This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Verify that the debugger has an array where to store lines before
[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
bbe252da
YO
45
46typedef struct regexp {
47 /* Generic details */
48 const struct regexp_engine* engine; /* what created this regexp? */
49 I32 refcnt; /* Refcount of this regexp */
50
51 /* The original string as passed to the compilation routine */
52 char *precomp; /* pre-compilation regular expression */
53 I32 prelen; /* length of precomp */
54
55 /* Used for generic optimisations by the perl core.
56 All engines are expected to provide this information. */
57 U32 extflags; /* Flags used both externally and internally */
de8c5301
YO
58 I32 minlen; /* mininum possible length of string to match */
59 I32 minlenret; /* mininum possible length of $& */
58e23c8d 60 U32 gofs; /* chars left of pos that we search from */
bbe252da
YO
61 U32 nparens; /* number of capture buffers */
62 HV *paren_names; /* Optional hash of paren names */
63 struct reg_substr_data *substrs; /* substring data about strings that must appear
64 in the final match, used for optimisations */
65
66 /* Data about the last/current match. Used by the core and therefore
67 must be populated by all engines. */
68 char *subbeg; /* saved or original string
69 so \digit works forever. */
70 I32 sublen; /* Length of string pointed by subbeg */
71 I32 *startp; /* Array of offsets from start of string (@-) */
72 I32 *endp; /* Array of offsets from start of string (@+) */
73
74 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
75 U32 lastparen; /* last open paren matched */
76 U32 lastcloseparen; /* last close paren matched */
77
78 /* Perl Regex Engine specific data. Other engines shouldn't need
79 to touch this. Should be refactored out into a different structure
80 and accessed via the *pprivate field. (except intflags) */
81 U32 intflags; /* Internal flags */
82 void *pprivate; /* Data private to the regex engine which
83 created this object. Perl will never mess with
84 this member at all. */
f8fc2ecf
YO
85} regexp;
86
87
88typedef struct regexp_internal {
bbe252da
YO
89 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
90 U32 *offsets; /* offset annotations 20001228 MJD
91 data about mapping the program to the
92 string*/
93 regnode *regstclass; /* Optional startclass as identified or constructed
94 by the optimiser */
95 struct reg_data *data; /* Additional miscellaneous data used by the program.
96 Used to make it easier to clone and free arbitrary
97 data that the regops need. Often the ARG field of
98 a regop is an index into this structure */
c277df42 99 regnode program[1]; /* Unwarranted chumminess with compiler. */
f8fc2ecf 100} regexp_internal;
378cc40b 101
f8fc2ecf
YO
102#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
103#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
104#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
f9f4320a
YO
105
106typedef struct re_scream_pos_data_s
107{
108 char **scream_olds; /* match pos */
109 I32 *scream_pos; /* Internal iterator of scream. */
110} re_scream_pos_data;
111
112typedef struct regexp_engine {
f2f78491
YO
113 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
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);
de8c5301 122 char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval);
f2f78491
YO
123#ifdef USE_ITHREADS
124 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
f9f4320a
YO
125#endif
126} regexp_engine;
127
bbe252da
YO
128/*
129 * Flags stored in regexp->intflags
130 * These are used only internally to the regexp engine
131 */
132#define PREGf_SKIP 0x00000001
133#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
134#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
135#define PREGf_VERBARG_SEEN 0x00000008
136#define PREGf_CUTGROUP_SEEN 0x00000010
137
138
139/* Flags stored in regexp->extflags
140 * These are used by code external to the regexp engine
141 */
142
143/* Anchor and GPOS related stuff */
144#define RXf_ANCH_BOL 0x00000001
145#define RXf_ANCH_MBOL 0x00000002
146#define RXf_ANCH_SBOL 0x00000004
147#define RXf_ANCH_GPOS 0x00000008
148#define RXf_GPOS_SEEN 0x00000010
149#define RXf_GPOS_FLOAT 0x00000020
150/* five bits here */
151#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
152#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
153#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
154/*
155 * 0xF800 of extflags is used by PMf_COMPILETIME
156 * These are the regex equivelent of the PMf_xyz stuff defined
157 * in op.h
158 */
159#define RXf_PMf_LOCALE 0x00000800
160#define RXf_PMf_MULTILINE 0x00001000
161#define RXf_PMf_SINGLELINE 0x00002000
162#define RXf_PMf_FOLD 0x00004000
163#define RXf_PMf_EXTENDED 0x00008000
164#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
165
166/* What we have seen */
167/* one bit here */
168#define RXf_LOOKBEHIND_SEEN 0x00020000
169#define RXf_EVAL_SEEN 0x00040000
170#define RXf_CANY_SEEN 0x00080000
171
172/* Special */
173#define RXf_NOSCAN 0x00100000
174#define RXf_CHECK_ALL 0x00200000
175
176/* UTF8 related */
177#define RXf_UTF8 0x00400000
178#define RXf_MATCH_UTF8 0x00800000
179
180/* Intuit related */
181#define RXf_USE_INTUIT_NOML 0x01000000
182#define RXf_USE_INTUIT_ML 0x02000000
183#define RXf_INTUIT_TAIL 0x04000000
184/* one bit here */
185#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
186
187/* Copy and tainted info */
