Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* regexp.h |
d6376244 | 2 | * |
4bb101f2 | 3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
1129b882 | 4 | * 2005, 2006, 2007, 2008 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 | ||
08e44740 NC |
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? */ \ | |
89f6f287 | 73 | HV *paren_names; /* Optional hash of paren names */ \ |
08e44740 NC |
74 | /* Information about the match that the perl core uses to */ \ |
75 | /* manage things */ \ | |
76 | U32 extflags; /* Flags used both externally and internally */ \ | |
77 | I32 minlen; /* mininum possible length of string to match */\ | |
78 | I32 minlenret; /* mininum possible length of $& */ \ | |
79 | U32 gofs; /* chars left of pos that we search from */ \ | |
80 | /* substring data about strings that must appear in the */ \ | |
81 | /* final match, used for optimisations */ \ | |
82 | struct reg_substr_data *substrs; \ | |
83 | U32 nparens; /* number of capture buffers */ \ | |
84 | /* private engine specific data */ \ | |
85 | U32 intflags; /* Engine Specific Internal flags */ \ | |
86 | void *pprivate; /* Data private to the regex engine which */ \ | |
87 | /* created this object. */ \ | |
88 | /* Data about the last/current match. These are modified */ \ | |
89 | /* during matching */ \ | |
90 | U32 lastparen; /* last open paren matched */ \ | |
91 | U32 lastcloseparen; /* last close paren matched */ \ | |
e9105d30 | 92 | regexp_paren_pair *swap; /* Unused: 5.10.1 and later */ \ |
08e44740 NC |
93 | /* Array of offsets for (@-) and (@+) */ \ |
94 | regexp_paren_pair *offs; \ | |
95 | /* saved or original string so \digit works forever. */ \ | |
96 | char *subbeg; \ | |
97 | SV_SAVED_COPY /* If non-NULL, SV which is COW from original */\ | |
98 | I32 sublen; /* Length of string pointed by subbeg */ \ | |
99 | /* Information about the match that isn't often used */ \ | |
100 | /* offset from wrapped to the start of precomp */ \ | |
654eccd5 | 101 | PERL_BITFIELD32 pre_prefix:4; \ |
08e44740 | 102 | /* number of eval groups in the pattern - for security checks */\ |
654eccd5 | 103 | PERL_BITFIELD32 seen_evals:28 |
08e44740 | 104 | |
bbe252da | 105 | typedef struct regexp { |
288b8c02 | 106 | _XPV_HEAD; |
08e44740 | 107 | _REGEXP_COMMON; |
f8fc2ecf YO |
108 | } regexp; |
109 | ||
89f6f287 | 110 | #define RXp_PAREN_NAMES(rx) ((rx)->paren_names) |
5daac39c | 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 { |
1593ad57 | 123 | REGEXP* (*comp) (pTHX_ SV * const pattern, 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 | |
7948fc08 | 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 | ||
7948fc08 | 218 | /* Flags stored in regexp->extflags |
bbe252da | 219 | * These are used by code external to the regexp engine |
e357fc67 | 220 | * |
5b126c84 KW |
221 | * Note that the flags whose names start with RXf_PMf_ are defined in |
222 | * op_reg_common.h, being copied from the parallel flags of op_pmflags | |
cb5027f2 YO |
223 | * |
224 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
7948fc08 | 225 | * so that regnodes.h is updated with the changes. |
cb5027f2 | 226 | * |
bbe252da YO |
227 | */ |
228 | ||
5b126c84 KW |
229 | #include "op_reg_common.h" |
230 | ||
cde0cee5 | 231 | #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED) |
