables:
464 * c-indentation-style: bsd
465 * c-basic-offset: 4
466 * indent-tabs-mode: t
467 * End:
468 *
469 * ex: set ts=8 sts=4 sw=4 noet:
470 */