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 |
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 | |
f0ab9afb |
53 | typedef struct regexp_paren_pair { |
54 | I32 start; |
55 | I32 end; |
56 | } regexp_paren_pair; |
28d8d7f4 |
57 | |
882227b7 |
58 | /* |
59 | The regexp/REGEXP struct, see L<perlreapi> for further documentation |
60 | on the individual fields. The struct is ordered so that the most |
61 | commonly used fields are placed at the start. |
62 | |
63 | Any patch that adds items to this struct will need to include |
64 | changes to F<sv.c> (C<Perl_re_dup()>) and F<regcomp.c> |
65 | (C<pregfree()>). This involves freeing or cloning items in the |
66 | regexp's data array based on the data item's type. |
67 | */ |
68 | |
08e44740 |
69 | #define _REGEXP_COMMON \ |
70 | /* what engine created this regexp? */ \ |
71 | const struct regexp_engine* engine; \ |
72 | REGEXP *mother_re; /* what re is this a lightweight copy of? */ \ |
73 | /* Information about the match that the perl core uses to */ \ |
74 | /* manage things */ \ |
75 | U32 extflags; /* Flags used both externally and internally */ \ |
76 | I32 minlen; /* mininum possible length of string to match */\ |
77 | I32 minlenret; /* mininum possible length of $& */ \ |
78 | U32 gofs; /* chars left of pos that we search from */ \ |
79 | /* substring data about strings that must appear in the */ \ |
80 | /* final match, used for optimisations */ \ |
81 | struct reg_substr_data *substrs; \ |
82 | U32 nparens; /* number of capture buffers */ \ |
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 | /* Data about the last/current match. These are modified */ \ |
88 | /* during matching */ \ |
89 | U32 lastparen; /* last open paren matched */ \ |
90 | U32 lastcloseparen; /* last close paren matched */ \ |
91 | regexp_paren_pair *swap; /* Swap copy of *offs */ \ |
92 | /* Array of offsets for (@-) and (@+) */ \ |
93 | regexp_paren_pair *offs; \ |
94 | /* saved or original string so \digit works forever. */ \ |
95 | char *subbeg; \ |
96 | SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\ |
97 | I32 sublen; /* Length of string pointed by subbeg */ \ |
98 | /* Information about the match that isn't often used */ \ |
99 | /* offset from wrapped to the start of precomp */ \ |
100 | unsigned pre_prefix:4; \ |
101 | /* number of eval groups in the pattern - for security checks */\ |
102 | unsigned seen_evals:28 |
103 | |
bbe252da |
104 | typedef struct regexp { |
288b8c02 |
105 | _XPV_HEAD; |
106 | _XPVMG_HEAD; |
08e44740 |
107 | _REGEXP_COMMON; |
f8fc2ecf |
108 | } regexp; |
109 | |
08e44740 |
110 | struct regexp_allocated { |
111 | _XPV_ALLOCATED_HEAD; |
112 | _XPVMG_HEAD; |
113 | _REGEXP_COMMON; |
114 | }; |
115 | |
1af6200d |
116 | /* HV *paren_names; Optional hash of paren names |
117 | now stored in the IV union */ |
118 | |
119 | #define RXp_PAREN_NAMES(rx) ((rx)->xiv_u.xivu_hv) |
5daac39c |
120 | |
785a26d5 |
121 | /* used for high speed searches */ |
f9f4320a |
122 | typedef struct re_scream_pos_data_s |
123 | { |
124 | char **scream_olds; /* match pos */ |
125 | I32 *scream_pos; /* Internal iterator of scream. */ |
126 | } re_scream_pos_data; |
127 | |
785a26d5 |
128 | /* regexp_engine structure. This is the dispatch table for regexes. |
129 | * Any regex engine implementation must be able to build one of these. |
130 | */ |
f9f4320a |
131 | typedef struct regexp_engine { |
0a0b43fa |
132 | REGEXP* (*comp) (pTHX_ const SV * const pattern, U32 flags); |
49d7dfbc |
133 | I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend, |
134 | char* strbeg, I32 minend, SV* screamer, |
135 | void* data, U32 flags); |
136 | char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos, |