cde0cee5 | 232 | |
a20207d7 YO |
233 | #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \ |
234 | case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \ | |
235 | case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \ | |
236 | case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \ | |
237 | case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break | |
bbe252da | 238 | |
9de15fec | 239 | /* Note, includes locale, unicode */ |
fb85c044 | 240 | #define STD_PMMOD_FLAGS_CLEAR(pmfl) \ |
9de15fec | 241 | *(pmfl) &= ~(RXf_PMf_FOLD|RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_EXTENDED|RXf_PMf_LOCALE|RXf_PMf_UNICODE) |
fb85c044 | 242 | |
bcdf7404 YO |
243 | /* chars and strings used as regex pattern modifiers |
244 | * Singlular is a 'c'har, plural is a "string" | |
87e95b7f YO |
245 | * |
246 | * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve | |
247 | * for compatibility reasons with Regexp::Common which highjacked (?k:...) | |
248 | * for its own uses. So 'k' is out as well. | |
bcdf7404 | 249 | */ |
85508812 | 250 | #define DEFAULT_PAT_MOD '^' /* Short for all the default modifiers */ |
bcdf7404 | 251 | #define EXEC_PAT_MOD 'e' |
87e95b7f | 252 | #define KEEPCOPY_PAT_MOD 'p' |
bcdf7404 YO |
253 | #define ONCE_PAT_MOD 'o' |
254 | #define GLOBAL_PAT_MOD 'g' | |
255 | #define CONTINUE_PAT_MOD 'c' | |
256 | #define MULTILINE_PAT_MOD 'm' | |
257 | #define SINGLE_PAT_MOD 's' | |
258 | #define IGNORE_PAT_MOD 'i' | |
259 | #define XTENDED_PAT_MOD 'x' | |
4f4d7508 | 260 | #define NONDESTRUCT_PAT_MOD 'r' |
9de15fec KW |
261 | #define LOCALE_PAT_MOD 'l' |
262 | #define UNICODE_PAT_MOD 'u' | |
263 | #define DUAL_PAT_MOD 'd' | |
bcdf7404 YO |
264 | |
265 | #define ONCE_PAT_MODS "o" | |
87e95b7f | 266 | #define KEEPCOPY_PAT_MODS "p" |
bcdf7404 YO |
267 | #define EXEC_PAT_MODS "e" |
268 | #define LOOP_PAT_MODS "gc" | |
4f4d7508 | 269 | #define NONDESTRUCT_PAT_MODS "r" |
9de15fec KW |
270 | #define LOCALE_PAT_MODS "l" |
271 | #define UNICODE_PAT_MODS "u" | |
272 | #define DUAL_PAT_MODS "d" | |
bcdf7404 | 273 | |
78fecfc8 | 274 | /* This string is expected by regcomp.c to be ordered so that the first |
c18d5d15 KW |
275 | * character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of extflags; the next |
276 | * character is bit +1, etc. */ | |
bcdf7404 YO |
277 | #define STD_PAT_MODS "msix" |
278 | ||
78fecfc8 | 279 | /* This string is expected by XS_re_regexp_pattern() in universal.c to be ordered |
c18d5d15 KW |
280 | * so that the first character is the flag in bit RXf_PMf_STD_PMMOD_SHIFT of |
281 | * extflags; the next character is in bit +1, etc. */ | |
bcdf7404 YO |
282 | #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS |
283 | ||
284 | #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS | |
285 | #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS | |
286 | #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS | |
4f4d7508 | 287 | #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS NONDESTRUCT_PAT_MODS |
bcdf7404 | 288 | |
cb5027f2 YO |
289 | /* |
290 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
7948fc08 | 291 | * so that regnodes.h is updated with the changes. |
cb5027f2 YO |
292 | * |
293 | */ | |
bcdf7404 | 294 | |
e795e964 KW |
295 | /* Leave some space, so future bit allocations can go either in the shared or |
