This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Remove duplicate declaration
[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;
c74340f9
YO
34typedef struct regexp_paren_ofs {
35 I32 *startp;
36 I32 *endp;
37} regexp_paren_ofs;
f9f4320a 38
378cc40b 39typedef struct regexp {
c74340f9 40 I32 *startp;
cf93c79d 41 I32 *endp;
c74340f9 42 regexp_paren_ofs *swap;
c277df42 43 regnode *regstclass;
2779dcf1 44 struct reg_substr_data *substrs;
cf93c79d 45 char *precomp; /* pre-compilation regular expression */
c277df42 46 struct reg_data *data; /* Additional data. */
cf93c79d
IZ
47 char *subbeg; /* saved or original string
48 so \digit works forever. */
f8c7b90f 49#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
50 SV *saved_copy; /* If non-NULL, SV which is COW from original */
51#endif
fac92740 52 U32 *offsets; /* offset annotations 20001228 MJD */
cf93c79d
IZ
53 I32 sublen; /* Length of string pointed by subbeg */
54 I32 refcnt;
de8c5301
YO
55 I32 minlen; /* mininum possible length of string to match */
56 I32 minlenret; /* mininum possible length of $& */
cf93c79d
IZ
57 I32 prelen; /* length of precomp */
58 U32 nparens; /* number of parentheses */
59 U32 lastparen; /* last paren matched */
a01268b5 60 U32 lastcloseparen; /* last paren matched */
cf93c79d
IZ
61 U32 reganch; /* Internal use only +
62 Tainted information used by regexec? */
81714fb9
YO
63 HV *paren_names; /* Paren names */
64 const struct regexp_engine* engine;
c277df42 65 regnode program[1]; /* Unwarranted chumminess with compiler. */
378cc40b
LW
66} regexp;
67
f9f4320a
YO
68
69typedef struct re_scream_pos_data_s
70{
71 char **scream_olds; /* match pos */
72 I32 *scream_pos; /* Internal iterator of scream. */
73} re_scream_pos_data;
74
75typedef struct regexp_engine {
f2f78491
YO
76 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
77 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
f9f4320a
YO
78 char* strbeg, I32 minend, SV* screamer,
79 void* data, U32 flags);
f2f78491
YO
80 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
81 char *strend, U32 flags,
82 struct re_scream_pos_data_s *data);
83 SV* (*checkstr) (pTHX_ regexp *prog);
84 void (*free) (pTHX_ struct regexp* r);
de8c5301 85 char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval);
f2f78491
YO
86#ifdef USE_ITHREADS
87 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
f9f4320a
YO
88#endif
89} regexp_engine;
90
cad2e5aa
JH
91#define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
92#define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
f2278c82
YO
93#define ROPT_ANCH_BOL 0x00000001
94#define ROPT_ANCH_MBOL 0x00000002
95#define ROPT_ANCH_SBOL 0x00000004
96#define ROPT_ANCH_GPOS 0x00000008
97#define ROPT_SKIP 0x00000010
98#define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */
99#define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */
100#define ROPT_GPOS_SEEN 0x00000080
101#define ROPT_CHECK_ALL 0x00000100
102#define ROPT_LOOKBEHIND_SEEN 0x00000200
103#define ROPT_EVAL_SEEN 0x00000400
104#define ROPT_CANY_SEEN 0x00000800
cce850e4 105#define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
f9f4320a 106#define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS)
a0ed51b3 107
5d458dd8 108/* 0xF800 of reganch is used by PMf_COMPILETIME */
c277df42 109
f2278c82
YO
110#define ROPT_UTF8 0x00010000
111#define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */
112#define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */
113#define ROPT_TAINTED_SEEN 0x00080000
a30b2f1f 114#define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
5d458dd8
YO
115#define ROPT_VERBARG_SEEN 0x20000000
116#define ROPT_CUTGROUP_SEEN 0x40000000
a0ed51b3 117
f2278c82
YO
118#define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
119#define RE_USE_INTUIT_ML 0x00200000
120#define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
be8e71aa 121#define REINT_AUTORITATIVE_ML 0x00800000
f2278c82
YO
122#define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
123#define REINT_ONCE_ML 0x02000000
124#define RE_INTUIT_ONECHAR 0x04000000
125#define RE_INTUIT_TAIL 0x08000000
126
cad2e5aa
JH
127
128#define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
129#define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
130#define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
131
5d458dd8 132#define RX_HAS_CUTGROUP(prog) ((prog)->reganch & ROPT_CUTGROUP_SEEN)
c277df42 133#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
72311751
GS
134#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
135#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
136#define RX_MATCH_TAINTED_set(prog, t) ((t) \
137 ? RX_MATCH_TAINTED_on(prog) \
138 : RX_MATCH_TAINTED_off(prog))
c277df42 139
cf93c79d
IZ
140#define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
141#define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
142#define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
143#define RX_MATCH_COPIED_set(prog,t) ((t) \
144 ? RX_MATCH_COPIED_on(prog) \
145 : RX_MATCH_COPIED_off(prog))
f9f4320a 146
be8e71aa
YO
147#endif /* PLUGGABLE_RE_EXTENSION */