9f61653a |
137 | char *strend, const U32 flags, |
138 | re_scream_pos_data *data); |
49d7dfbc |
139 | SV* (*checkstr) (pTHX_ REGEXP * const rx); |
140 | void (*free) (pTHX_ REGEXP * const rx); |
2fdbfb4d |
141 | void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren, |
192b9cd1 |
142 | SV * const sv); |
2fdbfb4d |
143 | void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren, |
144 | SV const * const value); |
145 | I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv, |
146 | const I32 paren); |
192b9cd1 |
147 | SV* (*named_buff) (pTHX_ REGEXP * const rx, SV * const key, |
148 | SV * const value, const U32 flags); |
149 | SV* (*named_buff_iter) (pTHX_ REGEXP * const rx, const SV * const lastkey, |
150 | const U32 flags); |
49d7dfbc |
151 | SV* (*qr_package)(pTHX_ REGEXP * const rx); |
f2f78491 |
152 | #ifdef USE_ITHREADS |
49d7dfbc |
153 | void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param); |
2fdbfb4d |
154 | #endif |
f9f4320a |
155 | } regexp_engine; |
156 | |
192b9cd1 |
157 | /* |
158 | These are passed to the numbered capture variable callbacks as the |
159 | paren name. >= 1 is reserved for actual numbered captures, i.e. $1, |
160 | $2 etc. |
161 | */ |
f1b875a0 |
162 | #define RX_BUFF_IDX_PREMATCH -2 /* $` / ${^PREMATCH} */ |
163 | #define RX_BUFF_IDX_POSTMATCH -1 /* $' / ${^POSTMATCH} */ |
164 | #define RX_BUFF_IDX_FULLMATCH 0 /* $& / ${^MATCH} */ |
192b9cd1 |
165 | |
166 | /* |
167 | Flags that are passed to the named_buff and named_buff_iter |
168 | callbacks above. Those routines are called from universal.c via the |
169 | Tie::Hash::NamedCapture interface for %+ and %- and the re:: |
170 | functions in the same file. |
171 | */ |
172 | |
173 | /* The Tie::Hash::NamedCapture operation this is part of, if any */ |
f1b875a0 |
174 | #define RXapif_FETCH 0x0001 |
175 | #define RXapif_STORE 0x0002 |
176 | #define RXapif_DELETE 0x0004 |
177 | #define RXapif_CLEAR 0x0008 |
178 | #define RXapif_EXISTS 0x0010 |
179 | #define RXapif_SCALAR 0x0020 |
180 | #define RXapif_FIRSTKEY 0x0040 |
181 | #define RXapif_NEXTKEY 0x0080 |
192b9cd1 |
182 | |
183 | /* Whether %+ or %- is being operated on */ |
f1b875a0 |
184 | #define RXapif_ONE 0x0100 /* %+ */ |
185 | #define RXapif_ALL 0x0200 /* %- */ |
192b9cd1 |
186 | |
187 | /* Whether this is being called from a re:: function */ |
f1b875a0 |
188 | #define RXapif_REGNAME 0x0400 |
189 | #define RXapif_REGNAMES 0x0800 |
190 | #define RXapif_REGNAMES_COUNT 0x1000 |
192b9cd1 |
191 | |
f7e71195 |
192 | /* |
193 | =head1 REGEXP Functions |
194 | |
195 | =for apidoc Am|REGEXP *|SvRX|SV *sv |
196 | |
197 | Convenience macro to get the REGEXP from a SV. This is approximately |
198 | equivalent to the following snippet: |
199 | |
200 | if (SvMAGICAL(sv)) |
201 | mg_get(sv); |
202 | if (SvROK(sv) && |
203 | (tmpsv = (SV*)SvRV(sv)) && |
204 | SvTYPE(tmpsv) == SVt_PVMG && |
205 | (tmpmg = mg_find(tmpsv, PERL_MAGIC_qr))) |
206 | { |
207 | return (REGEXP *)tmpmg->mg_obj; |
208 | } |
209 | |
210 | NULL will be returned if a REGEXP* is not found. |
211 | |
212 | =for apidoc Am|bool|SvRXOK|SV* sv |
213 | |
214 | Returns a boolean indicating whether the SV contains qr magic |
215 | (PERL_MAGIC_qr). |
216 | |
217 | If you want to do something with the REGEXP* later use SvRX instead |
218 | and check for NULL. |
219 | |
220 | =cut |
221 | */ |
222 | |
223 | #define SvRX(sv) (Perl_get_re_arg(aTHX_ sv)) |
224 | #define SvRXOK(sv) (Perl_get_re_arg(aTHX_ sv) ? TRUE : FALSE) |
225 | |
226 | |
bbe252da |
227 | /* Flags stored in regexp->extflags |
228 | * These are used by code external to the regexp engine |
e357fc67 |
229 | * |
