This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Further tweaks to make it easier to create regexp engine plug ins.
[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 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
119 */
120
121/* Anchor and GPOS related stuff */
122#define RXf_ANCH_BOL 0x00000001
123#define RXf_ANCH_MBOL 0x00000002
124#define RXf_ANCH_SBOL 0x00000004
125#define RXf_ANCH_GPOS 0x00000008
126#define RXf_GPOS_SEEN 0x00000010
127#define RXf_GPOS_FLOAT 0x00000020
128/* five bits here */
129#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
130#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
131#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
132/*
133 * 0xF800 of extflags is used by PMf_COMPILETIME
134 * These are the regex equivelent of the PMf_xyz stuff defined
135 * in op.h
136 */
137#define RXf_PMf_LOCALE 0x00000800
138#define RXf_PMf_MULTILINE 0x00001000
139#define RXf_PMf_SINGLELINE 0x00002000
140#define RXf_PMf_FOLD 0x00004000
141#define RXf_PMf_EXTENDED 0x00008000
142#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
143
144/* What we have seen */
145/* one bit here */
146#define RXf_LOOKBEHIND_SEEN 0x00020000
147#define RXf_EVAL_SEEN 0x00040000
148#define RXf_CANY_SEEN 0x00080000
149
150/* Special */
151#define RXf_NOSCAN 0x00100000
152#define RXf_CHECK_ALL 0x00200000
153
154/* UTF8 related */
155#define RXf_UTF8 0x00400000
156#define RXf_MATCH_UTF8 0x00800000
157
158/* Intuit related */
159#define RXf_USE_INTUIT_NOML 0x01000000
160#define RXf_USE_INTUIT_ML 0x02000000
161#define RXf_INTUIT_TAIL 0x04000000
162/* one bit here */
163#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
164
165/* Copy and tainted info */
166#define RXf_COPY_DONE 0x10000000
167#define RXf_TAINTED_SEEN 0x20000000
168/* two bits here */
169
170
171#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
172#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
173#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
174#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751
GS
175#define RX_MATCH_TAINTED_set(prog, t) ((t) \
176 ? RX_MATCH_TAINTED_on(prog) \
177 : RX_MATCH_TAINTED_off(prog))
c277df42 178
bbe252da
YO
179#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
180#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
181#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d
IZ
182#define RX_MATCH_COPIED_set(prog,t) ((t) \
183 ? RX_MATCH_COPIED_on(prog) \
184 : RX_MATCH_COPIED_off(prog))
f9f4320a 185
be8e71aa
YO
186#endif /* PLUGGABLE_RE_EXTENSION */
187
188/* Stuff that needs to be included in the plugable extension goes below here */
189
f8c7b90f 190#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
191#define RX_MATCH_COPY_FREE(rx) \
192 STMT_START {if (rx->saved_copy) { \
193 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
194 } \
195 if (RX_MATCH_COPIED(rx)) { \
196 Safefree(rx->subbeg); \
197 RX_MATCH_COPIED_off(rx); \
198 }} STMT_END
199#else
200#define RX_MATCH_COPY_FREE(rx) \
201 STMT_START {if (RX_MATCH_COPIED(rx)) { \
202 Safefree(rx->subbeg); \
203 RX_MATCH_COPIED_off(rx); \
204 }} STMT_END
205#endif
206
bbe252da
YO
207#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
208#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
209#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
210#define RX_MATCH_UTF8_set(prog, t) ((t) \
211 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
212 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
213
cad2e5aa
JH
214#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
215#define REXEC_CHECKED 0x02 /* check_substr already checked. */
216#define REXEC_SCREAM 0x04 /* use scream table. */
217#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 218#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 219
155aba94 220#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 221#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
222
223#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
224#define FBMcf_TAIL_DOLLARM 2
225#define FBMcf_TAIL_Z 4
226#define FBMcf_TAIL_z 8
227#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
228
229#define FBMrf_MULTILINE 1
cad2e5aa 230
5d9a96ca
DM
231/* an accepting state/position*/
232struct _reg_trie_accepted {
