Plug a memory leak (noticed by Yves)
[p5sagit/p5-mst-13.2.git] / regexp.h
CommitLineData
a0d0e21e 1/* regexp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
699a97de 4 * 2005, 2006 by Larry Wall and others
d6376244 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 9 */
10
378cc40b 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 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 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 31struct reg_data;
32
f9f4320a 33struct regexp_engine;
28d8d7f4 34struct regexp;
bbe252da 35
785a26d5 36struct reg_substr_datum {
37 I32 min_offset;
38 I32 max_offset;
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
41 I32 end_shift;
42};
43struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
45};
f9f4320a 46
f8c7b90f 47#ifdef PERL_OLD_COPY_ON_WRITE
bbe252da 48#define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
49#else
50#define SV_SAVED_COPY
ed252734 51#endif
28d8d7f4 52
f0ab9afb 53typedef struct regexp_paren_pair {
54 I32 start;
55 I32 end;
56} regexp_paren_pair;
28d8d7f4 57
882227b7 58/*
59 The regexp/REGEXP struct, see L<perlreapi> for further documentation
60 on the individual fields. The struct is ordered so that the most
61 commonly used fields are placed at the start.
62
63 Any patch that adds items to this struct will need to include
64 changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c>
65 (C<pregfree()>). This involves freeing or cloning items in the
66 regexp's data array based on the data item's type.
67*/
68
bbe252da 69typedef struct regexp {
f8149455 70 /* what engine created this regexp? */
71 const struct regexp_engine* engine;
28d8d7f4 72 struct regexp* mother_re; /* what re is this a lightweight copy of? */
f8149455 73
74 /* Information about the match that the perl core uses to manage things */
bbe252da 75 U32 extflags; /* Flags used both externally and internally */
de8c5301 76 I32 minlen; /* mininum possible length of string to match */
77 I32 minlenret; /* mininum possible length of $& */
58e23c8d 78 U32 gofs; /* chars left of pos that we search from */
f8149455 79 struct reg_substr_data *substrs; /* substring data about strings that must appear
bbe252da 80 in the final match, used for optimisations */
f8149455 81 U32 nparens; /* number of capture buffers */
bbe252da 82
f8149455 83 /* private engine specific data */
84 U32 intflags; /* Engine Specific Internal flags */
85 void *pprivate; /* Data private to the regex engine which
86 created this object. */
87
88 /* Data about the last/current match. These are modified during matching*/
89 U32 lastparen; /* last open paren matched */
90 U32 lastcloseparen; /* last close paren matched */
f0ab9afb 91 regexp_paren_pair *swap; /* Swap copy of *offs */
92 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
28d8d7f4 93
bbe252da 94 char *subbeg; /* saved or original string
95 so \digit works forever. */
bbe252da 96 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
1b5c067c 97 I32 sublen; /* Length of string pointed by subbeg */
f8149455 98
99
100 /* Information about the match that isn't often used */
f8149455 101 I32 prelen; /* length of precomp */
1b5c067c 102 const char *precomp; /* pre-compilation regular expression */
d7b33318 103 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
bcdf7404 104 char *wrapped; /* wrapped version of the pattern */
105 I32 wraplen; /* length of wrapped */
f8149455 106 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
107 HV *paren_names; /* Optional hash of paren names */
108
109 /* Refcount of this regexp */
110 I32 refcnt; /* Refcount of this regexp */
f8fc2ecf 111} regexp;
112
785a26d5 113/* used for high speed searches */
f9f4320a 114typedef struct re_scream_pos_data_s
115{
116 char **scream_olds; /* match pos */
117 I32 *scream_pos; /* Internal iterator of scream. */
118} re_scream_pos_data;
119
785a26d5 120/* regexp_engine structure. This is the dispatch table for regexes.
121 * Any regex engine implementation must be able to build one of these.
122 */
f9f4320a 123typedef struct regexp_engine {
3ab4a224 124 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
49d7dfbc 125 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
126 char* strbeg, I32 minend, SV* screamer,
127 void* data, U32 flags);
128 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
9f61653a 129 char *strend, const U32 flags,
130 re_scream_pos_data *data);
49d7dfbc 131 SV* (*checkstr) (pTHX_ REGEXP * const rx);
132 void (*free) (pTHX_ REGEXP * const rx);
2fdbfb4d 133 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
192b9cd1 134 SV * const sv);
2fdbfb4d 135 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
136 SV const * const value);
137 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
138 const I32 paren);
192b9cd1 139 SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key,
140 SV * const value, const U32 flags);
141 SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey,
142 const U32 flags);
49d7dfbc 143 SV* (*qr_package)(pTHX_ REGEXP * const rx);
f2f78491 144#ifdef USE_ITHREADS
49d7dfbc 145 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
2fdbfb4d 146#endif
f9f4320a 147} regexp_engine;
148
192b9cd1 149/*
150 These are passed to the numbered capture variable callbacks as the
151 paren name. >= 1 is reserved for actual numbered captures, i.e. $1,
152 $2 etc.
