Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* regexp.h |
d6376244 | 2 | * |
4bb101f2 | 3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
663f364b | 4 | * 2005, 2006, 2007, 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 */ | |
d7b33318 | 101 | /* wrapped can't be const char*, as it is returned by sv_2pv_flags */ |
bcdf7404 YO |
102 | char *wrapped; /* wrapped version of the pattern */ |
103 | I32 wraplen; /* length of wrapped */ | |
2c96c0d6 NC |
104 | unsigned pre_prefix:4; /* offset from wrapped to the start of precomp */ |
105 | unsigned seen_evals:28; /* number of eval groups in the pattern - for security checks */ | |
f8149455 YO |
106 | HV *paren_names; /* Optional hash of paren names */ |
107 | ||
108 | /* Refcount of this regexp */ | |
109 | I32 refcnt; /* Refcount of this regexp */ | |
f8fc2ecf YO |
110 | } regexp; |
111 | ||
785a26d5 | 112 | /* used for high speed searches */ |
f9f4320a YO |
113 | typedef struct re_scream_pos_data_s |
114 | { | |
115 | char **scream_olds; /* match pos */ | |
116 | I32 *scream_pos; /* Internal iterator of scream. */ | |
117 | } re_scream_pos_data; | |
118 | ||
785a26d5 YO |
119 | /* regexp_engine structure. This is the dispatch table for regexes. |
120 | * Any regex engine implementation must be able to build one of these. | |
121 | */ | |
f9f4320a | 122 | typedef struct regexp_engine { |
3ab4a224 | 123 | REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags); |
49d7dfbc AB |
124 | I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, |
125 | char* strbeg, I32 minend, SV* screamer, | |
126 | void* data, U32 flags); | |
127 | char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, | |
9f61653a YO |
128 | char *strend, const U32 flags, |
129 | re_scream_pos_data *data); | |
49d7dfbc AB |
130 | SV* (*checkstr) (pTHX_ REGEXP * const rx); |
131 | void (*free) (pTHX_ REGEXP * const rx); | |
2fdbfb4d | 132 | void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, |
192b9cd1 | 133 | SV * const sv); |
2fdbfb4d AB |
134 | void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, |
135 | SV const * const value); | |
136 | I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, | |
137 | const I32 paren); | |
192b9cd1 AB |
138 | SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, |
139 | SV * const value, const U32 flags); | |
140 | SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, | |
141 | const U32 flags); | |
49d7dfbc | 142 | SV* (*qr_package)(pTHX_ REGEXP * const rx); |
f2f78491 | 143 | #ifdef USE_ITHREADS |
49d7dfbc | 144 | void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); |
2fdbfb4d | 145 | #endif |
f9f4320a YO |
146 | } regexp_engine; |
147 | ||
192b9cd1 AB |
148 | /* |
149 | These are passed to the numbered capture variable callbacks as the | |
150 | paren name. >= 1 is reserved for actual numbered captures, i.e. $1, | |
151 | $2 etc. | |
152 | */ | |
f1b875a0 YO |
153 | #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ |
154 | #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ | |
155 | #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ | |
192b9cd1 AB |
156 | |
157 | /* | |
158 | Flags that are passed to the named_buff and named_buff_iter | |
159 | callbacks above. Those routines are called from universal.c via the | |
160 | Tie::Hash::NamedCapture interface for %+ and %- and the re:: | |
161 | functions in the same file. | |
162 | */ | |
163 | ||
164 | /* The Tie::Hash::NamedCapture operation this is part of, if any */ | |
f1b875a0 YO |
165 | #define RXapif_FETCH 0x0001 |
166 | #define RXapif_STORE 0x0002 | |
167 | #define RXapif_DELETE 0x0004 | |
168 | #define RXapif_CLEAR 0x0008 | |
169 | #define RXapif_EXISTS 0x0010 | |
170 | #define RXapif_SCALAR 0x0020 | |
171 | #define RXapif_FIRSTKEY 0x0040 | |