230 | * Note that flags starting with RXf_PMf_ have exact equivalents |
231 | * stored in op_pmflags and which are defined in op.h, they are defined |
232 | * numerically here only for clarity. |
cb5027f2 |
233 | * |
234 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl |
235 | * so that regnodes.h is updated with the changes. |
236 | * |
bbe252da |
237 | */ |
238 | |
52d81aa8 |
239 | /* 0x3F of extflags is used by (RXf_)PMf_COMPILETIME |
a1cf41cd |
240 | * If you change these you need to change the equivalent flags in op.h, and |
241 | * vice versa. */ |
52d81aa8 |
242 | #define RXf_PMf_MULTILINE 0x00000001 /* /m */ |
243 | #define RXf_PMf_SINGLELINE 0x00000002 /* /s */ |
244 | #define RXf_PMf_FOLD 0x00000004 /* /i */ |
245 | #define RXf_PMf_EXTENDED 0x00000008 /* /x */ |
246 | #define RXf_PMf_KEEPCOPY 0x00000010 /* /p */ |
247 | #define RXf_PMf_LOCALE 0x00000020 /* use locale */ |
e357fc67 |
248 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
52d81aa8 |
249 | #define RXf_PMf_STD_PMMOD_SHIFT 0 |
cde0cee5 |
250 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
251 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) |
252 | |
a20207d7 |
253 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
254 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ |
255 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ |
256 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ |
257 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break |
bbe252da |
258 | |
bcdf7404 |
259 | /* chars and strings used as regex pattern modifiers |
260 | * Singlular is a 'c'har, plural is a "string" |
87e95b7f |
261 | * |
262 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve |
263 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) |
264 | * for its own uses. So 'k' is out as well. |
bcdf7404 |
265 | */ |
266 | #define EXEC_PAT_MOD 'e' |
87e95b7f |
267 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 |
268 | #define ONCE_PAT_MOD 'o' |
269 | #define GLOBAL_PAT_MOD 'g' |
270 | #define CONTINUE_PAT_MOD 'c' |
271 | #define MULTILINE_PAT_MOD 'm' |
272 | #define SINGLE_PAT_MOD 's' |
273 | #define IGNORE_PAT_MOD 'i' |
274 | #define XTENDED_PAT_MOD 'x' |
275 | |
276 | #define ONCE_PAT_MODS "o" |
87e95b7f |
277 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 |
278 | #define EXEC_PAT_MODS "e" |
279 | #define LOOP_PAT_MODS "gc" |
280 | |
281 | #define STD_PAT_MODS "msix" |
282 | |
283 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS |
284 | |
285 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS |
286 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS |
287 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS |
288 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS |
289 | |
cb5027f2 |
290 | /* |
291 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl |
292 | * so that regnodes.h is updated with the changes. |
293 | * |
294 | */ |
bcdf7404 |
295 | |
52d81aa8 |
296 | /* Anchor and GPOS related stuff */ |
297 | #define RXf_ANCH_BOL 0x00000100 |
298 | #define RXf_ANCH_MBOL 0x00000200 |
299 | #define RXf_ANCH_SBOL 0x00000400 |
300 | #define RXf_ANCH_GPOS 0x00000800 |
301 | #define RXf_GPOS_SEEN 0x00001000 |
302 | #define RXf_GPOS_FLOAT 0x00002000 |
303 | /* two bits here */ |
304 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) |
305 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) |
306 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) |
307 | |
bbe252da |
308 | /* What we have seen */ |
52d81aa8 |
309 | #define RXf_LOOKBEHIND_SEEN 0x00004000 |
310 | #define RXf_EVAL_SEEN 0x00008000 |
311 | #define RXf_CANY_SEEN 0x00010000 |
bbe252da |
312 | |
313 | /* Special */ |
52d81aa8 |
314 | #define RXf_NOSCAN 0x00020000 |