153*/
154#define RXf_PREMATCH -2 /* $` / ${^PREMATCH} */
155#define RXf_POSTMATCH -1 /* $' / ${^POSTMATCH} */
156#define RXf_MATCH 0 /* $& / ${^MATCH} */
157
158/*
159 Flags that are passed to the named_buff and named_buff_iter
160 callbacks above. Those routines are called from universal.c via the
161 Tie::Hash::NamedCapture interface for %+ and %- and the re::
162 functions in the same file.
163*/
164
165/* The Tie::Hash::NamedCapture operation this is part of, if any */
166#define RXf_HASH_FETCH 0x0001
167#define RXf_HASH_STORE 0x0002
168#define RXf_HASH_DELETE 0x0004
169#define RXf_HASH_CLEAR 0x0008
170#define RXf_HASH_EXISTS 0x0010
171#define RXf_HASH_SCALAR 0x0020
172#define RXf_HASH_FIRSTKEY 0x0040
173#define RXf_HASH_NEXTKEY 0x0080
174
175/* Whether %+ or %- is being operated on */
176#define RXf_HASH_ONE 0x0100 /* %+ */
177#define RXf_HASH_ALL 0x0200 /* %- */
178
179/* Whether this is being called from a re:: function */
180#define RXf_HASH_REGNAME 0x0400
181#define RXf_HASH_REGNAMES 0x0800
182#define RXf_HASH_REGNAMES_COUNT 0x1000
183
bbe252da 184/* Flags stored in regexp->extflags
185 * These are used by code external to the regexp engine
e357fc67 186 *
187 * Note that flags starting with RXf_PMf_ have exact equivalents
188 * stored in op_pmflags and which are defined in op.h, they are defined
189 * numerically here only for clarity.
bbe252da 190 */
191
192/* Anchor and GPOS related stuff */
193#define RXf_ANCH_BOL 0x00000001
194#define RXf_ANCH_MBOL 0x00000002
195#define RXf_ANCH_SBOL 0x00000004
196#define RXf_ANCH_GPOS 0x00000008
197#define RXf_GPOS_SEEN 0x00000010
198#define RXf_GPOS_FLOAT 0x00000020
e357fc67 199/* two bits here */
bbe252da 200#define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
201#define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
e357fc67 202#define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
203
204/* Flags indicating special patterns */
c737faaf 205#define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */
e357fc67 206#define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
207#define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
208
bcdf7404 209/* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */
e357fc67 210#define RXf_PMf_LOCALE 0x00000800 /* use locale */
211#define RXf_PMf_MULTILINE 0x00001000 /* /m */
212#define RXf_PMf_SINGLELINE 0x00002000 /* /s */
213#define RXf_PMf_FOLD 0x00004000 /* /i */
214#define RXf_PMf_EXTENDED 0x00008000 /* /x */
cde0cee5 215#define RXf_PMf_KEEPCOPY 0x00010000 /* /k */
e357fc67 216/* these flags are transfered from the PMOP->op_pmflags member during compilation */
cde0cee5 217#define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
218#define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
219
a20207d7 220#define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
221 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
222 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
223 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
224 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
bbe252da 225
bcdf7404 226/* chars and strings used as regex pattern modifiers
227 * Singlular is a 'c'har, plural is a "string"
87e95b7f 228 *
229 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
230 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
231 * for its own uses. So 'k' is out as well.