172 | #define RXapif_NEXTKEY 0x0080 | |
192b9cd1 AB |
173 | |
174 | /* Whether %+ or %- is being operated on */ | |
f1b875a0 YO |
175 | #define RXapif_ONE 0x0100 /* %+ */ |
176 | #define RXapif_ALL 0x0200 /* %- */ | |
192b9cd1 AB |
177 | |
178 | /* Whether this is being called from a re:: function */ | |
f1b875a0 YO |
179 | #define RXapif_REGNAME 0x0400 |
180 | #define RXapif_REGNAMES 0x0800 | |
181 | #define RXapif_REGNAMES_COUNT 0x1000 | |
192b9cd1 | 182 | |
f7e71195 AB |
183 | /* |
184 | =head1 REGEXP Functions | |
185 | ||
186 | =for apidoc Am|REGEXP *|SvRX|SV *sv | |
187 | ||
188 | Convenience macro to get the REGEXP from a SV. This is approximately | |
189 | equivalent to the following snippet: | |
190 | ||
191 | if (SvMAGICAL(sv)) | |
192 | mg_get(sv); | |
193 | if (SvROK(sv) && | |
194 | (tmpsv = (SV*)SvRV(sv)) && | |
195 | SvTYPE(tmpsv) == SVt_PVMG && | |
196 | (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr))) | |
197 | { | |
198 | return (REGEXP *)tmpmg->mg_obj; | |
199 | } | |
200 | ||
201 | NULL will be returned if a REGEXP* is not found. | |
202 | ||
203 | =for apidoc Am|bool|SvRXOK|SV* sv | |
204 | ||
205 | Returns a boolean indicating whether the SV contains qr magic | |
206 | (PERL_MAGIC_qr). | |
207 | ||
208 | If you want to do something with the REGEXP* later use SvRX instead | |
209 | and check for NULL. | |
210 | ||
211 | =cut | |
212 | */ | |
213 | ||
214 | #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) | |
215 | #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) | |
216 | ||
217 | ||
bbe252da YO |
218 | /* Flags stored in regexp->extflags |
219 | * These are used by code external to the regexp engine | |
e357fc67 YO |
220 | * |
221 | * Note that flags starting with RXf_PMf_ have exact equivalents | |
222 | * stored in op_pmflags and which are defined in op.h, they are defined | |
223 | * numerically here only for clarity. | |
cb5027f2 YO |
224 | * |
225 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
226 | * so that regnodes.h is updated with the changes. | |
227 | * | |
bbe252da YO |
228 | */ |
229 | ||
52d81aa8 | 230 | /* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME |
a1cf41cd NC |
231 | * If you change these you need to change the equivalent flags in op.h, and |
232 | * vice versa. */ | |
52d81aa8 NC |
233 | #define RXf_PMf_MULTILINE 0x00000001 /* /m */ |
234 | #define RXf_PMf_SINGLELINE 0x00000002 /* /s */ | |
235 | #define RXf_PMf_FOLD 0x00000004 /* /i */ | |
236 | #define RXf_PMf_EXTENDED 0x00000008 /* /x */ | |
237 | #define RXf_PMf_KEEPCOPY 0x00000010 /* /p */ | |
238 | #define RXf_PMf_LOCALE 0x00000020 /* use locale */ | |
e357fc67 | 239 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
52d81aa8 | 240 | #define RXf_PMf_STD_PMMOD_SHIFT 0 |
cde0cee5 YO |
241 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
242 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) | |
243 | ||
a20207d7 YO |
244 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
245 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ | |
246 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ | |
247 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ | |
248 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break | |
bbe252da | 249 | |
bcdf7404 YO |
250 | /* chars and strings used as regex pattern modifiers |
251 | * Singlular is a 'c'har, plural is a "string" | |
87e95b7f YO |
252 | * |
253 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve | |
254 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) | |
255 | * for its own uses. So 'k' is out as well. | |
bcdf7404 YO |
256 | */ |
257 | #define EXEC_PAT_MOD 'e' | |
87e95b7f | 258 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 YO |
259 | #define ONCE_PAT_MOD 'o' |
260 | #define GLOBAL_PAT_MOD 'g' | |
