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 | |
bbe252da |
69 | typedef struct regexp { |
f8149455 |
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 |
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 |
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 |
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 |
91 | regexp_paren_pair *swap; /* Swap copy of *offs */ |
92 | regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */ |
28d8d7f4 |
93 | |
bbe252da |
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 |
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 |
102 | char *wrapped; /* wrapped version of the pattern */ |
103 | I32 wraplen; /* length of wrapped */ |
2c96c0d6 |
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 |
106 | HV *paren_names; /* Optional hash of paren names */ |
107 | |
108 | /* Refcount of this regexp */ |
109 | I32 refcnt; /* Refcount of this regexp */ |
f8fc2ecf |
110 | } regexp; |
111 | |
785a26d5 |
112 | /* used for high speed searches */ |
f9f4320a |
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 |
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 |
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 |
128 | char *strend, const U32 flags, |
129 | re_scream_pos_data *data); |
49d7dfbc |
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 |
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 |
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 |
146 | } regexp_engine; |
147 | |
192b9cd1 |
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 |
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 |
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 |
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 |
173 | |
174 | /* Whether %+ or %- is being operated on */ |
f1b875a0 |
175 | #define RXapif_ONE 0x0100 /* %+ */ |
176 | #define RXapif_ALL 0x0200 /* %- */ |
192b9cd1 |
177 | |
178 | /* Whether this is being called from a re:: function */ |
f1b875a0 |
179 | #define RXapif_REGNAME 0x0400 |
180 | #define RXapif_REGNAMES 0x0800 |
181 | #define RXapif_REGNAMES_COUNT 0x1000 |
192b9cd1 |
182 | |
f7e71195 |
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 |
218 | /* Flags stored in regexp->extflags |
219 | * These are used by code external to the regexp engine |
e357fc67 |
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 |
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 |
228 | */ |
229 | |
230 | /* Anchor and GPOS related stuff */ |
231 | #define RXf_ANCH_BOL 0x00000001 |
232 | #define RXf_ANCH_MBOL 0x00000002 |
233 | #define RXf_ANCH_SBOL 0x00000004 |
234 | #define RXf_ANCH_GPOS 0x00000008 |
235 | #define RXf_GPOS_SEEN 0x00000010 |
236 | #define RXf_GPOS_FLOAT 0x00000020 |
e357fc67 |
237 | /* two bits here */ |
bbe252da |
238 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) |
239 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) |
e357fc67 |
240 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) |
241 | |
242 | /* Flags indicating special patterns */ |
c737faaf |
243 | #define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */ |
e357fc67 |
244 | #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */ |
245 | #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */ |
640f820d |
246 | #define RXf_NULL 0x40000000 /* Pattern is // */ |
e357fc67 |
247 | |
bcdf7404 |
248 | /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */ |
e357fc67 |
249 | #define RXf_PMf_LOCALE 0x00000800 /* use locale */ |
250 | #define RXf_PMf_MULTILINE 0x00001000 /* /m */ |
251 | #define RXf_PMf_SINGLELINE 0x00002000 /* /s */ |
252 | #define RXf_PMf_FOLD 0x00004000 /* /i */ |
253 | #define RXf_PMf_EXTENDED 0x00008000 /* /x */ |
f7819f85 |
254 | #define RXf_PMf_KEEPCOPY 0x00010000 /* /p */ |
e357fc67 |
255 | /* these flags are transfered from the PMOP->op_pmflags member during compilation */ |
cde0cee5 |
256 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
257 | #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY) |
258 | |
a20207d7 |
259 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
260 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ |
261 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ |
262 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ |
263 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break |
bbe252da |
264 | |
bcdf7404 |
265 | /* chars and strings used as regex pattern modifiers |
266 | * Singlular is a 'c'har, plural is a "string" |
87e95b7f |
267 | * |
268 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve |
269 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) |
270 | * for its own uses. So 'k' is out as well. |
bcdf7404 |
271 | */ |
272 | #define EXEC_PAT_MOD 'e' |
87e95b7f |
273 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 |
274 | #define ONCE_PAT_MOD 'o' |
275 | #define GLOBAL_PAT_MOD 'g' |
276 | #define CONTINUE_PAT_MOD 'c' |
277 | #define MULTILINE_PAT_MOD 'm' |
278 | #define SINGLE_PAT_MOD 's' |
279 | #define IGNORE_PAT_MOD 'i' |
280 | #define XTENDED_PAT_MOD 'x' |
281 | |
282 | #define ONCE_PAT_MODS "o" |
87e95b7f |
283 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 |
284 | #define EXEC_PAT_MODS "e" |
285 | #define LOOP_PAT_MODS "gc" |
286 | |
287 | #define STD_PAT_MODS "msix" |
288 | |
289 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS |
290 | |
291 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS |
292 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS |
293 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS |
294 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS |
295 | |
cb5027f2 |
296 | /* |
297 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl |
298 | * so that regnodes.h is updated with the changes. |
299 | * |
300 | */ |
bcdf7404 |
301 | |
bbe252da |
302 | /* What we have seen */ |
bbe252da |
303 | #define RXf_LOOKBEHIND_SEEN 0x00020000 |
304 | #define RXf_EVAL_SEEN 0x00040000 |
305 | #define RXf_CANY_SEEN 0x00080000 |
306 | |
307 | /* Special */ |
308 | #define RXf_NOSCAN 0x00100000 |
309 | #define RXf_CHECK_ALL 0x00200000 |
310 | |
311 | /* UTF8 related */ |
312 | #define RXf_UTF8 0x00400000 |
995b21a2 |
313 | #define RXf_MATCH_UTF8 0x00800000 |
bbe252da |
314 | |
315 | /* Intuit related */ |
316 | #define RXf_USE_INTUIT_NOML 0x01000000 |
317 | #define RXf_USE_INTUIT_ML 0x02000000 |
318 | #define RXf_INTUIT_TAIL 0x04000000 |
0ac6acae |
319 | |
320 | /* |
321 | Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will |
322 | be used by regex engines to check whether they should set |
323 | RXf_SKIPWHITE |
324 | */ |
325 | #define RXf_SPLIT 0x08000000 |
326 | |
bbe252da |
327 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
328 | |
329 | /* Copy and tainted info */ |
330 | #define RXf_COPY_DONE 0x10000000 |
331 | #define RXf_TAINTED_SEEN 0x20000000 |
c737faaf |
332 | #define RXf_TAINTED 0x80000000 /* this pattern is tainted */ |
333 | |
cb5027f2 |
334 | /* |
335 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl |
336 | * so that regnodes.h is updated with the changes. |
337 | * |
338 | */ |
bbe252da |
339 | |
bbe252da |
340 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
341 | #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN) |
342 | #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN) |
343 | #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN) |
72311751 |
344 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
345 | ? RX_MATCH_TAINTED_on(prog) \ |
346 | : RX_MATCH_TAINTED_off(prog)) |
c277df42 |
347 | |
bbe252da |
348 | #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE) |
349 | #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE) |
350 | #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE) |
cf93c79d |
351 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
352 | ? RX_MATCH_COPIED_on(prog) \ |
353 | : RX_MATCH_COPIED_off(prog)) |
f9f4320a |
354 | |
220fc49f |
355 | /* For source compatibility. We used to store these explicitly. */ |
2c96c0d6 |
356 | #define RX_PRECOMP(prog) ((prog)->wrapped + (prog)->pre_prefix) |
51aa2004 |
357 | /* FIXME? Are we hardcoding too much here and constraining plugin extension |
358 | writers? Specifically, the value 1 assumes that the wrapped version always |