bcdf7404 232 */
233#define EXEC_PAT_MOD 'e'
87e95b7f 234#define KEEPCOPY_PAT_MOD 'p'
bcdf7404 235#define ONCE_PAT_MOD 'o'
236#define GLOBAL_PAT_MOD 'g'
237#define CONTINUE_PAT_MOD 'c'
238#define MULTILINE_PAT_MOD 'm'
239#define SINGLE_PAT_MOD 's'
240#define IGNORE_PAT_MOD 'i'
241#define XTENDED_PAT_MOD 'x'
242
243#define ONCE_PAT_MODS "o"
87e95b7f 244#define KEEPCOPY_PAT_MODS "p"
bcdf7404 245#define EXEC_PAT_MODS "e"
246#define LOOP_PAT_MODS "gc"
247
248#define STD_PAT_MODS "msix"
249
250#define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
251
252#define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
253#define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
254#define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
255#define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
256
257
bbe252da 258/* What we have seen */
bbe252da 259#define RXf_LOOKBEHIND_SEEN 0x00020000
260#define RXf_EVAL_SEEN 0x00040000
261#define RXf_CANY_SEEN 0x00080000
262
263/* Special */
264#define RXf_NOSCAN 0x00100000
265#define RXf_CHECK_ALL 0x00200000
266
267/* UTF8 related */
268#define RXf_UTF8 0x00400000
269#define RXf_MATCH_UTF8 0x00800000
270
271/* Intuit related */
272#define RXf_USE_INTUIT_NOML 0x01000000
273#define RXf_USE_INTUIT_ML 0x02000000
274#define RXf_INTUIT_TAIL 0x04000000
275/* one bit here */
276#define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
277
278/* Copy and tainted info */
279#define RXf_COPY_DONE 0x10000000
280#define RXf_TAINTED_SEEN 0x20000000
c737faaf 281#define RXf_TAINTED 0x80000000 /* this pattern is tainted */
282
bbe252da 283
bbe252da 284#define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
285#define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
286#define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
287#define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
72311751 288#define RX_MATCH_TAINTED_set(prog, t) ((t) \
289 ? RX_MATCH_TAINTED_on(prog) \
290 : RX_MATCH_TAINTED_off(prog))
c277df42 291
bbe252da 292#define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
293#define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
294#define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
cf93c79d 295#define RX_MATCH_COPIED_set(prog,t) ((t) \
296 ? RX_MATCH_COPIED_on(prog) \
297 : RX_MATCH_COPIED_off(prog))
f9f4320a 298
be8e71aa 299#endif /* PLUGGABLE_RE_EXTENSION */
300
301/* Stuff that needs to be included in the plugable extension goes below here */
302
f8c7b90f 303#ifdef PERL_OLD_COPY_ON_WRITE
ed252734 304#define RX_MATCH_COPY_FREE(rx) \
305 STMT_START {if (rx->saved_copy) { \
306 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
307 } \
308 if (RX_MATCH_COPIED(rx)) { \
309 Safefree(rx->subbeg); \
310 RX_MATCH_COPIED_off(rx); \
311 }} STMT_END
312#else
313#define RX_MATCH_COPY_FREE(rx) \
314 STMT_START {if (RX_MATCH_COPIED(rx)) { \
315 Safefree(rx->subbeg); \
316 RX_MATCH_COPIED_off(rx); \
317 }} STMT_END
318#endif
319
bbe252da 320#define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
321#define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
322#define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
a30b2f1f 323#define RX_MATCH_UTF8_set(prog, t) ((t) \
324 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
325 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
326
cad2e5aa 327#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
328#define REXEC_CHECKED 0x02 /* check_substr already checked. */
329#define REXEC_SCREAM 0x04 /* use scream table. */
330#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
cf93c79d 331#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
c277df42 332
155aba94 333#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
f9f4320a 334#define ReREFCNT_dec(re) CALLREGFREE(re)
cf93c79d 335
336#define FBMcf_TAIL_DOLLAR 1
cad2e5aa 337#define FBMcf_TAIL_DOLLARM 2
338#define FBMcf_TAIL_Z 4
339#define FBMcf_TAIL_z 8
340#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
cf93c79d 341
342#define FBMrf_MULTILINE 1
cad2e5aa 343
5d9a96ca 344/* an accepting state/position*/
345struct _reg_trie_accepted {
346 U8 *endpos;
347 U16 wordnum;
348};
349typedef struct _reg_trie_accepted reg_trie_accepted;
350
3b0527fe 351/* some basic information about the current match that is created by
352 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
353
354typedef struct {
355 regexp *prog;
356 char *bol;
357 char *till;
358 SV *sv;
359 char *ganch;
24b23f37 360 char *cutpoint;
3b0527fe 361} regmatch_info;
362
5d9a96ca 363
364/* structures for holding and saving the state maintained by regmatch() */
365
d56b3014 366#ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
4b196cd4 367#define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
d56b3014 368#endif
6bda09f9 369
5d9a96ca 370typedef I32 CHECKPOINT;
371
a0374537 372typedef struct regmatch_state {
