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 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 | 17 | #ifndef PLUGGABLE_RE_EXTENSION |
785a26d5 YO |
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 IZ |
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 JH |
31 | struct reg_data; |
32 | ||
f9f4320a | 33 | struct regexp_engine; |
28d8d7f4 | 34 | struct regexp; |
bbe252da | 35 | |
785a26d5 YO |
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 YO |
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 NC |
53 | typedef struct regexp_paren_pair { |
54 | I32 start; | |
55 | I32 end; | |
56 | } regexp_paren_pair; | |
28d8d7f4 | 57 | |
f8149455 YO |
58 | /* this is ordered such that the most commonly used |
59 | fields are at the start of the struct */ | |
bbe252da | 60 | typedef struct regexp { |
f8149455 YO |
61 | /* what engine created this regexp? */ |
62 | const struct regexp_engine* engine; | |
28d8d7f4 | 63 | struct regexp* mother_re; /* what re is this a lightweight copy of? */ |
f8149455 YO |
64 | |
65 | /* Information about the match that the perl core uses to manage things */ | |
bbe252da | 66 | U32 extflags; /* Flags used both externally and internally */ |
de8c5301 YO |
67 | I32 minlen; /* mininum possible length of string to match */ |
68 | I32 minlenret; /* mininum possible length of $& */ | |
58e23c8d | 69 | U32 gofs; /* chars left of pos that we search from */ |
f8149455 | 70 | struct reg_substr_data *substrs; /* substring data about strings that must appear |
bbe252da | 71 | in the final match, used for optimisations */ |
f8149455 | 72 | U32 nparens; /* number of capture buffers */ |
bbe252da | 73 | |
f8149455 YO |
74 | /* private engine specific data */ |
75 | U32 intflags; /* Engine Specific Internal flags */ | |
76 | void *pprivate; /* Data private to the regex engine which | |
77 | created this object. */ | |
78 | ||
79 | /* Data about the last/current match. These are modified during matching*/ | |
80 | U32 lastparen; /* last open paren matched */ | |
81 | U32 lastcloseparen; /* last close paren matched */ | |
f0ab9afb NC |
82 | regexp_paren_pair *swap; /* Swap copy of *offs */ |
83 | regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */ | |
28d8d7f4 | 84 | |
bbe252da YO |
85 | char *subbeg; /* saved or original string |
86 | so \digit works forever. */ | |
87 | I32 sublen; /* Length of string pointed by subbeg */ | |
bbe252da | 88 | SV_SAVED_COPY /* If non-NULL, SV which is COW from original */ |
f8149455 YO |
89 | |
90 | ||
91 | /* Information about the match that isn't often used */ | |
92 | char *precomp; /* pre-compilation regular expression */ | |
93 | I32 prelen; /* length of precomp */ | |
bcdf7404 YO |
94 | char *wrapped; /* wrapped version of the pattern */ |
95 | I32 wraplen; /* length of wrapped */ | |
f8149455 YO |
96 | I32 seen_evals; /* number of eval groups in the pattern - for security checks */ |
97 | HV *paren_names; /* Optional hash of paren names */ | |
98 | ||
99 | /* Refcount of this regexp */ | |
100 | I32 refcnt; /* Refcount of this regexp */ | |
f8fc2ecf YO |
101 | } regexp; |
102 | ||
785a26d5 | 103 | /* used for high speed searches */ |
f9f4320a YO |
104 | typedef struct re_scream_pos_data_s |
105 | { | |
106 | char **scream_olds; /* match pos */ | |
107 | I32 *scream_pos; /* Internal iterator of scream. */ | |
108 | } re_scream_pos_data; | |
109 | ||
785a26d5 YO |
110 | /* regexp_engine structure. This is the dispatch table for regexes. |
111 | * Any regex engine implementation must be able to build one of these. | |
112 | */ | |
f9f4320a | 113 | typedef struct regexp_engine { |
f2f78491 YO |
114 | regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm); |
115 | I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend, | |
f9f4320a YO |
116 | char* strbeg, I32 minend, SV* screamer, |
117 | void* data, U32 flags); | |
f2f78491 YO |
118 | char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos, |
119 | char *strend, U32 flags, | |
120 | struct re_scream_pos_data_s *data); | |
121 | SV* (*checkstr) (pTHX_ regexp *prog); | |
122 | void (*free) (pTHX_ struct regexp* r); | |
93b32b6d YO |
123 | SV* (*numbered_buff_get) (pTHX_ const REGEXP * const rx, I32 paren, SV* usesv); |
124 | SV* (*named_buff_get)(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags); | |
f2f78491 | 125 | #ifdef USE_ITHREADS |
f8149455 | 126 | void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param); |
f9f4320a YO |