359 | has exactly one character at the end, a ')'. Will that always be true? */ |
360 | #define RX_PRELEN(prog) ((prog)->wraplen - (prog)->pre_prefix - 1) |
220fc49f |
361 | |
be8e71aa |
362 | #endif /* PLUGGABLE_RE_EXTENSION */ |
363 | |
364 | /* Stuff that needs to be included in the plugable extension goes below here */ |
365 | |
f8c7b90f |
366 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 |
367 | #define RX_MATCH_COPY_FREE(rx) \ |
368 | STMT_START {if (rx->saved_copy) { \ |
369 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ |
370 | } \ |
371 | if (RX_MATCH_COPIED(rx)) { \ |
372 | Safefree(rx->subbeg); \ |
373 | RX_MATCH_COPIED_off(rx); \ |
374 | }} STMT_END |
375 | #else |
376 | #define RX_MATCH_COPY_FREE(rx) \ |
377 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
378 | Safefree(rx->subbeg); \ |
379 | RX_MATCH_COPIED_off(rx); \ |
380 | }} STMT_END |
381 | #endif |
382 | |
995b21a2 |
383 | #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8) |
384 | #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8) |
385 | #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8) |
a30b2f1f |
386 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
387 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
388 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
389 | |
cad2e5aa |
390 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
391 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
392 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
393 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
cf93c79d |
394 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 |
395 | |
155aba94 |
396 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a |
397 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d |
398 | |
399 | #define FBMcf_TAIL_DOLLAR 1 |
cad2e5aa |
400 | #define FBMcf_TAIL_DOLLARM 2 |
401 | #define FBMcf_TAIL_Z 4 |
402 | #define FBMcf_TAIL_z 8 |
403 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
cf93c79d |
404 | |
405 | #define FBMrf_MULTILINE 1 |
cad2e5aa |
406 | |
5d9a96ca |
407 | /* an accepting state/position*/ |
408 | struct _reg_trie_accepted { |
409 | U8 *endpos; |
410 | U16 wordnum; |
411 | }; |
412 | typedef struct _reg_trie_accepted reg_trie_accepted; |
413 | |
3b0527fe |
414 | /* some basic information about the current match that is created by |
415 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ |
416 | |
417 | typedef struct { |
418 | regexp *prog; |
419 | char *bol; |
420 | char *till; |
421 | SV *sv; |
422 | char *ganch; |
24b23f37 |
423 | char *cutpoint; |
3b0527fe |
424 | } regmatch_info; |
425 | |
5d9a96ca |
426 | |
427 | /* structures for holding and saving the state maintained by regmatch() */ |
428 | |
d56b3014 |
429 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 |
430 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 |
431 | #endif |
6bda09f9 |
432 | |
5d9a96ca |
433 | typedef I32 CHECKPOINT; |
434 | |
a0374537 |
435 | typedef struct regmatch_state { |
40a82448 |
436 | int resume_state; /* where to jump to on return */ |
24d3c4a9 |
437 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 |
438 | |
439 | union { |
77cb431f |
440 | |
441 | /* this is a fake union member that matches the first element |
442 | * of each member that needs to store positive backtrack |
443 | * information */ |
444 | struct { |
445 | struct regmatch_state *prev_yes_state; |
446 | } yes; |
447 | |
fae667d5 |
448 | /* branchlike members */ |
449 | /* this is a fake union member that matches the first elements |
450 | * of each member that needs to behave like a branch */ |
451 | struct { |
5d458dd8 |
452 | /* this first element must match u.yes */ |
453 | struct regmatch_state *prev_yes_state; |
fae667d5 |
454 | U32 lastparen; |
455 | CHECKPOINT cp; |
456 | |
457 | } branchlike; |
458 | |
459 | struct { |
460 | /* the first elements must match u.branchlike */ |
461 | struct regmatch_state *prev_yes_state; |
462 | U32 lastparen; |
463 | CHECKPOINT cp; |
464 | |
465 | regnode *next_branch; /* next branch node */ |
466 | } branch; |
467 | |
468 | struct { |
469 | /* the first elements must match u.branchlike */ |