233 U8 *endpos;
234 U16 wordnum;
235};
236typedef struct _reg_trie_accepted reg_trie_accepted;
237
3b0527fe
DM
238/* some basic information about the current match that is created by
239 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
240
241typedef struct {
242 regexp *prog;
243 char *bol;
244 char *till;
245 SV *sv;
246 char *ganch;
24b23f37 247 char *cutpoint;
3b0527fe
DM
248} regmatch_info;
249
5d9a96ca
DM
250
251/* structures for holding and saving the state maintained by regmatch() */
252
6bda09f9
YO
253#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
254
5d9a96ca
DM
255typedef I32 CHECKPOINT;
256
a0374537 257typedef struct regmatch_state {
40a82448 258 int resume_state; /* where to jump to on return */
24d3c4a9 259 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
260
261 union {
77cb431f
DM
262
263 /* this is a fake union member that matches the first element
264 * of each member that needs to store positive backtrack
265 * information */
266 struct {
267 struct regmatch_state *prev_yes_state;
268 } yes;
269
e822a8b4 270 struct {
5d458dd8
YO
271 /* this first element must match u.yes */
272 struct regmatch_state *prev_yes_state;
e822a8b4 273 reg_trie_accepted *accept_buff;
166ba7cd 274 U32 accepted; /* how many accepting states we have seen */
7f69552c 275 U16 *jump; /* positive offsets from me */
166ba7cd 276 regnode *B; /* node following the trie */
7f69552c 277 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
278 } trie;
279
280 struct {
77cb431f
DM
281 /* this first element must match u.yes */
282 struct regmatch_state *prev_yes_state;
faec1544
DM
283 struct regmatch_state *prev_eval;
284 struct regmatch_state *prev_curlyx;
aa283a38 285 regexp *prev_rex;
faec1544
DM
286 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
287 flip when transitioning between
288 inner and outer rexen */
aa283a38
DM
289 CHECKPOINT cp; /* remember current savestack indexes */
290 CHECKPOINT lastcp;
40a82448 291 regnode *B; /* the node following us */
6bda09f9 292 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
293 } eval;
294
295 struct {
c476f425 296 /* this first element must match u.yes */
262b90c4 297 struct regmatch_state *prev_yes_state;
c476f425
DM
298 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
299 CHECKPOINT cp; /* remember current savestack index */
300 bool minmod;
a0374537 301 int parenfloor;/* how far back to strip paren data */
c476f425
DM
302 int min; /* the minimal number of A's to match */
303 int max; /* the maximal number of A's to match */
304 regnode *A, *B; /* the nodes corresponding to /A*B/ */
305
306 /* these two are modified by WHILEM */
307 int count; /* how many instances of A we've matched */
308 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
309 } curlyx;
310
311 struct {
c476f425 312 /* this first element must match u.yes */
262b90c4 313 struct regmatch_state *prev_yes_state;
c476f425
DM
314 struct regmatch_state *save_curlyx;
315 CHECKPOINT cp; /* remember current savestack indexes */
316 CHECKPOINT lastcp;
317 char *save_lastloc; /* previous curlyx.lastloc */
318 I32 cache_offset;
319 I32 cache_mask;
e822a8b4
DM
320 } whilem;
321
322 struct {
5d458dd8
YO
323 /* this first element must match u.yes */
324 struct regmatch_state *prev_yes_state;
3b6647e0 325 U32 lastparen;
40a82448
DM
326 regnode *next_branch; /* next branch node */
327 CHECKPOINT cp;
328 } branch;
329
330 struct {
dad79028
DM
331 /* this first element must match u.yes */
332 struct regmatch_state *prev_yes_state;
e822a8b4 333 I32 c1, c2; /* case fold search */
40a82448
DM
334 CHECKPOINT cp;
335 I32 alen; /* length of first-matched A string */
336 I32 count;
0cadcf80 337 bool minmod;
40a82448
DM
338 regnode *A, *B; /* the nodes corresponding to /A*B/ */
339 regnode *me; /* the curlym node */
e822a8b4
DM
340 } curlym;
341
342 struct {
3b6647e0 343 U32 paren;
c255a977 344 CHECKPOINT cp;
e822a8b4 345 I32 c1, c2; /* case fold search */
c255a977
DM
346 char *maxpos; /* highest possible point in string to match */
347 char *oldloc; /* the previous locinput */
e822a8b4 348 int count;
c255a977
DM
349 int min, max; /* {m,n} */
350 regnode *A, *B; /* the nodes corresponding to /A*B/ */
351 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