127 | #endif |
128 | } regexp_engine; | |
129 | ||
bbe252da YO |
130 | /* Flags stored in regexp->extflags |
131 | * These are used by code external to the regexp engine | |
e357fc67 YO |
132 | * |
133 | * Note that flags starting with RXf_PMf_ have exact equivalents | |
134 | * stored in op_pmflags and which are defined in op.h, they are defined | |
135 | * numerically here only for clarity. | |
bbe252da YO |
136 | */ |
137 | ||
138 | /* Anchor and GPOS related stuff */ | |
139 | #define RXf_ANCH_BOL 0x00000001 | |
140 | #define RXf_ANCH_MBOL 0x00000002 | |
141 | #define RXf_ANCH_SBOL 0x00000004 | |
142 | #define RXf_ANCH_GPOS 0x00000008 | |
143 | #define RXf_GPOS_SEEN 0x00000010 | |
144 | #define RXf_GPOS_FLOAT 0x00000020 | |
e357fc67 | 145 | /* two bits here */ |
bbe252da YO |
146 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) |
147 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) | |
e357fc67 YO |
148 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) |
149 | ||
150 | /* Flags indicating special patterns */ | |
151 | #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ | |
152 | #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ | |
153 | ||
bcdf7404 | 154 | /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */ |
e357fc67 YO |
155 | #define RXf_PMf_LOCALE 0x00000800 /* use locale */ |
156 | #define RXf_PMf_MULTILINE 0x00001000 /* /m */ | |
157 | #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ | |
158 | #define RXf_PMf_FOLD 0x00004000 /* /i */ | |
159 | #define RXf_PMf_EXTENDED 0x00008000 /* /x */ | |
cde0cee5 | 160 | #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */ |
e357fc67 | 161 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
cde0cee5 YO |
162 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
163 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) | |
164 | ||
a20207d7 YO |
165 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
166 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ | |
167 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ | |
168 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ | |
169 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break | |
bbe252da | 170 | |
bcdf7404 YO |
171 | /* chars and strings used as regex pattern modifiers |
172 | * Singlular is a 'c'har, plural is a "string" | |
87e95b7f YO |
173 | * |
174 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve | |
175 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) | |
176 | * for its own uses. So 'k' is out as well. | |
bcdf7404 YO |
177 | */ |
178 | #define EXEC_PAT_MOD 'e' | |
87e95b7f | 179 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 YO |
180 | #define ONCE_PAT_MOD 'o' |
181 | #define GLOBAL_PAT_MOD 'g' | |
182 | #define CONTINUE_PAT_MOD 'c' | |
183 | #define MULTILINE_PAT_MOD 'm' | |
184 | #define SINGLE_PAT_MOD 's' | |
185 | #define IGNORE_PAT_MOD 'i' | |
186 | #define XTENDED_PAT_MOD 'x' | |
187 | ||
188 | #define ONCE_PAT_MODS "o" | |
87e95b7f | 189 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 YO |
190 | #define EXEC_PAT_MODS "e" |
191 | #define LOOP_PAT_MODS "gc" | |
192 | ||
193 | #define STD_PAT_MODS "msix" | |
194 | ||
195 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS | |
196 | ||
197 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS | |
198 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS | |
199 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS | |
200 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS | |
201 | ||
202 | ||
bbe252da | 203 | /* What we have seen */ |
bbe252da YO |
204 | #define RXf_LOOKBEHIND_SEEN 0x00020000 |
205 | #define RXf_EVAL_SEEN 0x00040000 | |
206 | #define RXf_CANY_SEEN 0x00080000 | |
207 | ||
208 | /* Special */ | |
209 | #define RXf_NOSCAN 0x00100000 | |
210 | #define RXf_CHECK_ALL 0x00200000 | |
211 | ||
212 | /* UTF8 related */ | |
213 | #define RXf_UTF8 0x00400000 | |
214 | #define RXf_MATCH_UTF8 0x00800000 | |
215 | ||
216 | /* Intuit related */ | |
217 | #define RXf_USE_INTUIT_NOML 0x01000000 | |
218 | #define RXf_USE_INTUIT_ML 0x02000000 | |
219 | #define RXf_INTUIT_TAIL 0x04000000 | |
220 | /* one bit here */ | |
221 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) | |
222 | ||
223 | /* Copy and tainted info */ | |
224 | #define RXf_COPY_DONE 0x10000000 | |
225 | #define RXf_TAINTED_SEEN 0x20000000 | |
226 | /* two bits here */ | |
227 | ||
bbe252da YO |
228 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
