This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Fwd: Memory leak with s/// and hashes
[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
YO
33struct regexp_engine;
34
378cc40b 35typedef struct regexp {
cf93c79d
IZ
36 I32 *startp;
37 I32 *endp;
c277df42 38 regnode *regstclass;
2779dcf1 39 struct reg_substr_data *substrs;
cf93c79d 40 char *precomp; /* pre-compilation regular expression */
c277df42 41 struct reg_data *data; /* Additional data. */
cf93c79d
IZ
42 char *subbeg; /* saved or original string
43 so \digit works forever. */
f8c7b90f 44#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
45 SV *saved_copy; /* If non-NULL, SV which is COW from original */
46#endif
fac92740 47 U32 *offsets; /* offset annotations 20001228 MJD */
cf93c79d
IZ
48 I32 sublen; /* Length of string pointed by subbeg */
49 I32 refcnt;
de8c5301
YO
50 I32 minlen; /* mininum possible length of string to match */
51 I32 minlenret; /* mininum possible length of $& */
cf93c79d
IZ
52 I32 prelen; /* length of precomp */
53 U32 nparens; /* number of parentheses */
54 U32 lastparen; /* last paren matched */
a01268b5 55 U32 lastcloseparen; /* last paren matched */
cf93c79d
IZ
56 U32 reganch; /* Internal use only +
57 Tainted information used by regexec? */
81714fb9
YO
58 HV *paren_names; /* Paren names */
59 const struct regexp_engine* engine;
c277df42 60 regnode program[1]; /* Unwarranted chumminess with compiler. */
378cc40b
LW
61} regexp;
62
f9f4320a
YO
63
64typedef struct re_scream_pos_data_s
65{
66 char **scream_olds; /* match pos */
67 I32 *scream_pos; /* Internal iterator of scream. */
68} re_scream_pos_data;
69
70typedef struct regexp_engine {
f2f78491
YO
71 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
72 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
f9f4320a
YO
73 char* strbeg, I32 minend, SV* screamer,
74 void* data, U32 flags);
f2f78491
YO
75 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
76 char *strend, U32 flags,
77 struct re_scream_pos_data_s *data);
78 SV* (*checkstr) (pTHX_ regexp *prog);
79 void (*free) (pTHX_ struct regexp* r);
de8c5301 80 char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval);
f2f78491
YO
81#ifdef USE_ITHREADS
82 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
f9f4320a
YO
83#endif
84} regexp_engine;
85
cad2e5aa
JH
86#define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
87#define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
f2278c82
YO
88#define ROPT_ANCH_BOL 0x00000001
89#define ROPT_ANCH_MBOL 0x00000002
90#define ROPT_ANCH_SBOL 0x00000004
91#define ROPT_ANCH_GPOS 0x00000008
92#define ROPT_SKIP 0x00000010
93#define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */
94#define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */
95#define ROPT_GPOS_SEEN 0x00000080
96#define ROPT_CHECK_ALL 0x00000100
97#define ROPT_LOOKBEHIND_SEEN 0x00000200
98#define ROPT_EVAL_SEEN 0x00000400
99#define ROPT_CANY_SEEN 0x00000800
cce850e4 100#define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
f9f4320a 101#define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS)
a0ed51b3 102
5d458dd8 103/* 0xF800 of reganch is used by PMf_COMPILETIME */
c277df42 104
f2278c82
YO
105#define ROPT_UTF8 0x00010000
106#define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */
107#define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */
108#define ROPT_TAINTED_SEEN 0x00080000
a30b2f1f 109#define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
5d458dd8
YO
110#define ROPT_VERBARG_SEEN 0x20000000
111#define ROPT_CUTGROUP_SEEN 0x40000000
a0ed51b3 112
f2278c82
YO
113#define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
114#define RE_USE_INTUIT_ML 0x00200000
115#define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
be8e71aa 116#define REINT_AUTORITATIVE_ML 0x00800000
f2278c82
YO
117#define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
118#define REINT_ONCE_ML 0x02000000
119#define RE_INTUIT_ONECHAR 0x04000000
120#define RE_INTUIT_TAIL 0x08000000
121
cad2e5aa
JH
122
123#define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
124#define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
125#define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
126
5d458dd8 127#define RX_HAS_CUTGROUP(prog) ((prog)->reganch & ROPT_CUTGROUP_SEEN)
c277df42 128#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
72311751
GS
129#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
130#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
131#define RX_MATCH_TAINTED_set(prog, t) ((t) \
132 ? RX_MATCH_TAINTED_on(prog) \
133 : RX_MATCH_TAINTED_off(prog))
c277df42 134
cf93c79d
IZ
135#define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
136#define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
137#define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
138#define RX_MATCH_COPIED_set(prog,t) ((t) \
139 ? RX_MATCH_COPIED_on(prog) \
140 : RX_MATCH_COPIED_off(prog))
f9f4320a 141
be8e71aa
YO
142#endif /* PLUGGABLE_RE_EXTENSION */
143
144/* Stuff that needs to be included in the plugable extension goes below here */
145