261 | #define CONTINUE_PAT_MOD 'c' | |
262 | #define MULTILINE_PAT_MOD 'm' | |
263 | #define SINGLE_PAT_MOD 's' | |
264 | #define IGNORE_PAT_MOD 'i' | |
265 | #define XTENDED_PAT_MOD 'x' | |
266 | ||
267 | #define ONCE_PAT_MODS "o" | |
87e95b7f | 268 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 YO |
269 | #define EXEC_PAT_MODS "e" |
270 | #define LOOP_PAT_MODS "gc" | |
271 | ||
272 | #define STD_PAT_MODS "msix" | |
273 | ||
274 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS | |
275 | ||
276 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS | |
277 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS | |
278 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS | |
279 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS | |
280 | ||
cb5027f2 YO |
281 | /* |
282 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
283 | * so that regnodes.h is updated with the changes. | |
284 | * | |
285 | */ | |
bcdf7404 | 286 | |
52d81aa8 NC |
287 | /* Anchor and GPOS related stuff */ |
288 | #define RXf_ANCH_BOL 0x00000100 | |
289 | #define RXf_ANCH_MBOL 0x00000200 | |
290 | #define RXf_ANCH_SBOL 0x00000400 | |
291 | #define RXf_ANCH_GPOS 0x00000800 | |
292 | #define RXf_GPOS_SEEN 0x00001000 | |
293 | #define RXf_GPOS_FLOAT 0x00002000 | |
294 | /* two bits here */ | |
295 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) | |
296 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) | |
297 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) | |
298 | ||
bbe252da | 299 | /* What we have seen */ |
52d81aa8 NC |
300 | #define RXf_LOOKBEHIND_SEEN 0x00004000 |
301 | #define RXf_EVAL_SEEN 0x00008000 | |
302 | #define RXf_CANY_SEEN 0x00010000 | |
bbe252da YO |
303 | |
304 | /* Special */ | |
52d81aa8 NC |
305 | #define RXf_NOSCAN 0x00020000 |
306 | #define RXf_CHECK_ALL 0x00040000 | |
bbe252da YO |
307 | |
308 | /* UTF8 related */ | |
52d81aa8 NC |
309 | #define RXf_UTF8 0x00080000 |
310 | #define RXf_MATCH_UTF8 0x00100000 | |
bbe252da YO |
311 | |
312 | /* Intuit related */ | |
52d81aa8 NC |
313 | #define RXf_USE_INTUIT_NOML 0x00200000 |
314 | #define RXf_USE_INTUIT_ML 0x00400000 | |
315 | #define RXf_INTUIT_TAIL 0x00800000 | |
0ac6acae AB |
316 | |
317 | /* | |
318 | Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will | |
319 | be used by regex engines to check whether they should set | |
320 | RXf_SKIPWHITE | |
321 | */ | |
52d81aa8 | 322 | #define RXf_SPLIT 0x01000000 |
0ac6acae | 323 | |
bbe252da YO |
324 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
325 | ||
326 | /* Copy and tainted info */ | |
52d81aa8 NC |
327 | #define RXf_COPY_DONE 0x02000000 |
328 | #define RXf_TAINTED_SEEN 0x04000000 | |
329 | #define RXf_TAINTED 0x08000000 /* this pattern is tainted */ | |
330 | ||
331 | /* Flags indicating special patterns */ | |
332 | #define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */ | |
333 | #define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */ | |
334 | #define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */ | |
335 | #define RXf_NULL 0x80000000 /* Pattern is // */ | |
c737faaf | 336 | |
cb5027f2 YO |
337 | /* |
338 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
339 | * so that regnodes.h is updated with the changes. | |
340 | * | |
341 | */ | |
bbe252da | 342 | |
bbe252da | 343 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
07bc277f NC |
344 | #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) |
345 | #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) | |
346 | #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) | |
347 | #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) | |
72311751 GS |
348 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
349 | ? RX_MATCH_TAINTED_on(prog) \ | |
350 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 351 | |