229 | #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN) | |
230 | #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN) | |
231 | #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN) | |
72311751 GS |
232 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
233 | ? RX_MATCH_TAINTED_on(prog) \ | |
234 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 235 | |
bbe252da YO |
236 | #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE) |
237 | #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE) | |
238 | #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE) | |
cf93c79d IZ |
239 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
240 | ? RX_MATCH_COPIED_on(prog) \ | |
241 | : RX_MATCH_COPIED_off(prog)) | |
f9f4320a | 242 | |
be8e71aa YO |
243 | #endif /* PLUGGABLE_RE_EXTENSION */ |
244 | ||
245 | /* Stuff that needs to be included in the plugable extension goes below here */ | |
246 | ||
f8c7b90f | 247 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
248 | #define RX_MATCH_COPY_FREE(rx) \ |
249 | STMT_START {if (rx->saved_copy) { \ | |
250 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ | |
251 | } \ | |
252 | if (RX_MATCH_COPIED(rx)) { \ | |
253 | Safefree(rx->subbeg); \ | |
254 | RX_MATCH_COPIED_off(rx); \ | |
255 | }} STMT_END | |
256 | #else | |
257 | #define RX_MATCH_COPY_FREE(rx) \ | |
258 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ | |
259 | Safefree(rx->subbeg); \ | |
260 | RX_MATCH_COPIED_off(rx); \ | |
261 | }} STMT_END | |
262 | #endif | |
263 | ||
bbe252da YO |
264 | #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8) |
265 | #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8) | |
266 | #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8) | |
a30b2f1f RGS |
267 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
268 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
269 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
270 | ||
cad2e5aa JH |
271 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
272 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
273 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
274 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 275 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 | 276 | |
155aba94 | 277 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a | 278 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d IZ |
279 | |
280 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
281 | #define FBMcf_TAIL_DOLLARM 2 |
282 | #define FBMcf_TAIL_Z 4 | |
283 | #define FBMcf_TAIL_z 8 | |
284 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
285 | |
286 | #define FBMrf_MULTILINE 1 | |
cad2e5aa | 287 | |
5d9a96ca DM |
288 | /* an accepting state/position*/ |
289 | struct _reg_trie_accepted { | |
290 | U8 *endpos; | |
291 | U16 wordnum; | |
292 | }; | |
293 | typedef struct _reg_trie_accepted reg_trie_accepted; | |
294 | ||
3b0527fe DM |
295 | /* some basic information about the current match that is created by |
296 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ | |
297 | ||
298 | typedef struct { | |
299 | regexp *prog; | |
300 | char *bol; | |
301 | char *till; | |
302 | SV *sv; | |
303 | char *ganch; | |
24b23f37 | 304 | char *cutpoint; |
3b0527fe DM |
305 | } regmatch_info; |
306 | ||
5d9a96ca DM |
307 | |
308 | /* structures for holding and saving the state maintained by regmatch() */ | |
309 | ||
d56b3014 | 310 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 | 311 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 | 312 | #endif |
6bda09f9 | 313 | |
5d9a96ca DM |
314 | typedef I32 CHECKPOINT; |
315 | ||
a0374537 | 316 | typedef struct regmatch_state { |
40a82448 | 317 | int resume_state; /* where to jump to on return */ |
24d3c4a9 | 318 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 DM |
319 | |
320 | union { | |
77cb431f DM |
321 | |
322 | /* this is a fake union member that matches the first element | |
323 | * of each member that needs to store positive backtrack | |
324 | * information */ | |
325 | struct { | |
326 | struct regmatch_state *prev_yes_state; | |
327 | } yes; | |
328 | ||
fae667d5 YO |
329 | /* branchlike members */ |
330 | /* this is a fake union member that matches the first elements | |
331 | * of each member that needs to behave like a branch */ | |
332 | struct { | |
5d458dd8 YO |
333 | /* this first element must match u.yes */ |
334 | struct regmatch_state *prev_yes_state; | |