315 | #define RXf_CHECK_ALL 0x00040000 |
bbe252da |
316 | |
317 | /* UTF8 related */ |
52d81aa8 |
318 | #define RXf_MATCH_UTF8 0x00100000 |
bbe252da |
319 | |
320 | /* Intuit related */ |
52d81aa8 |
321 | #define RXf_USE_INTUIT_NOML 0x00200000 |
322 | #define RXf_USE_INTUIT_ML 0x00400000 |
323 | #define RXf_INTUIT_TAIL 0x00800000 |
0ac6acae |
324 | |
325 | /* |
326 | Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will |
327 | be used by regex engines to check whether they should set |
328 | RXf_SKIPWHITE |
329 | */ |
52d81aa8 |
330 | #define RXf_SPLIT 0x01000000 |
0ac6acae |
331 | |
bbe252da |
332 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
333 | |
334 | /* Copy and tainted info */ |
52d81aa8 |
335 | #define RXf_COPY_DONE 0x02000000 |
336 | #define RXf_TAINTED_SEEN 0x04000000 |
337 | #define RXf_TAINTED 0x08000000 /* this pattern is tainted */ |
338 | |
339 | /* Flags indicating special patterns */ |
340 | #define RXf_START_ONLY 0x10000000 /* Pattern is /^/ */ |
341 | #define RXf_SKIPWHITE 0x20000000 /* Pattern is for a split / / */ |
342 | #define RXf_WHITE 0x40000000 /* Pattern is /\s+/ */ |
343 | #define RXf_NULL 0x80000000 /* Pattern is // */ |
c737faaf |
344 | |
cb5027f2 |
345 | /* |
346 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl |
347 | * so that regnodes.h is updated with the changes. |
348 | * |
349 | */ |
bbe252da |
350 | |
bbe252da |
351 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
07bc277f |
352 | #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) |
353 | #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) |
354 | #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) |
355 | #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) |
72311751 |
356 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
357 | ? RX_MATCH_TAINTED_on(prog) \ |
358 | : RX_MATCH_TAINTED_off(prog)) |
c277df42 |
359 | |
07bc277f |
360 | #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) |
361 | #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) |
362 | #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) |
363 | #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) |
364 | #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) |
365 | #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) |
cf93c79d |
366 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
367 | ? RX_MATCH_COPIED_on(prog) \ |
368 | : RX_MATCH_COPIED_off(prog)) |
f9f4320a |
369 | |
07bc277f |
370 | #define RXp_EXTFLAGS(rx) ((rx)->extflags) |
866c78d1 |
371 | |
07bc277f |
372 | /* For source compatibility. We used to store these explicitly. */ |
f7c278bf |
373 | #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) |
5509d87a |
374 | /* FIXME? Are we hardcoding too much here and constraining plugin extension |
375 | writers? Specifically, the value 1 assumes that the wrapped version always |
376 | has exactly one character at the end, a ')'. Will that always be true? */ |
377 | #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1) |
f7c278bf |
378 | #define RX_WRAPPED(prog) SvPVX(prog) |
9d17798d |
379 | #define RX_WRAPLEN(prog) SvCUR(prog) |
288b8c02 |
380 | #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr) |
288b8c02 |
381 | #define RX_REFCNT(prog) SvREFCNT(prog) |
4450b630 |
382 | #if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC) |
383 | # define RX_EXTFLAGS(prog) \ |
384 | (*({ \ |
385 | REGEXP *const thwape = (prog); \ |
386 | assert(SvTYPE(thwape) == SVt_REGEXP); \ |
387 | &RXp_EXTFLAGS(SvANY(thwape)); \ |
388 | })) |
389 | # define RX_ENGINE(prog) \ |
390 | (*({ \ |
391 | REGEXP *const thwape = (prog); \ |
392 | assert(SvTYPE(thwape) == SVt_REGEXP); \ |
393 | &SvANY(thwape)->engine; \ |