470 | struct regmatch_state *prev_yes_state; |
471 | U32 lastparen; |
472 | CHECKPOINT cp; |
473 | |
474 | reg_trie_accepted *accept_buff; /* accepting states we have seen */ |
166ba7cd |
475 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c |
476 | U16 *jump; /* positive offsets from me */ |
166ba7cd |
477 | regnode *B; /* node following the trie */ |
7f69552c |
478 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 |
479 | } trie; |
480 | |
fae667d5 |
481 | /* special types - these members are used to store state for special |
482 | regops like eval, if/then, lookaround and the markpoint state */ |
e822a8b4 |
483 | struct { |
77cb431f |
484 | /* this first element must match u.yes */ |
485 | struct regmatch_state *prev_yes_state; |
faec1544 |
486 | struct regmatch_state *prev_eval; |
487 | struct regmatch_state *prev_curlyx; |
aa283a38 |
488 | regexp *prev_rex; |
faec1544 |
489 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
490 | flip when transitioning between |
491 | inner and outer rexen */ |
aa283a38 |
492 | CHECKPOINT cp; /* remember current savestack indexes */ |
493 | CHECKPOINT lastcp; |
6bda09f9 |
494 | U32 close_paren; /* which close bracket is our end */ |
1b5c067c |
495 | regnode *B; /* the node following us */ |
e822a8b4 |
496 | } eval; |
497 | |
498 | struct { |
c476f425 |
499 | /* this first element must match u.yes */ |
262b90c4 |
500 | struct regmatch_state *prev_yes_state; |
fae667d5 |
501 | I32 wanted; |
502 | I32 logical; /* saved copy of 'logical' var */ |
503 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ |
504 | } ifmatch; /* and SUSPEND/UNLESSM */ |
505 | |
506 | struct { |
507 | /* this first element must match u.yes */ |
508 | struct regmatch_state *prev_yes_state; |
509 | struct regmatch_state *prev_mark; |
510 | SV* mark_name; |
511 | char *mark_loc; |
512 | } mark; |
513 | |
514 | struct { |
515 | int val; |
516 | } keeper; |
517 | |
518 | /* quantifiers - these members are used for storing state for |
519 | for the regops used to implement quantifiers */ |
520 | struct { |
521 | /* this first element must match u.yes */ |
522 | struct regmatch_state *prev_yes_state; |
c476f425 |
523 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
1b5c067c |
524 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
c476f425 |
525 | CHECKPOINT cp; /* remember current savestack index */ |
526 | bool minmod; |
a0374537 |
527 | int parenfloor;/* how far back to strip paren data */ |
c476f425 |
528 | int min; /* the minimal number of A's to match */ |
529 | int max; /* the maximal number of A's to match */ |
c476f425 |
530 | |
531 | /* these two are modified by WHILEM */ |
532 | int count; /* how many instances of A we've matched */ |
533 | char *lastloc;/* where previous A matched (0-len detect) */ |
e822a8b4 |
534 | } curlyx; |
535 | |
536 | struct { |
c476f425 |
537 | /* this first element must match u.yes */ |
262b90c4 |
538 | struct regmatch_state *prev_yes_state; |
c476f425 |
539 | struct regmatch_state *save_curlyx; |
540 | CHECKPOINT cp; /* remember current savestack indexes */ |
541 | CHECKPOINT lastcp; |
542 | char *save_lastloc; /* previous curlyx.lastloc */ |
543 | I32 cache_offset; |
544 | I32 cache_mask; |
e822a8b4 |
545 | } whilem; |
546 | |
547 | struct { |
5d458dd8 |
548 | /* this first element must match u.yes */ |
549 | struct regmatch_state *prev_yes_state; |
e822a8b4 |
550 | I32 c1, c2; /* case fold search */ |
40a82448 |
551 | CHECKPOINT cp; |
552 | I32 alen; /* length of first-matched A string */ |
553 | I32 count; |
0cadcf80 |
554 | bool minmod; |
40a82448 |
555 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
556 | regnode *me; /* the curlym node */ |
e822a8b4 |
557 | } curlym; |
558 | |
559 | struct { |
3b6647e0 |
560 | U32 paren; |
c255a977 |
561 | CHECKPOINT cp; |
e822a8b4 |
562 | I32 c1, c2; /* case fold search */ |
c255a977 |
563 | char *maxpos; /* highest possible point in string to match */ |
564 | char *oldloc; /* the previous locinput */ |
e822a8b4 |
565 | int count; |
c255a977 |
566 | int min, max; /* {m,n} */ |
567 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
