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 | |
882227b7 AB |
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 | 69 | typedef struct regexp { |
f8149455 YO |
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 YO |
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 YO |
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 YO |
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 NC |
91 | regexp_paren_pair *swap; /* Swap copy of *offs */ |
92 | regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */ | |
28d8d7f4 | 93 | |
bbe252da YO |
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 YO |
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 YO |
104 | char *wrapped; /* wrapped version of the pattern */ |
105 | I32 wraplen; /* length of wrapped */ | |
f8149455 YO |
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 YO |
111 | } regexp; |
112 | ||
785a26d5 | 113 | /* used for high speed searches */ |
f9f4320a YO |
114 | typedef 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 YO |
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 | 123 | typedef struct regexp_engine { |
3ab4a224 | 124 | REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags); |
49d7dfbc AB |
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 YO |
129 | char *strend, const U32 flags, |
130 | re_scream_pos_data *data); | |
49d7dfbc AB |
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 AB |
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 AB |
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 YO |
147 | } regexp_engine; |
148 | ||
192b9cd1 AB |
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 | */ | |
f1b875a0 YO |
154 | #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ |
155 | #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ | |
156 | #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ | |
192b9cd1 AB |
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 */ | |
f1b875a0 YO |
166 | #define RXapif_FETCH 0x0001 |
167 | #define RXapif_STORE 0x0002 | |
168 | #define RXapif_DELETE 0x0004 | |
169 | #define RXapif_CLEAR 0x0008 | |
170 | #define RXapif_EXISTS 0x0010 | |
171 | #define RXapif_SCALAR 0x0020 | |
172 | #define RXapif_FIRSTKEY 0x0040 | |
173 | #define RXapif_NEXTKEY 0x0080 | |
192b9cd1 AB |
174 | |
175 | /* Whether %+ or %- is being operated on */ | |
f1b875a0 YO |
176 | #define RXapif_ONE 0x0100 /* %+ */ |
177 | #define RXapif_ALL 0x0200 /* %- */ | |
192b9cd1 AB |
178 | |
179 | /* Whether this is being called from a re:: function */ | |
f1b875a0 YO |
180 | #define RXapif_REGNAME 0x0400 |
181 | #define RXapif_REGNAMES 0x0800 | |
182 | #define RXapif_REGNAMES_COUNT 0x1000 | |
192b9cd1 | 183 | |
f7e71195 AB |
184 | /* |
185 | =head1 REGEXP Functions | |
186 | ||
187 | =for apidoc Am|REGEXP *|SvRX|SV *sv | |
188 | ||
189 | Convenience macro to get the REGEXP from a SV. This is approximately | |
190 | equivalent to the following snippet: | |
191 | ||
192 | if (SvMAGICAL(sv)) | |
193 | mg_get(sv); | |
194 | if (SvROK(sv) && | |
195 | (tmpsv = (SV*)SvRV(sv)) && | |
196 | SvTYPE(tmpsv) == SVt_PVMG && | |
197 | (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr))) | |
198 | { | |
199 | return (REGEXP *)tmpmg->mg_obj; | |
200 | } | |
201 | ||
202 | NULL will be returned if a REGEXP* is not found. | |
203 | ||
204 | =for apidoc Am|bool|SvRXOK|SV* sv | |
205 | ||
206 | Returns a boolean indicating whether the SV contains qr magic | |
207 | (PERL_MAGIC_qr). | |
208 | ||
209 | If you want to do something with the REGEXP* later use SvRX instead | |
210 | and check for NULL. | |
211 | ||
212 | =cut | |
213 | */ | |
214 | ||
215 | #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) | |
216 | #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) | |
217 | ||
218 | ||
bbe252da YO |
219 | /* Flags stored in regexp->extflags |
220 | * These are used by code external to the regexp engine | |
e357fc67 YO |
221 | * |
222 | * Note that flags starting with RXf_PMf_ have exact equivalents | |
223 | * stored in op_pmflags and which are defined in op.h, they are defined | |
224 | * numerically here only for clarity. | |
bbe252da YO |
225 | */ |
226 | ||
227 | /* Anchor and GPOS related stuff */ | |
228 | #define RXf_ANCH_BOL 0x00000001 | |
229 | #define RXf_ANCH_MBOL 0x00000002 | |
230 | #define RXf_ANCH_SBOL 0x00000004 | |
231 | #define RXf_ANCH_GPOS 0x00000008 | |
232 | #define RXf_GPOS_SEEN 0x00000010 | |