40a82448 373 int resume_state; /* where to jump to on return */
24d3c4a9 374 char *locinput; /* where to backtrack in string on failure */
e822a8b4 375
376 union {
77cb431f 377
378 /* this is a fake union member that matches the first element
379 * of each member that needs to store positive backtrack
380 * information */
381 struct {
382 struct regmatch_state *prev_yes_state;
383 } yes;
384
fae667d5 385 /* branchlike members */
386 /* this is a fake union member that matches the first elements
387 * of each member that needs to behave like a branch */
388 struct {
5d458dd8 389 /* this first element must match u.yes */
390 struct regmatch_state *prev_yes_state;
fae667d5 391 U32 lastparen;
392 CHECKPOINT cp;
393
394 } branchlike;
395
396 struct {
397 /* the first elements must match u.branchlike */
398 struct regmatch_state *prev_yes_state;
399 U32 lastparen;
400 CHECKPOINT cp;
401
402 regnode *next_branch; /* next branch node */
403 } branch;
404
405 struct {
406 /* the first elements must match u.branchlike */
407 struct regmatch_state *prev_yes_state;
408 U32 lastparen;
409 CHECKPOINT cp;
410
411 reg_trie_accepted *accept_buff; /* accepting states we have seen */
166ba7cd 412 U32 accepted; /* how many accepting states we have seen */
7f69552c 413 U16 *jump; /* positive offsets from me */
166ba7cd 414 regnode *B; /* node following the trie */
7f69552c 415 regnode *me; /* Which node am I - needed for jump tries*/
e822a8b4 416 } trie;
417
fae667d5 418 /* special types - these members are used to store state for special
419 regops like eval, if/then, lookaround and the markpoint state */
e822a8b4 420 struct {
77cb431f 421 /* this first element must match u.yes */
422 struct regmatch_state *prev_yes_state;
faec1544 423 struct regmatch_state *prev_eval;
424 struct regmatch_state *prev_curlyx;
aa283a38 425 regexp *prev_rex;
faec1544 426 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
427 flip when transitioning between
428 inner and outer rexen */
aa283a38 429 CHECKPOINT cp; /* remember current savestack indexes */
430 CHECKPOINT lastcp;
6bda09f9 431 U32 close_paren; /* which close bracket is our end */
1b5c067c 432 regnode *B; /* the node following us */
e822a8b4 433 } eval;
434
435 struct {
c476f425 436 /* this first element must match u.yes */
262b90c4 437 struct regmatch_state *prev_yes_state;
fae667d5 438 I32 wanted;
439 I32 logical; /* saved copy of 'logical' var */
440 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
441 } ifmatch; /* and SUSPEND/UNLESSM */
442
443 struct {
444 /* this first element must match u.yes */
445 struct regmatch_state *prev_yes_state;
446 struct regmatch_state *prev_mark;
447 SV* mark_name;
448 char *mark_loc;
449 } mark;
450
451 struct {
452 int val;
453 } keeper;
454
455 /* quantifiers - these members are used for storing state for
456 for the regops used to implement quantifiers */
457 struct {
458 /* this first element must match u.yes */
459 struct regmatch_state *prev_yes_state;
c476f425 460 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
1b5c067c 461 regnode *A, *B; /* the nodes corresponding to /A*B/ */
c476f425 462 CHECKPOINT cp; /* remember current savestack index */
463 bool minmod;
a0374537 464 int parenfloor;/* how far back to strip paren data */
c476f425 465 int min; /* the minimal number of A's to match */
466 int max; /* the maximal number of A's to match */
c476f425 467
468 /* these two are modified by WHILEM */
469 int count; /* how many instances of A we've matched */
470 char *lastloc;/* where previous A matched (0-len detect) */
e822a8b4 471 } curlyx;
472
473 struct {
c476f425 474 /* this first element must match u.yes */
262b90c4 475 struct regmatch_state *prev_yes_state;
c476f425 476 struct regmatch_state *save_curlyx;
477 CHECKPOINT cp; /* remember current savestack indexes */
478 CHECKPOINT lastcp;
479 char *save_lastloc; /* previous curlyx.lastloc */
480 I32 cache_offset;
481 I32 cache_mask;
e822a8b4 482 } whilem;
483
484 struct {
5d458dd8 485 /* this first element must match u.yes */
486 struct regmatch_state *prev_yes_state;
e822a8b4 487 I32 c1, c2; /* case fold search */
40a82448 488 CHECKPOINT cp;
489 I32 alen; /* length of first-matched A string */
490 I32 count;
0cadcf80 491 bool minmod;
40a82448 492 regnode *A, *B; /* the nodes corresponding to /A*B/ */
493 regnode *me; /* the curlym node */
e822a8b4 494 } curlym;
495
496 struct {
3b6647e0 497 U32 paren;
c255a977 498 CHECKPOINT cp;
e822a8b4 499 I32 c1, c2; /* case fold search */
c255a977 500 char *maxpos; /* highest possible point in string to match */
501 char *oldloc; /* the previous locinput */
e822a8b4 502 int count;
c255a977 503 int min, max; /* {m,n} */
504 regnode *A, *B; /* the nodes corresponding to /A*B/ */
505 } curly; /* and CURLYN/PLUS/STAR */
dad79028 506
d8319b27 507 } u;
5d9a96ca 508} regmatch_state;
509
510/* how many regmatch_state structs to allocate as a single slab.