fae667d5 YO |
335 | U32 lastparen; |
336 | CHECKPOINT cp; | |
337 | ||
338 | } branchlike; | |
339 | ||
340 | struct { | |
341 | /* the first elements must match u.branchlike */ | |
342 | struct regmatch_state *prev_yes_state; | |
343 | U32 lastparen; | |
344 | CHECKPOINT cp; | |
345 | ||
346 | regnode *next_branch; /* next branch node */ | |
347 | } branch; | |
348 | ||
349 | struct { | |
350 | /* the first elements must match u.branchlike */ | |
351 | struct regmatch_state *prev_yes_state; | |
352 | U32 lastparen; | |
353 | CHECKPOINT cp; | |
354 | ||
355 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ | |
166ba7cd | 356 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c | 357 | U16 *jump; /* positive offsets from me */ |
166ba7cd | 358 | regnode *B; /* node following the trie */ |
7f69552c | 359 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 DM |
360 | } trie; |
361 | ||
fae667d5 YO |
362 | /* special types - these members are used to store state for special |
363 | regops like eval, if/then, lookaround and the markpoint state */ | |
e822a8b4 | 364 | struct { |
77cb431f DM |
365 | /* this first element must match u.yes */ |
366 | struct regmatch_state *prev_yes_state; | |
faec1544 DM |
367 | struct regmatch_state *prev_eval; |
368 | struct regmatch_state *prev_curlyx; | |
aa283a38 | 369 | regexp *prev_rex; |
faec1544 DM |
370 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
371 | flip when transitioning between | |
372 | inner and outer rexen */ | |
aa283a38 DM |
373 | CHECKPOINT cp; /* remember current savestack indexes */ |
374 | CHECKPOINT lastcp; | |
40a82448 | 375 | regnode *B; /* the node following us */ |
6bda09f9 | 376 | U32 close_paren; /* which close bracket is our end */ |
e822a8b4 DM |
377 | } eval; |
378 | ||
379 | struct { | |
c476f425 | 380 | /* this first element must match u.yes */ |
262b90c4 | 381 | struct regmatch_state *prev_yes_state; |
fae667d5 YO |
382 | I32 wanted; |
383 | I32 logical; /* saved copy of 'logical' var */ | |
384 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ | |
385 | } ifmatch; /* and SUSPEND/UNLESSM */ | |
386 | ||
387 | struct { | |
388 | /* this first element must match u.yes */ | |
389 | struct regmatch_state *prev_yes_state; | |
390 | struct regmatch_state *prev_mark; | |
391 | SV* mark_name; | |
392 | char *mark_loc; | |
393 | } mark; | |
394 | ||
395 | struct { | |
396 | int val; | |
397 | } keeper; | |
398 | ||
399 | /* quantifiers - these members are used for storing state for | |
400 | for the regops used to implement quantifiers */ | |
401 | struct { | |
402 | /* this first element must match u.yes */ | |
403 | struct regmatch_state *prev_yes_state; | |
c476f425 DM |
404 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
405 | CHECKPOINT cp; /* remember current savestack index */ | |
406 | bool minmod; | |
a0374537 | 407 | int parenfloor;/* how far back to strip paren data */ |
c476f425 DM |
408 | int min; /* the minimal number of A's to match */ |
409 | int max; /* the maximal number of A's to match */ | |
410 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ | |
411 | ||
412 | /* these two are modified by WHILEM */ | |
413 | int count; /* how many instances of A we've matched */ | |
414 | char *lastloc;/* where previous A matched (0-len detect) */ | |
e822a8b4 DM |
415 | } curlyx; |
416 | ||
417 | struct { | |
c476f425 | 418 | /* this first element must match u.yes */ |
262b90c4 | 419 | struct regmatch_state *prev_yes_state; |
c476f425 DM |
420 | struct regmatch_state *save_curlyx; |
421 | CHECKPOINT cp; /* remember current savestack indexes */ | |
422 | CHECKPOINT lastcp; | |
423 | char *save_lastloc; /* previous curlyx.lastloc */ | |
424 | I32 cache_offset; | |
425 | I32 cache_mask; | |
e822a8b4 DM |
426 | } whilem; |
427 | ||
428 | struct { | |
5d458dd8 YO |
429 | /* this first element must match u.yes */ |
430 | struct regmatch_state *prev_yes_state; | |
e822a8b4 | 431 | I32 c1, c2; /* case fold search */ |
40a82448 DM |
432 | CHECKPOINT cp; |
433 | I32 alen; /* length of first-matched A string */ | |
434 | I32 count; | |
0cadcf80 | 435 | bool minmod; |
40a82448 DM |
436 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
437 | regnode *me; /* the curlym node */ | |
e822a8b4 DM |
438 | } curlym; |
439 | ||
440 | struct { | |
3b6647e0 | 441 | U32 paren; |
c255a977 | 442 | CHECKPOINT cp; |
e822a8b4 | 443 | I32 c1, c2; /* case fold search */ |
c255a977 DM |