233 | #define RXf_GPOS_FLOAT 0x00000020 | |
e357fc67 | 234 | /* two bits here */ |
bbe252da YO |
235 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) |
236 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) | |
e357fc67 YO |
237 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) |
238 | ||
239 | /* Flags indicating special patterns */ | |
c737faaf | 240 | #define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */ |
e357fc67 YO |
241 | #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ |
242 | #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ | |
243 | ||
bcdf7404 | 244 | /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */ |
e357fc67 YO |
245 | #define RXf_PMf_LOCALE 0x00000800 /* use locale */ |
246 | #define RXf_PMf_MULTILINE 0x00001000 /* /m */ | |
247 | #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ | |
248 | #define RXf_PMf_FOLD 0x00004000 /* /i */ | |
249 | #define RXf_PMf_EXTENDED 0x00008000 /* /x */ | |
f7819f85 | 250 | #define RXf_PMf_KEEPCOPY 0x00010000 /* /p */ |
e357fc67 | 251 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
cde0cee5 YO |
252 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
253 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) | |
254 | ||
a20207d7 YO |
255 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
256 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ | |
257 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ | |
258 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ | |
259 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break | |
bbe252da | 260 | |
bcdf7404 YO |
261 | /* chars and strings used as regex pattern modifiers |
262 | * Singlular is a 'c'har, plural is a "string" | |
87e95b7f YO |
263 | * |
264 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve | |
265 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) | |
266 | * for its own uses. So 'k' is out as well. | |
bcdf7404 YO |
267 | */ |
268 | #define EXEC_PAT_MOD 'e' | |
87e95b7f | 269 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 YO |
270 | #define ONCE_PAT_MOD 'o' |
271 | #define GLOBAL_PAT_MOD 'g' | |
272 | #define CONTINUE_PAT_MOD 'c' | |
273 | #define MULTILINE_PAT_MOD 'm' | |
274 | #define SINGLE_PAT_MOD 's' | |
275 | #define IGNORE_PAT_MOD 'i' | |
276 | #define XTENDED_PAT_MOD 'x' | |
277 | ||
278 | #define ONCE_PAT_MODS "o" | |
87e95b7f | 279 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 YO |
280 | #define EXEC_PAT_MODS "e" |
281 | #define LOOP_PAT_MODS "gc" | |
282 | ||
283 | #define STD_PAT_MODS "msix" | |
284 | ||
285 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS | |
286 | ||
287 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS | |
288 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS | |
289 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS | |
290 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS | |
291 | ||
292 | ||
bbe252da | 293 | /* What we have seen */ |
bbe252da YO |
294 | #define RXf_LOOKBEHIND_SEEN 0x00020000 |
295 | #define RXf_EVAL_SEEN 0x00040000 | |
296 | #define RXf_CANY_SEEN 0x00080000 | |
297 | ||
298 | /* Special */ | |
299 | #define RXf_NOSCAN 0x00100000 | |
300 | #define RXf_CHECK_ALL 0x00200000 | |
301 | ||
302 | /* UTF8 related */ | |
303 | #define RXf_UTF8 0x00400000 | |
995b21a2 | 304 | #define RXf_MATCH_UTF8 0x00800000 |
bbe252da YO |
305 | |
306 | /* Intuit related */ | |
307 | #define RXf_USE_INTUIT_NOML 0x01000000 | |
308 | #define RXf_USE_INTUIT_ML 0x02000000 | |
309 | #define RXf_INTUIT_TAIL 0x04000000 | |
0ac6acae AB |
310 | |
311 | /* | |
312 | Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will | |
313 | be used by regex engines to check whether they should set | |
314 | RXf_SKIPWHITE | |
315 | */ | |
316 | #define RXf_SPLIT 0x08000000 | |
317 | ||
bbe252da YO |
318 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
319 | ||
320 | /* Copy and tainted info */ | |
321 | #define RXf_COPY_DONE 0x10000000 | |
322 | #define RXf_TAINTED_SEEN 0x20000000 | |
c737faaf YO |
323 | #define RXf_TAINTED 0x80000000 /* this pattern is tainted */ |
324 | ||
bbe252da | 325 | |
bbe252da YO |
326 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
327 | #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN) | |
328 | #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN) | |
329 | #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN) | |
72311751 GS |