296 | * unshared area without affecting binary compatibility */ | |
9de15fec | 297 | #define RXf_BASE_SHIFT (_RXf_PMf_SHIFT_NEXT+2) |
f4a4cecb | 298 | |
52d81aa8 | 299 | /* Anchor and GPOS related stuff */ |
1ad4ec54 KW |
300 | #define RXf_ANCH_BOL (1<<(RXf_BASE_SHIFT+0)) |
301 | #define RXf_ANCH_MBOL (1<<(RXf_BASE_SHIFT+1)) | |
302 | #define RXf_ANCH_SBOL (1<<(RXf_BASE_SHIFT+2)) | |
303 | #define RXf_ANCH_GPOS (1<<(RXf_BASE_SHIFT+3)) | |
304 | #define RXf_GPOS_SEEN (1<<(RXf_BASE_SHIFT+4)) | |
305 | #define RXf_GPOS_FLOAT (1<<(RXf_BASE_SHIFT+5)) | |
52d81aa8 NC |
306 | /* two bits here */ |
307 | #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL) | |
308 | #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS) | |
309 | #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS) | |
310 | ||
bbe252da | 311 | /* What we have seen */ |
1ad4ec54 KW |
312 | #define RXf_LOOKBEHIND_SEEN (1<<(RXf_BASE_SHIFT+6)) |
313 | #define RXf_EVAL_SEEN (1<<(RXf_BASE_SHIFT+7)) | |
314 | #define RXf_CANY_SEEN (1<<(RXf_BASE_SHIFT+8)) | |
bbe252da YO |
315 | |
316 | /* Special */ | |
1ad4ec54 KW |
317 | #define RXf_NOSCAN (1<<(RXf_BASE_SHIFT+9)) |
318 | #define RXf_CHECK_ALL (1<<(RXf_BASE_SHIFT+10)) | |
bbe252da YO |
319 | |
320 | /* UTF8 related */ | |
e795e964 | 321 | #define RXf_MATCH_UTF8 (1<<(RXf_BASE_SHIFT+11)) |
bbe252da YO |
322 | |
323 | /* Intuit related */ | |
e795e964 KW |
324 | #define RXf_USE_INTUIT_NOML (1<<(RXf_BASE_SHIFT+12)) |
325 | #define RXf_USE_INTUIT_ML (1<<(RXf_BASE_SHIFT+13)) | |
326 | #define RXf_INTUIT_TAIL (1<<(RXf_BASE_SHIFT+14)) | |
0ac6acae AB |
327 | |
328 | /* | |
329 | Set in Perl_pmruntime if op_flags & OPf_SPECIAL, i.e. split. Will | |
330 | be used by regex engines to check whether they should set | |
331 | RXf_SKIPWHITE | |
332 | */ | |
e795e964 | 333 | #define RXf_SPLIT (1<<(RXf_BASE_SHIFT+15)) |
0ac6acae | 334 | |
bbe252da YO |
335 | #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML) |
336 | ||
337 | /* Copy and tainted info */ | |
e795e964 KW |
338 | #define RXf_COPY_DONE (1<<(RXf_BASE_SHIFT+16)) |
339 | #define RXf_TAINTED_SEEN (1<<(RXf_BASE_SHIFT+17)) | |
340 | #define RXf_TAINTED (1<<(RXf_BASE_SHIFT+18)) /* this pattern is tainted */ | |
52d81aa8 NC |
341 | |
342 | /* Flags indicating special patterns */ | |
e795e964 KW |
343 | #define RXf_START_ONLY (1<<(RXf_BASE_SHIFT+19)) /* Pattern is /^/ */ |
344 | #define RXf_SKIPWHITE (1<<(RXf_BASE_SHIFT+20)) /* Pattern is for a split / / */ | |
345 | #define RXf_WHITE (1<<(RXf_BASE_SHIFT+21)) /* Pattern is /\s+/ */ | |
346 | #define RXf_NULL (1<<(RXf_BASE_SHIFT+22)) /* Pattern is // */ | |
347 | #if RXf_BASE_SHIFT+22 > 31 | |
1850c8f9 KW |
348 | # error Too many RXf_PMf bits used. See regnodes.h for any spare in middle |
349 | #endif | |
c737faaf | 350 | |
cb5027f2 YO |
351 | /* |
352 | * NOTE: if you modify any RXf flags you should run regen.pl or regcomp.pl | |
7948fc08 | 353 | * so that regnodes.h is updated with the changes. |
cb5027f2 YO |
354 | * |
355 | */ | |
bbe252da | 356 | |
bbe252da | 357 | #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN) |
07bc277f NC |
358 | #define RXp_MATCH_TAINTED(prog) (RXp_EXTFLAGS(prog) & RXf_TAINTED_SEEN) |
359 | #define RX_MATCH_TAINTED(prog) (RX_EXTFLAGS(prog) & RXf_TAINTED_SEEN) | |