f8c7b90f 146#ifdef PERL_OLD_COPY_ON_WRITE
ed252734
NC
147#define RX_MATCH_COPY_FREE(rx) \
148 STMT_START {if (rx->saved_copy) { \
149 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
150 } \
151 if (RX_MATCH_COPIED(rx)) { \
152 Safefree(rx->subbeg); \
153 RX_MATCH_COPIED_off(rx); \
154 }} STMT_END
155#else
156#define RX_MATCH_COPY_FREE(rx) \
157 STMT_START {if (RX_MATCH_COPIED(rx)) { \
158 Safefree(rx->subbeg); \
159 RX_MATCH_COPIED_off(rx); \
160 }} STMT_END
161#endif
162
a30b2f1f
RGS
163#define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
164#define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
165#define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
166#define RX_MATCH_UTF8_set(prog, t) ((t) \
167 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
168 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
169
cad2e5aa
JH
170#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
171#define REXEC_CHECKED 0x02 /* check_substr already checked. */
172#define REXEC_SCREAM 0x04 /* use scream table. */
173#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 174#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 175
155aba94 176#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 177#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d
IZ
178
179#define FBMcf_TAIL_DOLLAR 1
cad2e5aa
JH
180#define FBMcf_TAIL_DOLLARM 2
181#define FBMcf_TAIL_Z 4
182#define FBMcf_TAIL_z 8
183#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d
IZ
184
185#define FBMrf_MULTILINE 1
cad2e5aa
JH
186
187struct re_scream_pos_data_s;
5d9a96ca
DM
188
189/* an accepting state/position*/
190struct _reg_trie_accepted {
191 U8 *endpos;
192 U16 wordnum;
193};
194typedef struct _reg_trie_accepted reg_trie_accepted;
195
3b0527fe
DM
196/* some basic information about the current match that is created by
197 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
198
199typedef struct {
200 regexp *prog;
201 char *bol;
202 char *till;
203 SV *sv;
204 char *ganch;
24b23f37 205 char *cutpoint;
3b0527fe
DM
206} regmatch_info;
207
5d9a96ca
DM
208
209/* structures for holding and saving the state maintained by regmatch() */
210
6bda09f9
YO
211#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
212
5d9a96ca
DM
213typedef I32 CHECKPOINT;
214
a0374537 215typedef struct regmatch_state {
40a82448 216 int resume_state; /* where to jump to on return */
24d3c4a9 217 char *locinput; /* where to backtrack in string on failure */
e822a8b4
DM
218
219 union {
77cb431f
DM
220
221 /* this is a fake union member that matches the first element
222 * of each member that needs to store positive backtrack
223 * information */
224 struct {
225 struct regmatch_state *prev_yes_state;
226 } yes;
227
e822a8b4 228 struct {
5d458dd8
YO
229 /* this first element must match u.yes */
230 struct regmatch_state *prev_yes_state;
e822a8b4 231 reg_trie_accepted *accept_buff;
166ba7cd 232 U32 accepted; /* how many accepting states we have seen */
7f69552c 233 U16 *jump; /* positive offsets from me */
166ba7cd 234 regnode *B; /* node following the trie */
7f69552c 235 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4
DM
236 } trie;
237
238 struct {
77cb431f
DM
239 /* this first element must match u.yes */
240 struct regmatch_state *prev_yes_state;
faec1544
DM
241 struct regmatch_state *prev_eval;
242 struct regmatch_state *prev_curlyx;
aa283a38 243 regexp *prev_rex;
faec1544
DM
244 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
245 flip when transitioning between
246 inner and outer rexen */
aa283a38
DM
247 CHECKPOINT cp; /* remember current savestack indexes */
248 CHECKPOINT lastcp;
40a82448 249 regnode *B; /* the node following us */
6bda09f9 250 U32 close_paren; /* which close bracket is our end */
e822a8b4
DM
251 } eval;
252
253 struct {
c476f425 254 /* this first element must match u.yes */
262b90c4 255 struct regmatch_state *prev_yes_state;
c476f425
DM
256 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
257 CHECKPOINT cp; /* remember current savestack index */
258 bool minmod;
a0374537 259 int parenfloor;/* how far back to strip paren data */
c476f425
DM
260 int min; /* the minimal number of A's to match */
261 int max; /* the maximal number of A's to match */
262 regnode *A, *B; /* the nodes corresponding to /A*B/ */
263
264 /* these two are modified by WHILEM */
265 int count; /* how many instances of A we've matched */
266 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4
DM
267 } curlyx;
268
269 struct {
c476f425 270 /* this first element must match u.yes */
262b90c4 271 struct regmatch_state *prev_yes_state;
c476f425
DM
272 struct regmatch_state *save_curlyx;
273 CHECKPOINT cp; /* remember current savestack indexes */
274 CHECKPOINT lastcp;
275 char *save_lastloc; /* previous curlyx.lastloc */
276 I32 cache_offset;
277 I32 cache_mask;
e822a8b4
DM
278 } whilem;
279
280 struct {
5d458dd8
YO
281 /* this first element must match u.yes */
282 struct regmatch_state *prev_yes_state;
3b6647e0 283 U32 lastparen;
40a82448
DM
284 regnode *next_branch; /* next branch node */
285 CHECKPOINT cp;
286 } branch;
287
288 struct {
dad79028
DM