568 | } curly; /* and CURLYN/PLUS/STAR */ |
dad79028 |
569 | |
d8319b27 |
570 | } u; |
5d9a96ca |
571 | } regmatch_state; |
572 | |
573 | /* how many regmatch_state structs to allocate as a single slab. |
574 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
575 | * pointers, plus 1 for any mythical malloc overhead. */ |
576 | |
577 | #define PERL_REGMATCH_SLAB_SLOTS \ |
578 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
579 | |
580 | typedef struct regmatch_slab { |
581 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
582 | struct regmatch_slab *prev, *next; |
583 | } regmatch_slab; |
1ade1aa1 |
584 | |
46ab3289 |
585 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
586 | #define PL_bostr PL_reg_state.re_state_bostr |
587 | #define PL_reginput PL_reg_state.re_state_reginput |
46ab3289 |
588 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb |
589 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 |
590 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
591 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen |
46ab3289 |
592 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
593 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl |
594 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set |
46ab3289 |
595 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
596 | #define PL_reg_magic PL_reg_state.re_state_reg_magic |
597 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos |
598 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm |
599 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm |
600 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved |
601 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen |
602 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter |
603 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter |
604 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache |
605 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size |
606 | #define PL_regsize PL_reg_state.re_state_regsize |
607 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry |
608 | #define PL_nrs PL_reg_state.re_state_nrs |
609 | |
1ade1aa1 |
610 | struct re_save_state { |
611 | U32 re_state_reg_flags; /* from regexec.c */ |
1b5c067c |
612 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
613 | I32 re_state_reg_eval_set; /* from regexec.c */ |
614 | bool re_state_reg_match_utf8; /* from regexec.c */ |
1ade1aa1 |
615 | char *re_state_bostr; |
616 | char *re_state_reginput; /* String-input pointer. */ |
1ade1aa1 |
617 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb |
618 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 |
619 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
620 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ |
1ade1aa1 |
621 | char **re_state_reg_start_tmp; /* from regexec.c */ |
1ade1aa1 |
622 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
1ade1aa1 |
623 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
624 | PMOP *re_state_reg_curpm; /* from regexec.c */ |
625 | char *re_state_reg_oldsaved; /* old saved substr during match */ |
626 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ |
1b5c067c |
627 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
628 | I32 re_state_reg_oldpos; /* from regexec.c */ |
1ade1aa1 |
629 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
630 | I32 re_state_reg_leftiter; /* wait until caching pos */ |
3b6647e0 |
631 | U32 re_state_regsize; /* from regexec.c */ |
1b5c067c |
632 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
1ade1aa1 |
633 | char *re_state_reg_starttry; /* from regexec.c */ |
634 | #ifdef PERL_OLD_COPY_ON_WRITE |
46ab3289 |
635 | 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 |
636 | #endif |
637 | }; |
638 | |
639 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ |
640 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) |
cde0cee5 |
641 | |
1ade1aa1 |
642 | /* |
643 | * Local variables: |
644 | * c-indentation-style: bsd |
645 | * c-basic-offset: 4 |
646 | * indent-tabs-mode: t |
647 | * End: |
648 | * |
649 | * ex: set ts=8 sts=4 sw=4 noet: |
650 | */ |