Commit | Line | Data |
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 |
21 | struct regnode { |
22 | U8 flags; |
23 | U8 type; |
24 | U16 next_off; |
25 | }; |
26 | |
27 | typedef struct regnode regnode; |
28 | |
cad2e5aa |
29 | struct reg_substr_data; |
2779dcf1 |
30 | |
0ee3c8d0 |
31 | struct reg_data; |
32 | |
f9f4320a |
33 | struct regexp_engine; |
28d8d7f4 |
34 | struct regexp; |
bbe252da |
35 | |
785a26d5 |
36 | struct 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 | }; |
43 | struct 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 | |
53 | /* swap buffer for paren structs */ |
54 | typedef struct regexp_paren_ofs { |
55 | I32 *startp; |
56 | I32 *endp; |
57 | } regexp_paren_ofs; |
58 | |
f8149455 |
59 | /* this is ordered such that the most commonly used |
60 | fields are at the start of the struct */ |
bbe252da |
61 | typedef struct regexp { |
f8149455 |
62 | /* what engine created this regexp? */ |
63 | const struct regexp_engine* engine; |
28d8d7f4 |
64 | struct regexp* mother_re; /* what re is this a lightweight copy of? */ |
f8149455 |
65 | |
66 | /* Information about the match that the perl core uses to manage things */ |
bbe252da |
67 | U32 extflags; /* Flags used both externally and internally */ |
de8c5301 |
68 | I32 minlen; /* mininum possible length of string to match */ |
69 | I32 minlenret; /* mininum possible length of $& */ |
58e23c8d |
70 | U32 gofs; /* chars left of pos that we search from */ |
f8149455 |
71 | struct reg_substr_data *substrs; /* substring data about strings that must appear |
bbe252da |
72 | in the final match, used for optimisations */ |
f8149455 |
73 | U32 nparens; /* number of capture buffers */ |
bbe252da |
74 | |
f8149455 |
75 | /* private engine specific data */ |
76 | U32 intflags; /* Engine Specific Internal flags */ |
77 | void *pprivate; /* Data private to the regex engine which |
78 | created this object. */ |
79 | |
80 | /* Data about the last/current match. These are modified during matching*/ |
81 | U32 lastparen; /* last open paren matched */ |
82 | U32 lastcloseparen; /* last close paren matched */ |
28d8d7f4 |
83 | regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */ |
f8149455 |
84 | I32 *startp; /* Array of offsets from start of string (@-) */ |
85 | I32 *endp; /* Array of offsets from start of string (@+) */ |
28d8d7f4 |
86 | |
bbe252da |
87 | char *subbeg; /* saved or original string |
88 | so \digit works forever. */ |
89 | I32 sublen; /* Length of string pointed by subbeg */ |
bbe252da |
90 | SV_SAVED_COPY /* If non-NULL, SV which is COW from original */ |
f8149455 |
91 | |
92 | |
93 | /* Information about the match that isn't often used */ |
94 | char *precomp; /* pre-compilation regular expression */ |
95 | I32 prelen; /* length of precomp */ |
bcdf7404 |
96 | char *wrapped; /* wrapped version of the pattern */ |
97 | I32 wraplen; /* length of wrapped */ |
f8149455 |
98 | I32 seen_evals; /* number of eval groups in the pattern - for security checks */ |
99 | HV *paren_names; /* Optional hash of paren names */ |
100 | |
101 | /* Refcount of this regexp */ |
102 | I32 refcnt; /* Refcount of this regexp */ |
f8fc2ecf |
103 | } regexp; |
104 | |
785a26d5 |
105 | /* used for high speed searches */ |
f9f4320a |
106 | typedef 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 | |
785a26d5 |
112 | /* regexp_engine structure. This is the dispatch table for regexes. |
113 | * Any regex engine implementation must be able to build one of these. |
114 | */ |
f9f4320a |
115 | typedef struct regexp_engine { |
f2f78491 |
116 | regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm); |
117 | I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend, |
f9f4320a |
118 | char* strbeg, I32 minend, SV* screamer, |
119 | void* data, U32 flags); |
f2f78491 |
120 | char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos, |
121 | char *strend, U32 flags, |
122 | struct re_scream_pos_data_s *data); |
123 | SV* (*checkstr) (pTHX_ regexp *prog); |
124 | void (*free) (pTHX_ struct regexp* r); |
93b32b6d |
125 | SV* (*numbered_buff_get) (pTHX_ const REGEXP * const rx, I32 paren, SV* usesv); |
126 | SV* (*named_buff_get)(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags); |
f2f78491 |
127 | #ifdef USE_ITHREADS |
f8149455 |