444 | char *maxpos; /* highest possible point in string to match */ |
445 | char *oldloc; /* the previous locinput */ | |
e822a8b4 | 446 | int count; |
c255a977 DM |
447 | int min, max; /* {m,n} */ |
448 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ | |
449 | } curly; /* and CURLYN/PLUS/STAR */ | |
dad79028 | 450 | |
d8319b27 | 451 | } u; |
5d9a96ca DM |
452 | } regmatch_state; |
453 | ||
454 | /* how many regmatch_state structs to allocate as a single slab. | |
455 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev | |
456 | * pointers, plus 1 for any mythical malloc overhead. */ | |
457 | ||
458 | #define PERL_REGMATCH_SLAB_SLOTS \ | |
459 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) | |
460 | ||
461 | typedef struct regmatch_slab { | |
462 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; | |
463 | struct regmatch_slab *prev, *next; | |
464 | } regmatch_slab; | |
1ade1aa1 | 465 | |
46ab3289 NC |
466 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
467 | #define PL_bostr PL_reg_state.re_state_bostr | |
468 | #define PL_reginput PL_reg_state.re_state_reginput | |
46ab3289 | 469 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb | 470 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 NC |
471 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
472 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen | |
46ab3289 NC |
473 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
474 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl | |
475 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set | |
46ab3289 NC |
476 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
477 | #define PL_reg_magic PL_reg_state.re_state_reg_magic | |
478 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos | |
479 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm | |
480 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm | |
481 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved | |
482 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen | |
483 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter | |
484 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter | |
485 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache | |
486 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size | |
487 | #define PL_regsize PL_reg_state.re_state_regsize | |
488 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry | |
489 | #define PL_nrs PL_reg_state.re_state_nrs | |
490 | ||
1ade1aa1 NC |
491 | struct re_save_state { |
492 | U32 re_state_reg_flags; /* from regexec.c */ | |
493 | char *re_state_bostr; | |
494 | char *re_state_reginput; /* String-input pointer. */ | |
1ade1aa1 | 495 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb | 496 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 NC |
497 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
498 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ | |
1ade1aa1 NC |
499 | char **re_state_reg_start_tmp; /* from regexec.c */ |
500 | U32 re_state_reg_start_tmpl; /* from regexec.c */ | |
501 | I32 re_state_reg_eval_set; /* from regexec.c */ | |
1ade1aa1 NC |
502 | bool re_state_reg_match_utf8; /* from regexec.c */ |
503 | MAGIC *re_state_reg_magic; /* from regexec.c */ | |
504 | I32 re_state_reg_oldpos; /* from regexec.c */ | |
505 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ | |
506 | PMOP *re_state_reg_curpm; /* from regexec.c */ | |
507 | char *re_state_reg_oldsaved; /* old saved substr during match */ | |
508 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ | |
509 | I32 re_state_reg_maxiter; /* max wait until caching pos */ | |
510 | I32 re_state_reg_leftiter; /* wait until caching pos */ | |
511 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ | |
512 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ | |
3b6647e0 | 513 | U32 re_state_regsize; /* from regexec.c */ |
1ade1aa1 NC |
514 | char *re_state_reg_starttry; /* from regexec.c */ |
515 | #ifdef PERL_OLD_COPY_ON_WRITE | |
46ab3289 | 516 | 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 |
517 | #endif |
518 | }; | |
519 | ||
520 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ | |
521 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) | |
cde0cee5 | 522 | |
1ade1aa1 NC |
523 | /* |
524 | * Local variables: | |
525 | * c-indentation-style: bsd | |
526 | * c-basic-offset: 4 | |
527 | * indent-tabs-mode: t | |
528 | * End: | |
529 | * | |
530 | * ex: set ts=8 sts=4 sw=4 noet: | |
531 | */ |