188#define RXf_COPY_DONE 0x10000000
189#define RXf_TAINTED_SEEN 0x20000000
190/* two bits here */
191
192
193#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
194#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
195#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
196#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751
GS
197#define RX_MATCH_TAINTED_set(prog, t) ((t) \
198 ? RX_MATCH_TAINTED_on(prog) \
199 : RX_MATCH_TAINTED_off(prog))
c277df42 200
bbe252da
YO
201#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
202#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
203#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d
IZ
204#define RX_MATCH_COPIED_set(prog,t) ((t) \
205 ? RX_MATCH_COPIED_on(prog) \
206 : RX_MATCH_COPIED_off(prog))
f9f4320a 207
be8e71aa
YO
208#endif /* PLUGGABLE_RE_EXTENSION */
209
210/* Stuff that needs to be included in the plugable extension goes below here */
211
f8c7b90f 212#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
213#define RX_MATCH_COPY_FREE(rx) \
214 STMT_START {if (rx->saved_copy) { \
215 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
216 } \
217 if (RX_MATCH_COPIED(rx)) { \
218 Safefree(rx->subbeg); \
219 RX_MATCH_COPIED_off(rx); \
220 }} STMT_END
221#else
222#define RX_MATCH_COPY_FREE(rx) \
223 STMT_START {if (RX_MATCH_COPIED(rx)) { \
224 Safefree(rx->subbeg); \
225 RX_MATCH_COPIED_off(rx); \
226 }} STMT_END
227#endif
228
bbe252da
YO
229#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
230#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
231#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f
RGS
232#define RX_MATCH_UTF8_set(prog, t) ((t) \
233 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
234 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
235
cad2e5aa
JH
236#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
237#define REXEC_CHECKED 0x02 /* check_substr already checked. */
238#define REXEC_SCREAM 0x04 /* use scream table. */
239#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 240#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 241
155aba94 242#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 243#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
244
245#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
246#define FBMcf_TAIL_DOLLARM 2
247#define FBMcf_TAIL_Z 4
248#define FBMcf_TAIL_z 8
249#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
250
251#define FBMrf_MULTILINE 1
cad2e5aa 252
5d9a96ca
DM
253/* an accepting state/position*/
254struct _reg_trie_accepted {
255 U8 *endpos;
256 U16 wordnum;
257};
258typedef struct _reg_trie_accepted reg_trie_accepted;
259
3b0527fe
DM
260/* some basic information about the current match that is created by
261 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
262
263typedef struct {
264 regexp *prog;
265 char *bol;
266 char *till;
267 SV *sv;
268 char *ganch;
24b23f37 269 char *cutpoint;
3b0527fe
DM
270} regmatch_info;
271
5d9a96ca
DM
272
273/* structures for holding and saving the state maintained by regmatch() */
274
6bda09f9
YO
275#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
276
5d9a96ca
DM
277typedef I32 CHECKPOINT;
278
a0374537 279typedef struct regmatch_state {
40a82448 280 int resume_state; /* where to jump to on return */
24d3c4a9 281 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
282
283 union {
77cb431f
DM
284
285 /* this is a fake union member that matches the first element
286 * of each member that needs to store positive backtrack
287 * information */
288 struct {
289 struct regmatch_state *prev_yes_state;
290 } yes;
291
e822a8b4 292 struct {
5d458dd8
YO
293 /* this first element must match u.yes */
294 struct regmatch_state *prev_yes_state;
e822a8b4 295 reg_trie_accepted *accept_buff;
166ba7cd 296 U32 accepted; /* how many accepting states we have seen */
7f69552c 297 U16 *jump; /* positive offsets from me */
166ba7cd 298 regnode *B; /* node following the trie */
7f69552c 299 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
300 } trie;
301
302 struct {
77cb431f
DM
303 /* this first element must match u.yes */
304 struct regmatch_state *prev_yes_state;
faec1544
DM
305 struct regmatch_state *prev_eval;
306 struct regmatch_state *prev_curlyx;
aa283a38 307 regexp *prev_rex;
faec1544
DM
308 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
309 flip when transitioning between
310 inner and outer rexen */
aa283a38
DM
311 CHECKPOINT cp; /* remember current savestack indexes */
312 CHECKPOINT lastcp;
40a82448 313 regnode *B; /* the node following us */
6bda09f9 314 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
315 } eval;
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 *prev_curlyx; /* previous cur_curlyx */
321 CHECKPOINT cp; /* remember current savestack index */
322 bool minmod;
a0374537 323 int parenfloor;/* how far back to strip paren data */
c476f425
DM
324 int min; /* the minimal number of A's to match */
325 int max; /* the maximal number of A's to match */
326 regnode *A, *B; /* the nodes corresponding to /A*B/ */
327
328 /* these two are modified by WHILEM */
329 int count; /* how many instances of A we've matched */
330 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
331 } curlyx;
332
333 struct {
c476f425 334 /* this first element must match u.yes */
262b90c4 335 struct regmatch_state *prev_yes_state;
c476f425
DM
336 struct regmatch_state *save_curlyx;
337 CHECKPOINT cp; /* remember current savestack indexes */
338 CHECKPOINT lastcp;