128 | void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param); |
f9f4320a |
129 | #endif |
130 | } regexp_engine; |
131 | |
bbe252da |
132 | /* Flags stored in regexp->extflags |
133 | * These are used by code external to the regexp engine |
e357fc67 |
134 | * |
135 | * Note that flags starting with RXf_PMf_ have exact equivalents |
136 | * stored in op_pmflags and which are defined in op.h, they are defined |
137 | * numerically here only for clarity. |
bbe252da |
138 | */ |
139 | |
140 | /* Anchor and GPOS related stuff */ |
141 | #define RXf_ANCH_BOL 0x00000001 |
142 | #define RXf_ANCH_MBOL 0x00000002 |
143 | #define RXf_ANCH_SBOL 0x00000004 |
144 | #define RXf_ANCH_GPOS 0x00000008 |
145 | #define RXf_GPOS_SEEN 0x00000010 |
146 | #define RXf_GPOS_FLOAT 0x00000020 |
e357fc67 |
147 | /* two bits here */ |
bbe252da |
148 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) |
149 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) |
e357fc67 |
150 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) |
151 | |
152 | /* Flags indicating special patterns */ |
153 | #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ |
154 | #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ |
155 | |
bcdf7404 |
156 | /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */ |
e357fc67 |
157 | #define RXf_PMf_LOCALE 0x00000800 /* use locale */ |
158 | #define RXf_PMf_MULTILINE 0x00001000 /* /m */ |
159 | #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ |
160 | #define RXf_PMf_FOLD 0x00004000 /* /i */ |
161 | #define RXf_PMf_EXTENDED 0x00008000 /* /x */ |
cde0cee5 |
162 | #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */ |
e357fc67 |
163 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
cde0cee5 |
164 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
165 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) |
166 | |
a20207d7 |
167 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
168 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ |
169 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ |
170 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ |
171 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break |
bbe252da |
172 | |
bcdf7404 |
173 | /* chars and strings used as regex pattern modifiers |
174 | * Singlular is a 'c'har, plural is a "string" |
87e95b7f |
175 | * |
176 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve |
177 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) |
178 | * for its own uses. So 'k' is out as well. |
bcdf7404 |
179 | */ |
180 | #define EXEC_PAT_MOD 'e' |
87e95b7f |
181 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 |
182 | #define ONCE_PAT_MOD 'o' |
183 | #define GLOBAL_PAT_MOD 'g' |
184 | #define CONTINUE_PAT_MOD 'c' |
185 | #define MULTILINE_PAT_MOD 'm' |
186 | #define SINGLE_PAT_MOD 's' |
187 | #define IGNORE_PAT_MOD 'i' |
188 | #define XTENDED_PAT_MOD 'x' |
189 | |
190 | #define ONCE_PAT_MODS "o" |
87e95b7f |
191 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 |
192 | #define EXEC_PAT_MODS "e" |
193 | #define LOOP_PAT_MODS "gc" |
194 | |
195 | #define STD_PAT_MODS "msix" |
196 | |
197 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS |
198 | |
199 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS |
200 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS |
201 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS |
202 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS |
203 | |
204 | |
bbe252da |
205 | /* What we have seen */ |
bbe252da |
206 | #define RXf_LOOKBEHIND_SEEN 0x00020000 |
207 | #define RXf_EVAL_SEEN 0x00040000 |
208 | #define RXf_CANY_SEEN 0x00080000 |
209 | |
210 | /* Special */ |
211 | #define RXf_NOSCAN 0x00100000 |
212 | #define RXf_CHECK_ALL 0x00200000 |
213 | |
214 | /* UTF8 related */ |
215 | #define RXf_UTF8 0x00400000 |
216 | #define RXf_MATCH_UTF8 0x00800000 |
217 | |
218 | /* Intuit related */ |
219 | #define RXf_USE_INTUIT_NOML 0x01000000 |
220 | #define RXf_USE_INTUIT_ML 0x02000000 |
221 | #define RXf_INTUIT_TAIL 0x04000000 |
222 | /* one bit here */ |
223 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
224 | |
225 | /* Copy and tainted info */ |
226 | #define RXf_COPY_DONE 0x10000000 |
227 | #define RXf_TAINTED_SEEN 0x20000000 |
228 | /* two bits here */ |
229 | |
bbe252da |
230 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