07bc277f NC |
352 | #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) |
353 | #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) | |
354 | #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) | |
355 | #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) | |
356 | #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) | |
357 | #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) | |
cf93c79d IZ |
358 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
359 | ? RX_MATCH_COPIED_on(prog) \ | |
360 | : RX_MATCH_COPIED_off(prog)) | |
f9f4320a | 361 | |
07bc277f | 362 | #define RXp_PRECOMP(rx) ((rx)->wrapped + (rx)->pre_prefix) |
51aa2004 NC |
363 | /* FIXME? Are we hardcoding too much here and constraining plugin extension |
364 | writers? Specifically, the value 1 assumes that the wrapped version always | |
365 | has exactly one character at the end, a ')'. Will that always be true? */ | |
07bc277f NC |
366 | #define RXp_PRELEN(rx) ((rx)->wraplen - (rx)->pre_prefix - 1) |
367 | #define RXp_WRAPPED(rx) ((rx)->wrapped) | |
368 | #define RXp_WRAPLEN(rx) ((rx)->wraplen) | |
369 | #define RXp_EXTFLAGS(rx) ((rx)->extflags) | |
866c78d1 | 370 | |
07bc277f NC |
371 | /* For source compatibility. We used to store these explicitly. */ |
372 | #define RX_PRECOMP(prog) ((prog)->wrapped + (prog)->pre_prefix) | |
373 | #define RX_PRELEN(prog) ((prog)->wraplen - (prog)->pre_prefix - 1) | |
374 | #define RX_WRAPPED(prog) ((prog)->wrapped) | |
375 | #define RX_WRAPLEN(prog) ((prog)->wraplen) | |
376 | #define RX_CHECK_SUBSTR(prog) ((prog)->check_substr) | |
377 | #define RX_EXTFLAGS(prog) ((prog)->extflags) | |
378 | #define RX_REFCNT(prog) ((prog)->refcnt) | |
379 | #define RX_ENGINE(prog) ((prog)->engine) | |
380 | #define RX_SUBBEG(prog) ((prog)->subbeg) | |
381 | #define RX_OFFS(prog) ((prog)->offs) | |
382 | #define RX_NPARENS(prog) ((prog)->nparens) | |
383 | #define RX_SUBLEN(prog) ((prog)->sublen) | |
384 | #define RX_SUBBEG(prog) ((prog)->subbeg) | |
385 | #define RX_MINLEN(prog) ((prog)->minlen) | |
386 | #define RX_MINLENRET(prog) ((prog)->minlenret) | |
387 | #define RX_GOFS(prog) ((prog)->gofs) | |
388 | #define RX_LASTPAREN(prog) ((prog)->lastparen) | |
389 | #define RX_LASTCLOSEPAREN(prog) ((prog)->lastcloseparen) | |
390 | #define RX_SEEN_EVALS(prog) ((prog)->seen_evals) | |
220fc49f | 391 | |
be8e71aa YO |
392 | #endif /* PLUGGABLE_RE_EXTENSION */ |
393 | ||
394 | /* Stuff that needs to be included in the plugable extension goes below here */ | |
395 | ||
f8c7b90f | 396 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
397 | #define RX_MATCH_COPY_FREE(rx) \ |
398 | STMT_START {if (rx->saved_copy) { \ | |
399 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ | |
400 | } \ | |
401 | if (RX_MATCH_COPIED(rx)) { \ | |
07bc277f | 402 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 NC |
403 | RX_MATCH_COPIED_off(rx); \ |
404 | }} STMT_END | |
405 | #else | |
406 | #define RX_MATCH_COPY_FREE(rx) \ | |
407 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ | |
07bc277f | 408 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 NC |
409 | RX_MATCH_COPIED_off(rx); \ |
410 | }} STMT_END | |
411 | #endif | |
412 | ||
07bc277f NC |
413 | #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) |
414 | #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) | |
415 | #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) | |
416 | #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) | |
a30b2f1f RGS |
417 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
418 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
419 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
420 | ||
cad2e5aa JH |
421 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
422 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