339 char *save_lastloc; /* previous curlyx.lastloc */
340 I32 cache_offset;
341 I32 cache_mask;
e822a8b4
DM
342 } whilem;
343
344 struct {
5d458dd8
YO
345 /* this first element must match u.yes */
346 struct regmatch_state *prev_yes_state;
3b6647e0 347 U32 lastparen;
40a82448
DM
348 regnode *next_branch; /* next branch node */
349 CHECKPOINT cp;
350 } branch;
351
352 struct {
dad79028
DM
353 /* this first element must match u.yes */
354 struct regmatch_state *prev_yes_state;
e822a8b4 355 I32 c1, c2; /* case fold search */
40a82448
DM
356 CHECKPOINT cp;
357 I32 alen; /* length of first-matched A string */
358 I32 count;
0cadcf80 359 bool minmod;
40a82448
DM
360 regnode *A, *B; /* the nodes corresponding to /A*B/ */
361 regnode *me; /* the curlym node */
e822a8b4
DM
362 } curlym;
363
364 struct {
3b6647e0 365 U32 paren;
c255a977 366 CHECKPOINT cp;
e822a8b4 367 I32 c1, c2; /* case fold search */
c255a977
DM
368 char *maxpos; /* highest possible point in string to match */
369 char *oldloc; /* the previous locinput */
e822a8b4 370 int count;
c255a977
DM
371 int min, max; /* {m,n} */
372 regnode *A, *B; /* the nodes corresponding to /A*B/ */
373 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
374
375 struct {
376 /* this first element must match u.yes */
377 struct regmatch_state *prev_yes_state;
378 I32 wanted;
24d3c4a9 379 I32 logical; /* saved copy of 'logical' var */
40a82448 380 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 381 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
382
383 struct {
384 /* this first element must match u.yes */
385 struct regmatch_state *prev_yes_state;
386 struct regmatch_state *prev_mark;
387 SV* mark_name;
388 char *mark_loc;
389 } mark;
d8319b27 390 } u;
5d9a96ca
DM
391} regmatch_state;
392
393/* how many regmatch_state structs to allocate as a single slab.
394 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
395 * pointers, plus 1 for any mythical malloc overhead. */
396
397#define PERL_REGMATCH_SLAB_SLOTS \
398 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
399
400typedef struct regmatch_slab {
401 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
402 struct regmatch_slab *prev, *next;
403} regmatch_slab;
1ade1aa1 404
46ab3289
NC
405#define PL_reg_flags PL_reg_state.re_state_reg_flags
406#define PL_bostr PL_reg_state.re_state_bostr
407#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
408#define PL_regeol PL_reg_state.re_state_regeol
409#define PL_regstartp PL_reg_state.re_state_regstartp
410#define PL_regendp PL_reg_state.re_state_regendp
411#define PL_reglastparen PL_reg_state.re_state_reglastparen
412#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
413#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
414#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
415#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
416#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
417#define PL_reg_magic PL_reg_state.re_state_reg_magic
418#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
419#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
420#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
421#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
422#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
423#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
424#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
425#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
426#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
427#define PL_regsize PL_reg_state.re_state_regsize
428#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
429#define PL_nrs PL_reg_state.re_state_nrs
430
1ade1aa1
NC
431struct re_save_state {
432 U32 re_state_reg_flags; /* from regexec.c */
433 char *re_state_bostr;
434 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
435 char *re_state_regeol; /* End of input, for $ check. */
436 I32 *re_state_regstartp; /* Pointer to startp array. */
437 I32 *re_state_regendp; /* Ditto for endp. */
438 U32 *re_state_reglastparen; /* Similarly for lastparen. */
439 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
440 char **re_state_reg_start_tmp; /* from regexec.c */
441 U32 re_state_reg_start_tmpl; /* from regexec.c */
442 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
443 bool re_state_reg_match_utf8; /* from regexec.c */
444 MAGIC *re_state_reg_magic; /* from regexec.c */
445 I32 re_state_reg_oldpos; /* from regexec.c */
446 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
447 PMOP *re_state_reg_curpm; /* from regexec.c */
448 char *re_state_reg_oldsaved; /* old saved substr during match */
449 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
450 I32 re_state_reg_maxiter; /* max wait until caching pos */
451 I32 re_state_reg_leftiter; /* wait until caching pos */
452 char *re_state_reg_poscache; /* cache of pos of WHILEM */
453 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 454 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
455 char *re_state_reg_starttry; /* from regexec.c */
456#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 457 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
458#endif
459};
460
461#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
462 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
463/*
464 * Local variables:
465 * c-indentation-style: bsd
466 * c-basic-offset: 4
467 * indent-tabs-mode: t
468 * End:
469 *
470 * ex: set ts=8 sts=4 sw=4 noet:
471 */