231 | #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN) |
232 | #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN) |
233 | #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN) |
72311751 |
234 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
235 | ? RX_MATCH_TAINTED_on(prog) \ |
236 | : RX_MATCH_TAINTED_off(prog)) |
c277df42 |
237 | |
bbe252da |
238 | #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE) |
239 | #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE) |
240 | #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE) |
cf93c79d |
241 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
242 | ? RX_MATCH_COPIED_on(prog) \ |
243 | : RX_MATCH_COPIED_off(prog)) |
f9f4320a |
244 | |
be8e71aa |
245 | #endif /* PLUGGABLE_RE_EXTENSION */ |
246 | |
247 | /* Stuff that needs to be included in the plugable extension goes below here */ |
248 | |
f8c7b90f |
249 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 |
250 | #define RX_MATCH_COPY_FREE(rx) \ |
251 | STMT_START {if (rx->saved_copy) { \ |
252 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ |
253 | } \ |
254 | if (RX_MATCH_COPIED(rx)) { \ |
255 | Safefree(rx->subbeg); \ |
256 | RX_MATCH_COPIED_off(rx); \ |
257 | }} STMT_END |
258 | #else |
259 | #define RX_MATCH_COPY_FREE(rx) \ |
260 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
261 | Safefree(rx->subbeg); \ |
262 | RX_MATCH_COPIED_off(rx); \ |
263 | }} STMT_END |
264 | #endif |
265 | |
bbe252da |
266 | #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8) |
267 | #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8) |
268 | #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8) |
a30b2f1f |
269 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
270 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
271 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
272 | |
cad2e5aa |
273 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
274 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
275 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
276 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
cf93c79d |
277 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 |
278 | |
155aba94 |
279 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a |
280 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d |
281 | |
282 | #define FBMcf_TAIL_DOLLAR 1 |
cad2e5aa |
283 | #define FBMcf_TAIL_DOLLARM 2 |
284 | #define FBMcf_TAIL_Z 4 |
285 | #define FBMcf_TAIL_z 8 |
286 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
cf93c79d |
287 | |
288 | #define FBMrf_MULTILINE 1 |
cad2e5aa |
289 | |
5d9a96ca |
290 | /* an accepting state/position*/ |
291 | struct _reg_trie_accepted { |
292 | U8 *endpos; |
293 | U16 wordnum; |
294 | }; |
295 | typedef struct _reg_trie_accepted reg_trie_accepted; |
296 | |
3b0527fe |
297 | /* some basic information about the current match that is created by |
298 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ |
299 | |
300 | typedef struct { |
301 | regexp *prog; |
302 | char *bol; |
303 | char *till; |
304 | SV *sv; |
305 | char *ganch; |
24b23f37 |
306 | char *cutpoint; |
3b0527fe |
307 | } regmatch_info; |
308 | |
5d9a96ca |
309 | |
310 | /* structures for holding and saving the state maintained by regmatch() */ |
311 | |
d56b3014 |
312 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 |
313 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 |
314 | #endif |
6bda09f9 |
315 | |
5d9a96ca |
316 | typedef I32 CHECKPOINT; |
317 | |
a0374537 |
318 | typedef struct regmatch_state { |
40a82448 |
319 | int resume_state; /* where to jump to on return */ |
24d3c4a9 |
320 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 |
321 | |
322 | union { |
77cb431f |
323 | |
324 | /* this is a fake union member that matches the first element |
325 | * of each member that needs to store positive backtrack |
326 | * information */ |
327 | struct { |
328 | struct regmatch_state *prev_yes_state; |
329 | } yes; |
330 | |
fae667d5 |
331 | /* branchlike members */ |
332 | /* this is a fake union member that matches the first elements |
333 | * of each member that needs to behave like a branch */ |
334 | struct { |
5d458dd8 |
335 | /* this first element must match u.yes */ |
336 | struct regmatch_state *prev_yes_state; |
fae667d5 |
337 | U32 lastparen; |