148
149/* Stuff that needs to be included in the plugable extension goes below here */
150
f8c7b90f 151#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
152#define RX_MATCH_COPY_FREE(rx) \
153 STMT_START {if (rx->saved_copy) { \
154 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
155 } \
156 if (RX_MATCH_COPIED(rx)) { \
157 Safefree(rx->subbeg); \
158 RX_MATCH_COPIED_off(rx); \
159 }} STMT_END
160#else
161#define RX_MATCH_COPY_FREE(rx) \
162 STMT_START {if (RX_MATCH_COPIED(rx)) { \
163 Safefree(rx->subbeg); \
164 RX_MATCH_COPIED_off(rx); \
165 }} STMT_END
166#endif
167
a30b2f1f
RGS
168#define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
169#define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
170#define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
171#define RX_MATCH_UTF8_set(prog, t) ((t) \
172 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
173 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
174
cad2e5aa
JH
175#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
176#define REXEC_CHECKED 0x02 /* check_substr already checked. */
177#define REXEC_SCREAM 0x04 /* use scream table. */
178#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 179#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 180
155aba94 181#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 182#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
183
184#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
185#define FBMcf_TAIL_DOLLARM 2
186#define FBMcf_TAIL_Z 4
187#define FBMcf_TAIL_z 8
188#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
189
190#define FBMrf_MULTILINE 1
cad2e5aa 191
5d9a96ca
DM
192/* an accepting state/position*/
193struct _reg_trie_accepted {
194 U8 *endpos;
195 U16 wordnum;
196};
197typedef struct _reg_trie_accepted reg_trie_accepted;
198
3b0527fe
DM
199/* some basic information about the current match that is created by
200 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
201
202typedef struct {
203 regexp *prog;
204 char *bol;
205 char *till;
206 SV *sv;
207 char *ganch;
24b23f37 208 char *cutpoint;
3b0527fe
DM
209} regmatch_info;
210
5d9a96ca
DM
211
212/* structures for holding and saving the state maintained by regmatch() */
213
6bda09f9
YO
214#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
215
5d9a96ca
DM
216typedef I32 CHECKPOINT;
217
a0374537 218typedef struct regmatch_state {
40a82448 219 int resume_state; /* where to jump to on return */
24d3c4a9 220 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
221
222 union {
77cb431f
DM
223
224 /* this is a fake union member that matches the first element
225 * of each member that needs to store positive backtrack
226 * information */
227 struct {
228 struct regmatch_state *prev_yes_state;
229 } yes;
230
e822a8b4 231 struct {
5d458dd8
YO
232 /* this first element must match u.yes */
233 struct regmatch_state *prev_yes_state;
e822a8b4 234 reg_trie_accepted *accept_buff;
166ba7cd 235 U32 accepted; /* how many accepting states we have seen */
7f69552c 236 U16 *jump; /* positive offsets from me */
166ba7cd 237 regnode *B; /* node following the trie */
7f69552c 238 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
239 } trie;
240
241 struct {
77cb431f
DM
242 /* this first element must match u.yes */
243 struct regmatch_state *prev_yes_state;
faec1544
DM
244 struct regmatch_state *prev_eval;
245 struct regmatch_state *prev_curlyx;
aa283a38 246 regexp *prev_rex;
faec1544
DM
247 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
248 flip when transitioning between
249 inner and outer rexen */
aa283a38
DM
250 CHECKPOINT cp; /* remember current savestack indexes */
251 CHECKPOINT lastcp;
40a82448 252 regnode *B; /* the node following us */
6bda09f9 253 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
254 } eval;
255
256 struct {
c476f425 257 /* this first element must match u.yes */
262b90c4 258 struct regmatch_state *prev_yes_state;
c476f425
DM
259 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
260 CHECKPOINT cp; /* remember current savestack index */
261 bool minmod;
a0374537 262 int parenfloor;/* how far back to strip paren data */
c476f425
DM
263 int min; /* the minimal number of A's to match */
264 int max; /* the maximal number of A's to match */
265 regnode *A, *B; /* the nodes corresponding to /A*B/ */
266
267 /* these two are modified by WHILEM */
268 int count; /* how many instances of A we've matched */
269 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
270 } curlyx;
271
272 struct {
c476f425 273 /* this first element must match u.yes */
262b90c4 274 struct regmatch_state *prev_yes_state;
c476f425
DM
275 struct regmatch_state *save_curlyx;
276 CHECKPOINT cp; /* remember current savestack indexes */
277 CHECKPOINT lastcp;
278 char *save_lastloc; /* previous curlyx.lastloc */
279 I32 cache_offset;
280 I32 cache_mask;
e822a8b4
DM
281 } whilem;
282
283 struct {
5d458dd8
YO
284 /* this first element must match u.yes */
285 struct regmatch_state *prev_yes_state;
3b6647e0 286 U32 lastparen;
40a82448
DM
287 regnode *next_branch; /* next branch node */
288 CHECKPOINT cp;