394 | })) |
395 | # define RX_SUBBEG(prog) \ |
396 | (*({ \ |
397 | REGEXP *const thwape = (prog); \ |
398 | assert(SvTYPE(thwape) == SVt_REGEXP); \ |
399 | &SvANY(thwape)->subbeg; \ |
400 | })) |
200ff815 |
401 | # define RX_OFFS(prog) \ |
4450b630 |
402 | (*({ \ |
403 | REGEXP *const thwape = (prog); \ |
404 | assert(SvTYPE(thwape) == SVt_REGEXP); \ |
200ff815 |
405 | &SvANY(thwape)->offs; \ |
4450b630 |
406 | })) |
407 | # define RX_NPARENS(prog) \ |
408 | (*({ \ |
409 | REGEXP *const thwape = (prog); \ |
410 | assert(SvTYPE(thwape) == SVt_REGEXP); \ |
411 | &SvANY(thwape)->nparens; \ |
412 | })) |
413 | #else |
414 | # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog)) |
415 | # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine) |
416 | # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg) |
417 | # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs) |
418 | # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens) |
419 | #endif |
288b8c02 |
420 | #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen) |
288b8c02 |
421 | #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen) |
422 | #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret) |
423 | #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs) |
424 | #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen) |
425 | #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen) |
426 | #define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals) |
bdd9a1b1 |
427 | #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy) |
220fc49f |
428 | |
be8e71aa |
429 | #endif /* PLUGGABLE_RE_EXTENSION */ |
430 | |
431 | /* Stuff that needs to be included in the plugable extension goes below here */ |
432 | |
f8c7b90f |
433 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 |
434 | #define RX_MATCH_COPY_FREE(rx) \ |
bdd9a1b1 |
435 | STMT_START {if (RX_SAVED_COPY(rx)) { \ |
436 | SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \ |
ed252734 |
437 | } \ |
438 | if (RX_MATCH_COPIED(rx)) { \ |
07bc277f |
439 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 |
440 | RX_MATCH_COPIED_off(rx); \ |
441 | }} STMT_END |
442 | #else |
443 | #define RX_MATCH_COPY_FREE(rx) \ |
444 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
07bc277f |
445 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 |
446 | RX_MATCH_COPIED_off(rx); \ |
447 | }} STMT_END |
448 | #endif |
449 | |
07bc277f |
450 | #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) |
451 | #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) |
452 | #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) |
453 | #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) |
a30b2f1f |
454 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
455 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
456 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
efd26800 |
457 | |
458 | /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ |
8f6ae13c |
459 | #define RX_UTF8(prog) SvUTF8(prog) |
a30b2f1f |
460 | |
cad2e5aa |
461 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
462 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
463 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
464 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
cf93c79d |
465 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 |
466 | |
288b8c02 |
467 | #if defined(__GNUC__) && !defined(__STRICT_ANSI__) && !defined(PERL_GCC_PEDANTIC) |
468 | # define ReREFCNT_inc(re) \ |
469 | ({ \ |
470 | /* This is here to generate a casting warning if incorrect. */ \ |
471 | REGEXP *const zwapp = (re); \ |
4450b630 |
472 | assert(SvTYPE(zwapp) == SVt_REGEXP); \ |
b7e0bd9e |
473 | (REGEXP *) SvREFCNT_inc(zwapp); \ |