338 | CHECKPOINT cp; |
339 | |
340 | } branchlike; |
341 | |
342 | struct { |
343 | /* the first elements must match u.branchlike */ |
344 | struct regmatch_state *prev_yes_state; |
345 | U32 lastparen; |
346 | CHECKPOINT cp; |
347 | |
348 | regnode *next_branch; /* next branch node */ |
349 | } branch; |
350 | |
351 | struct { |
352 | /* the first elements must match u.branchlike */ |
353 | struct regmatch_state *prev_yes_state; |
354 | U32 lastparen; |
355 | CHECKPOINT cp; |
356 | |
357 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ |
166ba7cd |
358 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c |
359 | U16 *jump; /* positive offsets from me */ |
166ba7cd |
360 | regnode *B; /* node following the trie */ |
7f69552c |
361 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 |
362 | } trie; |
363 | |
fae667d5 |
364 | /* special types - these members are used to store state for special |
365 | regops like eval, if/then, lookaround and the markpoint state */ |
e822a8b4 |
366 | struct { |
77cb431f |
367 | /* this first element must match u.yes */ |
368 | struct regmatch_state *prev_yes_state; |
faec1544 |
369 | struct regmatch_state *prev_eval; |
370 | struct regmatch_state *prev_curlyx; |
aa283a38 |
371 | regexp *prev_rex; |
faec1544 |
372 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
373 | flip when transitioning between |
374 | inner and outer rexen */ |
aa283a38 |
375 | CHECKPOINT cp; /* remember current savestack indexes */ |
376 | CHECKPOINT lastcp; |
40a82448 |
377 | regnode *B; /* the node following us */ |
6bda09f9 |
378 | U32 close_paren; /* which close bracket is our end */ |
e822a8b4 |
379 | } eval; |
380 | |
381 | struct { |
c476f425 |
382 | /* this first element must match u.yes */ |
262b90c4 |
383 | struct regmatch_state *prev_yes_state; |
fae667d5 |
384 | I32 wanted; |
385 | I32 logical; /* saved copy of 'logical' var */ |
386 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ |
387 | } ifmatch; /* and SUSPEND/UNLESSM */ |
388 | |
389 | struct { |
390 | /* this first element must match u.yes */ |
391 | struct regmatch_state *prev_yes_state; |
392 | struct regmatch_state *prev_mark; |
393 | SV* mark_name; |
394 | char *mark_loc; |
395 | } mark; |
396 | |
397 | struct { |
398 | int val; |
399 | } keeper; |
400 | |
401 | /* quantifiers - these members are used for storing state for |
402 | for the regops used to implement quantifiers */ |
403 | struct { |
404 | /* this first element must match u.yes */ |
405 | struct regmatch_state *prev_yes_state; |
c476f425 |
406 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
407 | CHECKPOINT cp; /* remember current savestack index */ |
408 | bool minmod; |
a0374537 |
409 | int parenfloor;/* how far back to strip paren data */ |
c476f425 |
410 | int min; /* the minimal number of A's to match */ |
411 | int max; /* the maximal number of A's to match */ |
412 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
413 | |
414 | /* these two are modified by WHILEM */ |
415 | int count; /* how many instances of A we've matched */ |
416 | char *lastloc;/* where previous A matched (0-len detect) */ |
e822a8b4 |
417 | } curlyx; |
418 | |
419 | struct { |
c476f425 |
420 | /* this first element must match u.yes */ |
262b90c4 |
421 | struct regmatch_state *prev_yes_state; |
c476f425 |
422 | struct regmatch_state *save_curlyx; |
423 | CHECKPOINT cp; /* remember current savestack indexes */ |
424 | CHECKPOINT lastcp; |
425 | char *save_lastloc; /* previous curlyx.lastloc */ |
426 | I32 cache_offset; |
427 | I32 cache_mask; |
e822a8b4 |
428 | } whilem; |
429 | |
430 | struct { |
5d458dd8 |
431 | /* this first element must match u.yes */ |
432 | struct regmatch_state *prev_yes_state; |
e822a8b4 |
433 | I32 c1, c2; /* case fold search */ |
40a82448 |
434 | CHECKPOINT cp; |
435 | I32 alen; /* length of first-matched A string */ |
436 | I32 count; |
0cadcf80 |
437 | bool minmod; |
40a82448 |
438 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
439 | regnode *me; /* the curlym node */ |
e822a8b4 |
440 | } curlym; |
441 | |
442 | struct { |
3b6647e0 |
443 | U32 paren; |
c255a977 |
444 | CHECKPOINT cp; |
e822a8b4 |
445 | I32 c1, c2; /* case fold search */ |
c255a977 |
446 | char *maxpos; /* highest possible point in string to match */ |
447 | char *oldloc; /* the previous locinput */ |