289 } branch;
290
291 struct {
dad79028
DM
292 /* this first element must match u.yes */
293 struct regmatch_state *prev_yes_state;
e822a8b4 294 I32 c1, c2; /* case fold search */
40a82448
DM
295 CHECKPOINT cp;
296 I32 alen; /* length of first-matched A string */
297 I32 count;
0cadcf80 298 bool minmod;
40a82448
DM
299 regnode *A, *B; /* the nodes corresponding to /A*B/ */
300 regnode *me; /* the curlym node */
e822a8b4
DM
301 } curlym;
302
303 struct {
3b6647e0 304 U32 paren;
c255a977 305 CHECKPOINT cp;
e822a8b4 306 I32 c1, c2; /* case fold search */
c255a977
DM
307 char *maxpos; /* highest possible point in string to match */
308 char *oldloc; /* the previous locinput */
e822a8b4 309 int count;
c255a977
DM
310 int min, max; /* {m,n} */
311 regnode *A, *B; /* the nodes corresponding to /A*B/ */
312 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
313
314 struct {
315 /* this first element must match u.yes */
316 struct regmatch_state *prev_yes_state;
317 I32 wanted;
24d3c4a9 318 I32 logical; /* saved copy of 'logical' var */
40a82448 319 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 320 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
321
322 struct {
323 /* this first element must match u.yes */
324 struct regmatch_state *prev_yes_state;
325 struct regmatch_state *prev_mark;
326 SV* mark_name;
327 char *mark_loc;
328 } mark;
d8319b27 329 } u;
5d9a96ca
DM
330} regmatch_state;
331
332/* how many regmatch_state structs to allocate as a single slab.
333 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
334 * pointers, plus 1 for any mythical malloc overhead. */
335
336#define PERL_REGMATCH_SLAB_SLOTS \
337 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
338
339typedef struct regmatch_slab {
340 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
341 struct regmatch_slab *prev, *next;
342} regmatch_slab;
1ade1aa1 343
46ab3289
NC
344#define PL_reg_flags PL_reg_state.re_state_reg_flags
345#define PL_bostr PL_reg_state.re_state_bostr
346#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
347#define PL_regeol PL_reg_state.re_state_regeol
348#define PL_regstartp PL_reg_state.re_state_regstartp
349#define PL_regendp PL_reg_state.re_state_regendp
350#define PL_reglastparen PL_reg_state.re_state_reglastparen
351#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
352#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
353#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
354#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
355#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
356#define PL_reg_magic PL_reg_state.re_state_reg_magic
357#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
358#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
359#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
360#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
361#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
362#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
363#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
364#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
365#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
366#define PL_regsize PL_reg_state.re_state_regsize
367#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
368#define PL_nrs PL_reg_state.re_state_nrs
369
1ade1aa1
NC
370struct re_save_state {
371 U32 re_state_reg_flags; /* from regexec.c */
372 char *re_state_bostr;
373 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
374 char *re_state_regeol; /* End of input, for $ check. */
375 I32 *re_state_regstartp; /* Pointer to startp array. */
376 I32 *re_state_regendp; /* Ditto for endp. */
377 U32 *re_state_reglastparen; /* Similarly for lastparen. */
378 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
379 char **re_state_reg_start_tmp; /* from regexec.c */
380 U32 re_state_reg_start_tmpl; /* from regexec.c */
381 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
382 bool re_state_reg_match_utf8; /* from regexec.c */
383 MAGIC *re_state_reg_magic; /* from regexec.c */
384 I32 re_state_reg_oldpos; /* from regexec.c */
385 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
386 PMOP *re_state_reg_curpm; /* from regexec.c */
387 char *re_state_reg_oldsaved; /* old saved substr during match */
388 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
389 I32 re_state_reg_maxiter; /* max wait until caching pos */
390 I32 re_state_reg_leftiter; /* wait until caching pos */
391 char *re_state_reg_poscache; /* cache of pos of WHILEM */
392 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 393 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
394 char *re_state_reg_starttry; /* from regexec.c */
395#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 396 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
397#endif
398};
399
400#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
401 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
402/*
403 * Local variables:
404 * c-indentation-style: bsd
405 * c-basic-offset: 4
406 * indent-tabs-mode: t
407 * End:
408 *
409 * ex: set ts=8 sts=4 sw=4 noet:
410 */