330 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
331 | ? RX_MATCH_TAINTED_on(prog) \ | |
332 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 333 | |
bbe252da YO |
334 | #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE) |
335 | #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE) | |
336 | #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE) | |
cf93c79d IZ |
337 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
338 | ? RX_MATCH_COPIED_on(prog) \ | |
339 | : RX_MATCH_COPIED_off(prog)) | |
f9f4320a | 340 | |
be8e71aa YO |
341 | #endif /* PLUGGABLE_RE_EXTENSION */ |
342 | ||
343 | /* Stuff that needs to be included in the plugable extension goes below here */ | |
344 | ||
f8c7b90f | 345 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
346 | #define RX_MATCH_COPY_FREE(rx) \ |
347 | STMT_START {if (rx->saved_copy) { \ | |
348 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ | |
349 | } \ | |
350 | if (RX_MATCH_COPIED(rx)) { \ | |
351 | Safefree(rx->subbeg); \ | |
352 | RX_MATCH_COPIED_off(rx); \ | |
353 | }} STMT_END | |
354 | #else | |
355 | #define RX_MATCH_COPY_FREE(rx) \ | |
356 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ | |
357 | Safefree(rx->subbeg); \ | |
358 | RX_MATCH_COPIED_off(rx); \ | |
359 | }} STMT_END | |
360 | #endif | |
361 | ||
995b21a2 YO |
362 | #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8) |
363 | #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8) | |
364 | #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8) | |
a30b2f1f RGS |
365 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
366 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
367 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
368 | ||
cad2e5aa JH |
369 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
370 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
371 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
372 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 373 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 | 374 | |
155aba94 | 375 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a | 376 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d IZ |
377 | |
378 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
379 | #define FBMcf_TAIL_DOLLARM 2 |
380 | #define FBMcf_TAIL_Z 4 | |
381 | #define FBMcf_TAIL_z 8 | |
382 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
383 | |
384 | #define FBMrf_MULTILINE 1 | |
cad2e5aa | 385 | |
5d9a96ca DM |
386 | /* an accepting state/position*/ |
387 | struct _reg_trie_accepted { | |
388 | U8 *endpos; | |
389 | U16 wordnum; | |
390 | }; | |
391 | typedef struct _reg_trie_accepted reg_trie_accepted; | |
392 | ||
3b0527fe DM |
393 | /* some basic information about the current match that is created by |
394 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ | |
395 | ||
396 | typedef struct { | |
397 | regexp *prog; | |
398 | char *bol; | |
399 | char *till; | |
400 | SV *sv; | |
401 | char *ganch; | |
24b23f37 | 402 | char *cutpoint; |
3b0527fe DM |
403 | } regmatch_info; |
404 | ||
5d9a96ca DM |
405 | |
406 | /* structures for holding and saving the state maintained by regmatch() */ | |
407 | ||
d56b3014 | 408 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 | 409 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 | 410 | #endif |
6bda09f9 | 411 | |
5d9a96ca DM |
412 | typedef I32 CHECKPOINT; |
413 | ||
a0374537 | 414 | typedef struct regmatch_state { |
40a82448 | 415 | int resume_state; /* where to jump to on return */ |
24d3c4a9 | 416 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 DM |
417 | |
418 | union { | |
77cb431f DM |
419 | |
420 | /* this is a fake union member that matches the first element | |
421 | * of each member that needs to store positive backtrack | |
422 | * information */ | |
423 | struct { | |
424 | struct regmatch_state *prev_yes_state; | |
425 | } yes; | |
426 | ||
fae667d5 YO |
427 | /* branchlike members */ |
428 | /* this is a fake union member that matches the first elements | |
429 | * of each member that needs to behave like a branch */ | |
430 | struct { | |
5d458dd8 YO |
431 | /* this first element must match u.yes */ |
432 | struct regmatch_state *prev_yes_state; | |
fae667d5 YO |
433 | U32 lastparen; |
434 | CHECKPOINT cp; | |
435 | ||
436 | } branchlike; | |
437 | ||
438 | struct { | |
439 | /* the first elements must match u.branchlike */ | |