511 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
512 * pointers, plus 1 for any mythical malloc overhead. */
513
514#define PERL_REGMATCH_SLAB_SLOTS \
515 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
516
517typedef struct regmatch_slab {
518 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
519 struct regmatch_slab *prev, *next;
520} regmatch_slab;
1ade1aa1 521
46ab3289 522#define PL_reg_flags PL_reg_state.re_state_reg_flags
523#define PL_bostr PL_reg_state.re_state_bostr
524#define PL_reginput PL_reg_state.re_state_reginput
46ab3289 525#define PL_regeol PL_reg_state.re_state_regeol
f0ab9afb 526#define PL_regoffs PL_reg_state.re_state_regoffs
46ab3289 527#define PL_reglastparen PL_reg_state.re_state_reglastparen
528#define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
46ab3289 529#define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
530#define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
531#define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
46ab3289 532#define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
533#define PL_reg_magic PL_reg_state.re_state_reg_magic
534#define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
535#define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
536#define PL_reg_curpm PL_reg_state.re_state_reg_curpm
537#define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
538#define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
539#define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
540#define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
541#define PL_reg_poscache PL_reg_state.re_state_reg_poscache
542#define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
543#define PL_regsize PL_reg_state.re_state_regsize
544#define PL_reg_starttry PL_reg_state.re_state_reg_starttry
545#define PL_nrs PL_reg_state.re_state_nrs
546
1ade1aa1 547struct re_save_state {
548 U32 re_state_reg_flags; /* from regexec.c */
1b5c067c 549 U32 re_state_reg_start_tmpl; /* from regexec.c */
550 I32 re_state_reg_eval_set; /* from regexec.c */
551 bool re_state_reg_match_utf8; /* from regexec.c */
1ade1aa1 552 char *re_state_bostr;
553 char *re_state_reginput; /* String-input pointer. */
1ade1aa1 554 char *re_state_regeol; /* End of input, for $ check. */
f0ab9afb 555 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
1ade1aa1 556 U32 *re_state_reglastparen; /* Similarly for lastparen. */
557 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
1ade1aa1 558 char **re_state_reg_start_tmp; /* from regexec.c */
1ade1aa1 559 MAGIC *re_state_reg_magic; /* from regexec.c */
1ade1aa1 560 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
561 PMOP *re_state_reg_curpm; /* from regexec.c */
562 char *re_state_reg_oldsaved; /* old saved substr during match */
563 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
1b5c067c 564 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
565 I32 re_state_reg_oldpos; /* from regexec.c */
1ade1aa1 566 I32 re_state_reg_maxiter; /* max wait until caching pos */
567 I32 re_state_reg_leftiter; /* wait until caching pos */
3b6647e0 568 U32 re_state_regsize; /* from regexec.c */
1b5c067c 569 char *re_state_reg_poscache; /* cache of pos of WHILEM */
1ade1aa1 570 char *re_state_reg_starttry; /* from regexec.c */
571#ifdef PERL_OLD_COPY_ON_WRITE
46ab3289 572 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 573#endif
574};
575
576#define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
577 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
cde0cee5 578
1ade1aa1 579/*
580 * Local variables:
581 * c-indentation-style: bsd
582 * c-basic-offset: 4
583 * indent-tabs-mode: t
584 * End:
585 *
586 * ex: set ts=8 sts=4 sw=4 noet:
587 */