423 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
424 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 425 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 | 426 | |
155aba94 | 427 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a | 428 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d IZ |
429 | |
430 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
431 | #define FBMcf_TAIL_DOLLARM 2 |
432 | #define FBMcf_TAIL_Z 4 | |
433 | #define FBMcf_TAIL_z 8 | |
434 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
435 | |
436 | #define FBMrf_MULTILINE 1 | |
cad2e5aa | 437 | |
5d9a96ca DM |
438 | /* an accepting state/position*/ |
439 | struct _reg_trie_accepted { | |
440 | U8 *endpos; | |
441 | U16 wordnum; | |
442 | }; | |
443 | typedef struct _reg_trie_accepted reg_trie_accepted; | |
444 | ||
3b0527fe DM |
445 | /* some basic information about the current match that is created by |
446 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ | |
447 | ||
448 | typedef struct { | |
449 | regexp *prog; | |
450 | char *bol; | |
451 | char *till; | |
452 | SV *sv; | |
453 | char *ganch; | |
24b23f37 | 454 | char *cutpoint; |
3b0527fe DM |
455 | } regmatch_info; |
456 | ||
5d9a96ca DM |
457 | |
458 | /* structures for holding and saving the state maintained by regmatch() */ | |
459 | ||
d56b3014 | 460 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 | 461 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 | 462 | #endif |
6bda09f9 | 463 | |
5d9a96ca DM |
464 | typedef I32 CHECKPOINT; |
465 | ||
a0374537 | 466 | typedef struct regmatch_state { |
40a82448 | 467 | int resume_state; /* where to jump to on return */ |
24d3c4a9 | 468 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 DM |
469 | |
470 | union { | |
77cb431f DM |
471 | |
472 | /* this is a fake union member that matches the first element | |
473 | * of each member that needs to store positive backtrack | |
474 | * information */ | |
475 | struct { | |
476 | struct regmatch_state *prev_yes_state; | |
477 | } yes; | |
478 | ||
fae667d5 YO |
479 | /* branchlike members */ |
480 | /* this is a fake union member that matches the first elements | |
481 | * of each member that needs to behave like a branch */ | |
482 | struct { | |
5d458dd8 YO |
483 | /* this first element must match u.yes */ |
484 | struct regmatch_state *prev_yes_state; | |
fae667d5 YO |
485 | U32 lastparen; |
486 | CHECKPOINT cp; | |
487 | ||
488 | } branchlike; | |
489 | ||
490 | struct { | |
491 | /* the first elements must match u.branchlike */ | |
492 | struct regmatch_state *prev_yes_state; | |
493 | U32 lastparen; | |
494 | CHECKPOINT cp; | |
495 | ||
496 | regnode *next_branch; /* next branch node */ | |
497 | } branch; | |
498 | ||
499 | struct { | |
500 | /* the first elements must match u.branchlike */ | |
501 | struct regmatch_state *prev_yes_state; | |
502 | U32 lastparen; | |
503 | CHECKPOINT cp; | |
504 | ||
505 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ | |
166ba7cd | 506 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c | 507 | U16 *jump; /* positive offsets from me */ |
166ba7cd | 508 | regnode *B; /* node following the trie */ |
7f69552c | 509 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 DM |
510 | } trie; |
511 | ||
fae667d5 YO |
512 | /* special types - these members are used to store state for special |
513 | regops like eval, if/then, lookaround and the markpoint state */ | |
e822a8b4 | 514 | struct { |
77cb431f DM |
515 | /* this first element must match u.yes */ |
516 | struct regmatch_state *prev_yes_state; | |
faec1544 DM |
517 | struct regmatch_state *prev_eval; |
518 | struct regmatch_state *prev_curlyx; | |
aa283a38 | 519 | regexp *prev_rex; |
faec1544 DM |
520 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