288b8c02 |
474 | }) |
475 | # define ReREFCNT_dec(re) \ |
476 | ({ \ |
477 | /* This is here to generate a casting warning if incorrect. */ \ |
478 | REGEXP *const boff = (re); \ |
479 | SvREFCNT_dec(boff); \ |
480 | }) |
481 | #else |
482 | # define ReREFCNT_dec(re) SvREFCNT_dec(re) |
b7e0bd9e |
483 | # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re)) |
288b8c02 |
484 | #endif |
485 | |
486 | /* FIXME for plugins. */ |
cf93c79d |
487 | |
488 | #define FBMcf_TAIL_DOLLAR 1 |
cad2e5aa |
489 | #define FBMcf_TAIL_DOLLARM 2 |
490 | #define FBMcf_TAIL_Z 4 |
491 | #define FBMcf_TAIL_z 8 |
492 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
cf93c79d |
493 | |
494 | #define FBMrf_MULTILINE 1 |
cad2e5aa |
495 | |
5d9a96ca |
496 | /* an accepting state/position*/ |
497 | struct _reg_trie_accepted { |
498 | U8 *endpos; |
499 | U16 wordnum; |
500 | }; |
501 | typedef struct _reg_trie_accepted reg_trie_accepted; |
502 | |
3b0527fe |
503 | /* some basic information about the current match that is created by |
504 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ |
505 | |
506 | typedef struct { |
288b8c02 |
507 | REGEXP *prog; |
3b0527fe |
508 | char *bol; |
509 | char *till; |
510 | SV *sv; |
511 | char *ganch; |
24b23f37 |
512 | char *cutpoint; |
3b0527fe |
513 | } regmatch_info; |
514 | |
5d9a96ca |
515 | |
516 | /* structures for holding and saving the state maintained by regmatch() */ |
517 | |
d56b3014 |
518 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 |
519 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 |
520 | #endif |
6bda09f9 |
521 | |
5d9a96ca |
522 | typedef I32 CHECKPOINT; |
523 | |
a0374537 |
524 | typedef struct regmatch_state { |
40a82448 |
525 | int resume_state; /* where to jump to on return */ |
24d3c4a9 |
526 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 |
527 | |
528 | union { |
77cb431f |
529 | |
530 | /* this is a fake union member that matches the first element |
531 | * of each member that needs to store positive backtrack |
532 | * information */ |
533 | struct { |
534 | struct regmatch_state *prev_yes_state; |
535 | } yes; |
536 | |
fae667d5 |
537 | /* branchlike members */ |
538 | /* this is a fake union member that matches the first elements |
539 | * of each member that needs to behave like a branch */ |
540 | struct { |
5d458dd8 |
541 | /* this first element must match u.yes */ |
542 | struct regmatch_state *prev_yes_state; |
fae667d5 |
543 | U32 lastparen; |
544 | CHECKPOINT cp; |
545 | |
546 | } branchlike; |
547 | |
548 | struct { |
549 | /* the first elements must match u.branchlike */ |
550 | struct regmatch_state *prev_yes_state; |
551 | U32 lastparen; |
552 | CHECKPOINT cp; |
553 | |
554 | regnode *next_branch; /* next branch node */ |
555 | } branch; |
556 | |
557 | struct { |
558 | /* the first elements must match u.branchlike */ |
559 | struct regmatch_state *prev_yes_state; |
560 | U32 lastparen; |
561 | CHECKPOINT cp; |
562 | |
563 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ |
166ba7cd |
564 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c |
565 | U16 *jump; /* positive offsets from me */ |
166ba7cd |
566 | regnode *B; /* node following the trie */ |
7f69552c |
567 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 |
568 | } trie; |
569 | |
fae667d5 |
570 | /* special types - these members are used to store state for special |
571 | regops like eval, if/then, lookaround and the markpoint state */ |
e822a8b4 |
572 | struct { |
77cb431f |
573 | /* this first element must match u.yes */ |
574 | struct regmatch_state *prev_yes_state; |
faec1544 |
575 | struct regmatch_state *prev_eval; |
576 | struct regmatch_state *prev_curlyx; |