e822a8b4 |
448 | int count; |
c255a977 |
449 | int min, max; /* {m,n} */ |
450 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
451 | } curly; /* and CURLYN/PLUS/STAR */ |
dad79028 |
452 | |
d8319b27 |
453 | } u; |
5d9a96ca |
454 | } regmatch_state; |
455 | |
456 | /* how many regmatch_state structs to allocate as a single slab. |
457 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
458 | * pointers, plus 1 for any mythical malloc overhead. */ |
459 | |
460 | #define PERL_REGMATCH_SLAB_SLOTS \ |
461 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
462 | |
463 | typedef struct regmatch_slab { |
464 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
465 | struct regmatch_slab *prev, *next; |
466 | } regmatch_slab; |
1ade1aa1 |
467 | |
46ab3289 |
468 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
469 | #define PL_bostr PL_reg_state.re_state_bostr |
470 | #define PL_reginput PL_reg_state.re_state_reginput |
46ab3289 |
471 | #define PL_regeol PL_reg_state.re_state_regeol |
472 | #define PL_regstartp PL_reg_state.re_state_regstartp |
473 | #define PL_regendp PL_reg_state.re_state_regendp |
474 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
475 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen |
46ab3289 |
476 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
477 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl |
478 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set |
46ab3289 |
479 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
480 | #define PL_reg_magic PL_reg_state.re_state_reg_magic |
481 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos |
482 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm |
483 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm |
484 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved |
485 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen |
486 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter |
487 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter |
488 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache |
489 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size |
490 | #define PL_regsize PL_reg_state.re_state_regsize |
491 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry |
492 | #define PL_nrs PL_reg_state.re_state_nrs |
493 | |
1ade1aa1 |
494 | struct re_save_state { |
495 | U32 re_state_reg_flags; /* from regexec.c */ |
496 | char *re_state_bostr; |
497 | char *re_state_reginput; /* String-input pointer. */ |
1ade1aa1 |
498 | char *re_state_regeol; /* End of input, for $ check. */ |
499 | I32 *re_state_regstartp; /* Pointer to startp array. */ |
500 | I32 *re_state_regendp; /* Ditto for endp. */ |
501 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
502 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ |
1ade1aa1 |
503 | char **re_state_reg_start_tmp; /* from regexec.c */ |
504 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
505 | I32 re_state_reg_eval_set; /* from regexec.c */ |
1ade1aa1 |
506 | bool re_state_reg_match_utf8; /* from regexec.c */ |
507 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
508 | I32 re_state_reg_oldpos; /* from regexec.c */ |
509 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
510 | PMOP *re_state_reg_curpm; /* from regexec.c */ |
511 | char *re_state_reg_oldsaved; /* old saved substr during match */ |
512 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ |
513 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
514 | I32 re_state_reg_leftiter; /* wait until caching pos */ |
515 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
516 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
3b6647e0 |
517 | U32 re_state_regsize; /* from regexec.c */ |
1ade1aa1 |
518 | char *re_state_reg_starttry; /* from regexec.c */ |
519 | #ifdef PERL_OLD_COPY_ON_WRITE |
46ab3289 |
520 | 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 |
521 | #endif |
522 | }; |
523 | |
524 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ |
525 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) |
cde0cee5 |
526 | |
1ade1aa1 |
527 | /* |
528 | * Local variables: |
529 | * c-indentation-style: bsd |
530 | * c-basic-offset: 4 |
531 | * indent-tabs-mode: t |
532 | * End: |
533 | * |
534 | * ex: set ts=8 sts=4 sw=4 noet: |
535 | */ |