521 | flip when transitioning between | |
522 | inner and outer rexen */ | |
aa283a38 DM |
523 | CHECKPOINT cp; /* remember current savestack indexes */ |
524 | CHECKPOINT lastcp; | |
6bda09f9 | 525 | U32 close_paren; /* which close bracket is our end */ |
1b5c067c | 526 | regnode *B; /* the node following us */ |
e822a8b4 DM |
527 | } eval; |
528 | ||
529 | struct { | |
c476f425 | 530 | /* this first element must match u.yes */ |
262b90c4 | 531 | struct regmatch_state *prev_yes_state; |
fae667d5 YO |
532 | I32 wanted; |
533 | I32 logical; /* saved copy of 'logical' var */ | |
534 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ | |
535 | } ifmatch; /* and SUSPEND/UNLESSM */ | |
536 | ||
537 | struct { | |
538 | /* this first element must match u.yes */ | |
539 | struct regmatch_state *prev_yes_state; | |
540 | struct regmatch_state *prev_mark; | |
541 | SV* mark_name; | |
542 | char *mark_loc; | |
543 | } mark; | |
544 | ||
545 | struct { | |
546 | int val; | |
547 | } keeper; | |
548 | ||
549 | /* quantifiers - these members are used for storing state for | |
550 | for the regops used to implement quantifiers */ | |
551 | struct { | |
552 | /* this first element must match u.yes */ | |
553 | struct regmatch_state *prev_yes_state; | |
c476f425 | 554 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
1b5c067c | 555 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
c476f425 DM |
556 | CHECKPOINT cp; /* remember current savestack index */ |
557 | bool minmod; | |
a0374537 | 558 | int parenfloor;/* how far back to strip paren data */ |
c476f425 DM |
559 | int min; /* the minimal number of A's to match */ |
560 | int max; /* the maximal number of A's to match */ | |
c476f425 DM |
561 | |
562 | /* these two are modified by WHILEM */ | |
563 | int count; /* how many instances of A we've matched */ | |
564 | char *lastloc;/* where previous A matched (0-len detect) */ | |
e822a8b4 DM |
565 | } curlyx; |
566 | ||
567 | struct { | |
c476f425 | 568 | /* this first element must match u.yes */ |
262b90c4 | 569 | struct regmatch_state *prev_yes_state; |
c476f425 DM |
570 | struct regmatch_state *save_curlyx; |
571 | CHECKPOINT cp; /* remember current savestack indexes */ | |
572 | CHECKPOINT lastcp; | |
573 | char *save_lastloc; /* previous curlyx.lastloc */ | |
574 | I32 cache_offset; | |
575 | I32 cache_mask; | |
e822a8b4 DM |
576 | } whilem; |
577 | ||
578 | struct { | |
5d458dd8 YO |
579 | /* this first element must match u.yes */ |
580 | struct regmatch_state *prev_yes_state; | |
e822a8b4 | 581 | I32 c1, c2; /* case fold search */ |
40a82448 DM |
582 | CHECKPOINT cp; |
583 | I32 alen; /* length of first-matched A string */ | |
584 | I32 count; | |
0cadcf80 | 585 | bool minmod; |
40a82448 DM |
586 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
587 | regnode *me; /* the curlym node */ | |
e822a8b4 DM |
588 | } curlym; |
589 | ||
590 | struct { | |
3b6647e0 | 591 | U32 paren; |
c255a977 | 592 | CHECKPOINT cp; |
e822a8b4 | 593 | I32 c1, c2; /* case fold search */ |
c255a977 DM |
594 | char *maxpos; /* highest possible point in string to match */ |
595 | char *oldloc; /* the previous locinput */ | |
e822a8b4 | 596 | int count; |
c255a977 DM |
597 | int min, max; /* {m,n} */ |
598 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ | |
599 | } curly; /* and CURLYN/PLUS/STAR */ | |
dad79028 | 600 | |
d8319b27 | 601 | } u; |
5d9a96ca DM |
602 | } regmatch_state; |
603 | ||
604 | /* how many regmatch_state structs to allocate as a single slab. | |
605 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev | |
606 | * pointers, plus 1 for any mythical malloc overhead. */ | |
607 | ||
608 | #define PERL_REGMATCH_SLAB_SLOTS \ | |
609 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) | |
610 | ||