440 | struct regmatch_state *prev_yes_state; | |
441 | U32 lastparen; | |
442 | CHECKPOINT cp; | |
443 | ||
444 | regnode *next_branch; /* next branch node */ | |
445 | } branch; | |
446 | ||
447 | struct { | |
448 | /* the first elements must match u.branchlike */ | |
449 | struct regmatch_state *prev_yes_state; | |
450 | U32 lastparen; | |
451 | CHECKPOINT cp; | |
452 | ||
453 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ | |
166ba7cd | 454 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c | 455 | U16 *jump; /* positive offsets from me */ |
166ba7cd | 456 | regnode *B; /* node following the trie */ |
7f69552c | 457 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 DM |
458 | } trie; |
459 | ||
fae667d5 YO |
460 | /* special types - these members are used to store state for special |
461 | regops like eval, if/then, lookaround and the markpoint state */ | |
e822a8b4 | 462 | struct { |
77cb431f DM |
463 | /* this first element must match u.yes */ |
464 | struct regmatch_state *prev_yes_state; | |
faec1544 DM |
465 | struct regmatch_state *prev_eval; |
466 | struct regmatch_state *prev_curlyx; | |
aa283a38 | 467 | regexp *prev_rex; |
faec1544 DM |
468 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
469 | flip when transitioning between | |
470 | inner and outer rexen */ | |
aa283a38 DM |
471 | CHECKPOINT cp; /* remember current savestack indexes */ |
472 | CHECKPOINT lastcp; | |
6bda09f9 | 473 | U32 close_paren; /* which close bracket is our end */ |
1b5c067c | 474 | regnode *B; /* the node following us */ |
e822a8b4 DM |
475 | } eval; |
476 | ||
477 | struct { | |
c476f425 | 478 | /* this first element must match u.yes */ |
262b90c4 | 479 | struct regmatch_state *prev_yes_state; |
fae667d5 YO |
480 | I32 wanted; |
481 | I32 logical; /* saved copy of 'logical' var */ | |
482 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ | |
483 | } ifmatch; /* and SUSPEND/UNLESSM */ | |
484 | ||
485 | struct { | |
486 | /* this first element must match u.yes */ | |
487 | struct regmatch_state *prev_yes_state; | |
488 | struct regmatch_state *prev_mark; | |
489 | SV* mark_name; | |
490 | char *mark_loc; | |
491 | } mark; | |
492 | ||
493 | struct { | |
494 | int val; | |
495 | } keeper; | |
496 | ||
497 | /* quantifiers - these members are used for storing state for | |
498 | for the regops used to implement quantifiers */ | |
499 | struct { | |
500 | /* this first element must match u.yes */ | |
501 | struct regmatch_state *prev_yes_state; | |
c476f425 | 502 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
1b5c067c | 503 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
c476f425 DM |
504 | CHECKPOINT cp; /* remember current savestack index */ |
505 | bool minmod; | |
a0374537 | 506 | int parenfloor;/* how far back to strip paren data */ |
c476f425 DM |
507 | int min; /* the minimal number of A's to match */ |
508 | int max; /* the maximal number of A's to match */ | |
c476f425 DM |
509 | |
510 | /* these two are modified by WHILEM */ | |
511 | int count; /* how many instances of A we've matched */ | |
512 | char *lastloc;/* where previous A matched (0-len detect) */ | |
e822a8b4 DM |
513 | } curlyx; |
514 | ||
515 | struct { | |
c476f425 | 516 | /* this first element must match u.yes */ |
262b90c4 | 517 | struct regmatch_state *prev_yes_state; |
c476f425 DM |
518 | struct regmatch_state *save_curlyx; |
519 | CHECKPOINT cp; /* remember current savestack indexes */ | |
520 | CHECKPOINT lastcp; | |
521 | char *save_lastloc; /* previous curlyx.lastloc */ | |
522 | I32 cache_offset; | |
523 | I32 cache_mask; | |
e822a8b4 DM |
524 | } whilem; |
525 | ||
526 | struct { | |
5d458dd8 YO |
527 | /* this first element must match u.yes */ |
528 | struct regmatch_state *prev_yes_state; | |
e822a8b4 | 529 | I32 c1, c2; /* case fold search */ |
40a82448 DM |
530 | CHECKPOINT cp; |
531 | I32 alen; /* length of first-matched A string */ | |
532 | I32 count; | |
0cadcf80 | 533 | bool minmod; |
40a82448 DM |
534 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
535 | regnode *me; /* the curlym node */ | |
e822a8b4 DM |
536 | } curlym; |
537 | ||
538 | struct { | |
3b6647e0 | 539 | U32 paren; |
c255a977 | 540 | CHECKPOINT cp; |
e822a8b4 | 541 | I32 c1, c2; /* case fold search */ |
c255a977 DM |
542 | char *maxpos; /* highest possible point in string to match */ |
543 | char *oldloc; /* the previous locinput */ | |