288b8c02 |
577 | REGEXP *prev_rex; |
faec1544 |
578 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
579 | flip when transitioning between |
580 | inner and outer rexen */ |
aa283a38 |
581 | CHECKPOINT cp; /* remember current savestack indexes */ |
582 | CHECKPOINT lastcp; |
6bda09f9 |
583 | U32 close_paren; /* which close bracket is our end */ |
1b5c067c |
584 | regnode *B; /* the node following us */ |
e822a8b4 |
585 | } eval; |
586 | |
587 | struct { |
c476f425 |
588 | /* this first element must match u.yes */ |
262b90c4 |
589 | struct regmatch_state *prev_yes_state; |
fae667d5 |
590 | I32 wanted; |
591 | I32 logical; /* saved copy of 'logical' var */ |
592 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ |
593 | } ifmatch; /* and SUSPEND/UNLESSM */ |
594 | |
595 | struct { |
596 | /* this first element must match u.yes */ |
597 | struct regmatch_state *prev_yes_state; |
598 | struct regmatch_state *prev_mark; |
599 | SV* mark_name; |
600 | char *mark_loc; |
601 | } mark; |
602 | |
603 | struct { |
604 | int val; |
605 | } keeper; |
606 | |
607 | /* quantifiers - these members are used for storing state for |
608 | for the regops used to implement quantifiers */ |
609 | struct { |
610 | /* this first element must match u.yes */ |
611 | struct regmatch_state *prev_yes_state; |
c476f425 |
612 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
1b5c067c |
613 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
c476f425 |
614 | CHECKPOINT cp; /* remember current savestack index */ |
615 | bool minmod; |
a0374537 |
616 | int parenfloor;/* how far back to strip paren data */ |
c476f425 |
617 | int min; /* the minimal number of A's to match */ |
618 | int max; /* the maximal number of A's to match */ |
c476f425 |
619 | |
620 | /* these two are modified by WHILEM */ |
621 | int count; /* how many instances of A we've matched */ |
622 | char *lastloc;/* where previous A matched (0-len detect) */ |
e822a8b4 |
623 | } curlyx; |
624 | |
625 | struct { |
c476f425 |
626 | /* this first element must match u.yes */ |
262b90c4 |
627 | struct regmatch_state *prev_yes_state; |
c476f425 |
628 | struct regmatch_state *save_curlyx; |
629 | CHECKPOINT cp; /* remember current savestack indexes */ |
630 | CHECKPOINT lastcp; |
631 | char *save_lastloc; /* previous curlyx.lastloc */ |
632 | I32 cache_offset; |
633 | I32 cache_mask; |
e822a8b4 |
634 | } whilem; |
635 | |
636 | struct { |
5d458dd8 |
637 | /* this first element must match u.yes */ |
638 | struct regmatch_state *prev_yes_state; |
e822a8b4 |
639 | I32 c1, c2; /* case fold search */ |
40a82448 |
640 | CHECKPOINT cp; |
641 | I32 alen; /* length of first-matched A string */ |
642 | I32 count; |
0cadcf80 |
643 | bool minmod; |
40a82448 |
644 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
645 | regnode *me; /* the curlym node */ |
e822a8b4 |
646 | } curlym; |
647 | |
648 | struct { |
3b6647e0 |
649 | U32 paren; |
c255a977 |
650 | CHECKPOINT cp; |
e822a8b4 |
651 | I32 c1, c2; /* case fold search */ |
c255a977 |
652 | char *maxpos; /* highest possible point in string to match */ |
653 | char *oldloc; /* the previous locinput */ |
e822a8b4 |
654 | int count; |
c255a977 |
655 | int min, max; /* {m,n} */ |
656 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
657 | } curly; /* and CURLYN/PLUS/STAR */ |
dad79028 |
658 | |
d8319b27 |
659 | } u; |
5d9a96ca |
660 | } regmatch_state; |
661 | |
662 | /* how many regmatch_state structs to allocate as a single slab. |
663 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
664 | * pointers, plus 1 for any mythical malloc overhead. */ |
665 | |
666 | #define PERL_REGMATCH_SLAB_SLOTS \ |
667 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