611 | typedef struct regmatch_slab { | |
612 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; | |
613 | struct regmatch_slab *prev, *next; | |
614 | } regmatch_slab; | |
1ade1aa1 | 615 | |
46ab3289 NC |
616 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
617 | #define PL_bostr PL_reg_state.re_state_bostr | |
618 | #define PL_reginput PL_reg_state.re_state_reginput | |
46ab3289 | 619 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb | 620 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 NC |
621 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
622 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen | |
46ab3289 NC |
623 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
624 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl | |
625 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set | |
46ab3289 NC |
626 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
627 | #define PL_reg_magic PL_reg_state.re_state_reg_magic | |
628 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos | |
629 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm | |
630 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm | |
631 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved | |
632 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen | |
633 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter | |
634 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter | |
635 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache | |
636 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size | |
637 | #define PL_regsize PL_reg_state.re_state_regsize | |
638 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry | |
639 | #define PL_nrs PL_reg_state.re_state_nrs | |
640 | ||
1ade1aa1 NC |
641 | struct re_save_state { |
642 | U32 re_state_reg_flags; /* from regexec.c */ | |
1b5c067c NC |
643 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
644 | I32 re_state_reg_eval_set; /* from regexec.c */ | |
645 | bool re_state_reg_match_utf8; /* from regexec.c */ | |
1ade1aa1 NC |
646 | char *re_state_bostr; |
647 | char *re_state_reginput; /* String-input pointer. */ | |
1ade1aa1 | 648 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb | 649 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 NC |
650 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
651 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ | |
1ade1aa1 | 652 | char **re_state_reg_start_tmp; /* from regexec.c */ |
1ade1aa1 | 653 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
1ade1aa1 NC |
654 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
655 | PMOP *re_state_reg_curpm; /* from regexec.c */ | |
656 | char *re_state_reg_oldsaved; /* old saved substr during match */ | |
657 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ | |
1b5c067c NC |
658 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
659 | I32 re_state_reg_oldpos; /* from regexec.c */ | |
1ade1aa1 NC |
660 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
661 | I32 re_state_reg_leftiter; /* wait until caching pos */ | |
3b6647e0 | 662 | U32 re_state_regsize; /* from regexec.c */ |
1b5c067c | 663 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
1ade1aa1 NC |
664 | char *re_state_reg_starttry; /* from regexec.c */ |
665 | #ifdef PERL_OLD_COPY_ON_WRITE | |
46ab3289 | 666 | 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 |
667 | #endif |
668 | }; | |
669 | ||
670 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ | |
671 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) | |
cde0cee5 | 672 | |
1ade1aa1 NC |
673 | /* |
674 | * Local variables: | |
675 | * c-indentation-style: bsd | |
676 | * c-basic-offset: 4 | |
677 | * indent-tabs-mode: t | |
678 | * End: | |
679 | * | |
680 | * ex: set ts=8 sts=4 sw=4 noet: | |
681 | */ |