e822a8b4 | 544 | int count; |
c255a977 DM |
545 | int min, max; /* {m,n} */ |
546 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ | |
547 | } curly; /* and CURLYN/PLUS/STAR */ | |
dad79028 | 548 | |
d8319b27 | 549 | } u; |
5d9a96ca DM |
550 | } regmatch_state; |
551 | ||
552 | /* how many regmatch_state structs to allocate as a single slab. | |
553 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev | |
554 | * pointers, plus 1 for any mythical malloc overhead. */ | |
555 | ||
556 | #define PERL_REGMATCH_SLAB_SLOTS \ | |
557 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) | |
558 | ||
559 | typedef struct regmatch_slab { | |
560 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; | |
561 | struct regmatch_slab *prev, *next; | |
562 | } regmatch_slab; | |
1ade1aa1 | 563 | |
46ab3289 NC |
564 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
565 | #define PL_bostr PL_reg_state.re_state_bostr | |
566 | #define PL_reginput PL_reg_state.re_state_reginput | |
46ab3289 | 567 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb | 568 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 NC |
569 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
570 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen | |
46ab3289 NC |
571 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
572 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl | |
573 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set | |
46ab3289 NC |
574 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
575 | #define PL_reg_magic PL_reg_state.re_state_reg_magic | |
576 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos | |
577 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm | |
578 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm | |
579 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved | |
580 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen | |
581 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter | |
582 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter | |
583 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache | |
584 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size | |
585 | #define PL_regsize PL_reg_state.re_state_regsize | |
586 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry | |
587 | #define PL_nrs PL_reg_state.re_state_nrs | |
588 | ||
1ade1aa1 NC |
589 | struct re_save_state { |
590 | U32 re_state_reg_flags; /* from regexec.c */ | |
1b5c067c NC |
591 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
592 | I32 re_state_reg_eval_set; /* from regexec.c */ | |
593 | bool re_state_reg_match_utf8; /* from regexec.c */ | |
1ade1aa1 NC |
594 | char *re_state_bostr; |
595 | char *re_state_reginput; /* String-input pointer. */ | |
1ade1aa1 | 596 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb | 597 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 NC |
598 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
599 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ | |
1ade1aa1 | 600 | char **re_state_reg_start_tmp; /* from regexec.c */ |
1ade1aa1 | 601 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
1ade1aa1 NC |
602 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
603 | PMOP *re_state_reg_curpm; /* from regexec.c */ | |
604 | char *re_state_reg_oldsaved; /* old saved substr during match */ | |
605 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ | |
1b5c067c NC |
606 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
607 | I32 re_state_reg_oldpos; /* from regexec.c */ | |
1ade1aa1 NC |
608 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
609 | I32 re_state_reg_leftiter; /* wait until caching pos */ | |
3b6647e0 | 610 | U32 re_state_regsize; /* from regexec.c */ |
1b5c067c | 611 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
1ade1aa1 NC |
612 | char *re_state_reg_starttry; /* from regexec.c */ |
613 | #ifdef PERL_OLD_COPY_ON_WRITE | |
46ab3289 | 614 | 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 |
615 | #endif |
616 | }; | |
617 | ||
618 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ | |
619 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) | |
cde0cee5 | 620 | |
1ade1aa1 NC |
621 | /* |
622 | * Local variables: | |
623 | * c-indentation-style: bsd | |
624 | * c-basic-offset: 4 | |
625 | * indent-tabs-mode: t | |
626 | * End: | |
627 | * | |
628 | * ex: set ts=8 sts=4 sw=4 noet: | |
629 | */ |