289 /* this first element must match u.yes */
290 struct regmatch_state *prev_yes_state;
e822a8b4 291 I32 c1, c2; /* case fold search */
40a82448
DM
292 CHECKPOINT cp;
293 I32 alen; /* length of first-matched A string */
294 I32 count;
0cadcf80 295 bool minmod;
40a82448
DM
296 regnode *A, *B; /* the nodes corresponding to /A*B/ */
297 regnode *me; /* the curlym node */
e822a8b4
DM
298 } curlym;
299
300 struct {
3b6647e0 301 U32 paren;
c255a977 302 CHECKPOINT cp;
e822a8b4 303 I32 c1, c2; /* case fold search */
c255a977
DM
304 char *maxpos; /* highest possible point in string to match */
305 char *oldloc; /* the previous locinput */
e822a8b4 306 int count;
c255a977
DM
307 int min, max; /* {m,n} */
308 regnode *A, *B; /* the nodes corresponding to /A*B/ */
309 } curly; /* and CURLYN/PLUS/STAR */
dad79028
DM
310
311 struct {
312 /* this first element must match u.yes */
313 struct regmatch_state *prev_yes_state;
314 I32 wanted;
24d3c4a9 315 I32 logical; /* saved copy of 'logical' var */
40a82448 316 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
dad79028 317 } ifmatch; /* and SUSPEND/UNLESSM */
e2e6a0f1
YO
318
319 struct {
320 /* this first element must match u.yes */
321 struct regmatch_state *prev_yes_state;
322 struct regmatch_state *prev_mark;
323 SV* mark_name;
324 char *mark_loc;
325 } mark;
d8319b27 326 } u;
5d9a96ca
DM
327} regmatch_state;
328
329/* how many regmatch_state structs to allocate as a single slab.
330 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
331 * pointers, plus 1 for any mythical malloc overhead. */
332
333#define PERL_REGMATCH_SLAB_SLOTS \
334 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
335
336typedef struct regmatch_slab {
337 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
338 struct regmatch_slab *prev, *next;
339} regmatch_slab;
1ade1aa1 340
46ab3289
NC
341#define PL_reg_flags PL_reg_state.re_state_reg_flags
342#define PL_bostr PL_reg_state.re_state_bostr
343#define PL_reginput PL_reg_state.re_state_reginput
46ab3289
NC
344#define PL_regeol PL_reg_state.re_state_regeol
345#define PL_regstartp PL_reg_state.re_state_regstartp
346#define PL_regendp PL_reg_state.re_state_regendp
347#define PL_reglastparen PL_reg_state.re_state_reglastparen
348#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289
NC
349#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
350#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
351#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289
NC
352#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
353#define PL_reg_magic PL_reg_state.re_state_reg_magic
354#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
355#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
356#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
357#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
358#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
359#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
360#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
361#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
362#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
363#define PL_regsize PL_reg_state.re_state_regsize
364#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
365#define PL_nrs PL_reg_state.re_state_nrs
366
1ade1aa1
NC
367struct re_save_state {
368 U32 re_state_reg_flags; /* from regexec.c */
369 char *re_state_bostr;
370 char *re_state_reginput; /* String-input pointer. */
1ade1aa1
NC
371 char *re_state_regeol; /* End of input, for $ check. */
372 I32 *re_state_regstartp; /* Pointer to startp array. */
373 I32 *re_state_regendp; /* Ditto for endp. */
374 U32 *re_state_reglastparen; /* Similarly for lastparen. */
375 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1
NC
376 char **re_state_reg_start_tmp; /* from regexec.c */
377 U32 re_state_reg_start_tmpl; /* from regexec.c */
378 I32 re_state_reg_eval_set; /* from regexec.c */
1ade1aa1
NC
379 bool re_state_reg_match_utf8; /* from regexec.c */
380 MAGIC *re_state_reg_magic; /* from regexec.c */
381 I32 re_state_reg_oldpos; /* from regexec.c */
382 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
383 PMOP *re_state_reg_curpm; /* from regexec.c */
384 char *re_state_reg_oldsaved; /* old saved substr during match */
385 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
386 I32 re_state_reg_maxiter; /* max wait until caching pos */
387 I32 re_state_reg_leftiter; /* wait until caching pos */
388 char *re_state_reg_poscache; /* cache of pos of WHILEM */
389 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
3b6647e0 390 U32 re_state_regsize; /* from regexec.c */
1ade1aa1
NC
391 char *re_state_reg_starttry; /* from regexec.c */
392#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 393 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
394#endif
395};
396
397#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
398 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
399/*
400 * Local variables:
401 * c-indentation-style: bsd
402 * c-basic-offset: 4
403 * indent-tabs-mode: t
404 * End:
405 *
406 * ex: set ts=8 sts=4 sw=4 noet:
407 */