668 | |
669 | typedef struct regmatch_slab { |
670 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
671 | struct regmatch_slab *prev, *next; |
672 | } regmatch_slab; |
1ade1aa1 |
673 | |
46ab3289 |
674 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
675 | #define PL_bostr PL_reg_state.re_state_bostr |
676 | #define PL_reginput PL_reg_state.re_state_reginput |
46ab3289 |
677 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb |
678 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 |
679 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
680 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen |
46ab3289 |
681 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
682 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl |
683 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set |
46ab3289 |
684 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
685 | #define PL_reg_magic PL_reg_state.re_state_reg_magic |
686 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos |
687 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm |
688 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm |
689 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved |
690 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen |
691 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter |
692 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter |
693 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache |
694 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size |
695 | #define PL_regsize PL_reg_state.re_state_regsize |
696 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry |
697 | #define PL_nrs PL_reg_state.re_state_nrs |
698 | |
1ade1aa1 |
699 | struct re_save_state { |
700 | U32 re_state_reg_flags; /* from regexec.c */ |
1b5c067c |
701 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
702 | I32 re_state_reg_eval_set; /* from regexec.c */ |
703 | bool re_state_reg_match_utf8; /* from regexec.c */ |
1ade1aa1 |
704 | char *re_state_bostr; |
705 | char *re_state_reginput; /* String-input pointer. */ |
1ade1aa1 |
706 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb |
707 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 |
708 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
709 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ |
1ade1aa1 |
710 | char **re_state_reg_start_tmp; /* from regexec.c */ |
1ade1aa1 |
711 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
1ade1aa1 |
712 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
713 | PMOP *re_state_reg_curpm; /* from regexec.c */ |
714 | char *re_state_reg_oldsaved; /* old saved substr during match */ |
715 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ |
1b5c067c |
716 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
717 | I32 re_state_reg_oldpos; /* from regexec.c */ |
1ade1aa1 |
718 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
719 | I32 re_state_reg_leftiter; /* wait until caching pos */ |
3b6647e0 |
720 | U32 re_state_regsize; /* from regexec.c */ |
1b5c067c |
721 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
1ade1aa1 |
722 | char *re_state_reg_starttry; /* from regexec.c */ |
723 | #ifdef PERL_OLD_COPY_ON_WRITE |
46ab3289 |
724 | 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 |
725 | #endif |
726 | }; |
727 | |
728 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ |
729 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) |
cde0cee5 |
730 | |
1ade1aa1 |
731 | /* |
732 | * Local variables: |
733 | * c-indentation-style: bsd |
734 | * c-basic-offset: 4 |
735 | * indent-tabs-mode: t |
736 | * End: |
737 | * |
738 | * ex: set ts=8 sts=4 sw=4 noet: |
739 | */ |