352
353 struct {
354 /* this first element must match u.yes */
355 struct regmatch_state *prev_yes_state;
356 I32 wanted;
24d3c4a9 357 I32 logical; /* saved copy of 'logical' var */
40a82448 358 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 359 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
360
361 struct {
362 /* this first element must match u.yes */
363 struct regmatch_state *prev_yes_state;
364 struct regmatch_state *prev_mark;
365 SV* mark_name;
366 char *mark_loc;
367 } mark;
d8319b27 368 } u;
5d9a96ca
DM
369} regmatch_state;
370
371/* how many regmatch_state structs to allocate as a single slab.
372 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
373 * pointers, plus 1 for any mythical malloc overhead. */
374
375#define PERL_REGMATCH_SLAB_SLOTS \
376 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
377
378typedef struct regmatch_slab {
379 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
380 struct regmatch_slab *prev, *next;
381} regmatch_slab;
1ade1aa1 382
46ab3289
NC
383#define PL_reg_flags PL_reg_state.re_state_reg_flags
384#define PL_bostr PL_reg_state.re_state_bostr
385#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
386#define PL_regeol PL_reg_state.re_state_regeol
387#define PL_regstartp PL_reg_state.re_state_regstartp
388#define PL_regendp PL_reg_state.re_state_regendp
389#define PL_reglastparen PL_reg_state.re_state_reglastparen
390#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
391#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
392#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
393#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
394#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
395#define PL_reg_magic PL_reg_state.re_state_reg_magic
396#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
397#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
398#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
399#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
400#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
401#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
402#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
403#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
404#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
405#define PL_regsize PL_reg_state.re_state_regsize
406#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
407#define PL_nrs PL_reg_state.re_state_nrs
408
1ade1aa1
NC
409struct re_save_state {
410 U32 re_state_reg_flags; /* from regexec.c */
411 char *re_state_bostr;
412 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
413 char *re_state_regeol; /* End of input, for $ check. */
414 I32 *re_state_regstartp; /* Pointer to startp array. */
415 I32 *re_state_regendp; /* Ditto for endp. */
416 U32 *re_state_reglastparen; /* Similarly for lastparen. */
417 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
418 char **re_state_reg_start_tmp; /* from regexec.c */
419 U32 re_state_reg_start_tmpl; /* from regexec.c */
420 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
421 bool re_state_reg_match_utf8; /* from regexec.c */
422 MAGIC *re_state_reg_magic; /* from regexec.c */
423 I32 re_state_reg_oldpos; /* from regexec.c */
424 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
425 PMOP *re_state_reg_curpm; /* from regexec.c */
426 char *re_state_reg_oldsaved; /* old saved substr during match */
427 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
428 I32 re_state_reg_maxiter; /* max wait until caching pos */
429 I32 re_state_reg_leftiter; /* wait until caching pos */
430 char *re_state_reg_poscache; /* cache of pos of WHILEM */
431 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 432 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
433 char *re_state_reg_starttry; /* from regexec.c */
434#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 435 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
436#endif
437};
438
439#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
440 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
441/*
442 * Local variables:
443 * c-indentation-style: bsd
444 * c-basic-offset: 4
445 * indent-tabs-mode: t
446 * End:
447 *
448 * ex: set ts=8 sts=4 sw=4 noet:
449 */