360 | #define RX_MATCH_TAINTED_on(prog) (RX_EXTFLAGS(prog) |= RXf_TAINTED_SEEN) | |
361 | #define RX_MATCH_TAINTED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_TAINTED_SEEN) | |
72311751 GS |
362 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
363 | ? RX_MATCH_TAINTED_on(prog) \ | |
364 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 365 | |
07bc277f NC |
366 | #define RXp_MATCH_COPIED(prog) (RXp_EXTFLAGS(prog) & RXf_COPY_DONE) |
367 | #define RX_MATCH_COPIED(prog) (RX_EXTFLAGS(prog) & RXf_COPY_DONE) | |
368 | #define RXp_MATCH_COPIED_on(prog) (RXp_EXTFLAGS(prog) |= RXf_COPY_DONE) | |
369 | #define RX_MATCH_COPIED_on(prog) (RX_EXTFLAGS(prog) |= RXf_COPY_DONE) | |
370 | #define RXp_MATCH_COPIED_off(prog) (RXp_EXTFLAGS(prog) &= ~RXf_COPY_DONE) | |
371 | #define RX_MATCH_COPIED_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_COPY_DONE) | |
cf93c79d IZ |
372 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
373 | ? RX_MATCH_COPIED_on(prog) \ | |
374 | : RX_MATCH_COPIED_off(prog)) | |
f9f4320a | 375 | |
07bc277f | 376 | #define RXp_EXTFLAGS(rx) ((rx)->extflags) |
866c78d1 | 377 | |
07bc277f | 378 | /* For source compatibility. We used to store these explicitly. */ |
f7c278bf | 379 | #define RX_PRECOMP(prog) (RX_WRAPPED(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) |
d2c6dc5e | 380 | #define RX_PRECOMP_const(prog) (RX_WRAPPED_const(prog) + ((struct regexp *)SvANY(prog))->pre_prefix) |
5509d87a NC |
381 | /* FIXME? Are we hardcoding too much here and constraining plugin extension |
382 | writers? Specifically, the value 1 assumes that the wrapped version always | |
383 | has exactly one character at the end, a ')'. Will that always be true? */ | |
384 | #define RX_PRELEN(prog) (RX_WRAPLEN(prog) - ((struct regexp *)SvANY(prog))->pre_prefix - 1) | |
f7c278bf | 385 | #define RX_WRAPPED(prog) SvPVX(prog) |
d2c6dc5e | 386 | #define RX_WRAPPED_const(prog) SvPVX_const(prog) |
9d17798d | 387 | #define RX_WRAPLEN(prog) SvCUR(prog) |
288b8c02 | 388 | #define RX_CHECK_SUBSTR(prog) (((struct regexp *)SvANY(prog))->check_substr) |
288b8c02 | 389 | #define RX_REFCNT(prog) SvREFCNT(prog) |
7fe44985 | 390 | #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) |
4450b630 NC |
391 | # define RX_EXTFLAGS(prog) \ |
392 | (*({ \ | |
7948fc08 RU |
393 | const REGEXP *const _rx_extflags = (prog); \ |
394 | assert(SvTYPE(_rx_extflags) == SVt_REGEXP); \ | |
395 | &RXp_EXTFLAGS(SvANY(_rx_extflags)); \ | |
4450b630 NC |
396 | })) |
397 | # define RX_ENGINE(prog) \ | |
398 | (*({ \ | |
7948fc08 RU |
399 | const REGEXP *const _rx_engine = (prog); \ |
400 | assert(SvTYPE(_rx_engine) == SVt_REGEXP); \ | |
401 | &SvANY(_rx_engine)->engine; \ | |
4450b630 NC |
402 | })) |
403 | # define RX_SUBBEG(prog) \ | |
404 | (*({ \ | |
7948fc08 RU |
405 | const REGEXP *const _rx_subbeg = (prog); \ |
406 | assert(SvTYPE(_rx_subbeg) == SVt_REGEXP); \ | |
407 | &SvANY(_rx_subbeg)->subbeg; \ | |
4450b630 | 408 | })) |
200ff815 | 409 | # define RX_OFFS(prog) \ |
4450b630 | 410 | (*({ \ |
7948fc08 RU |
411 | const REGEXP *const _rx_offs = (prog); \ |
412 | assert(SvTYPE(_rx_offs) == SVt_REGEXP); \ | |
413 | &SvANY(_rx_offs)->offs; \ | |
4450b630 NC |
414 | })) |
415 | # define RX_NPARENS(prog) \ | |
416 | (*({ \ | |
7948fc08 RU |
417 | const REGEXP *const _rx_nparens = (prog); \ |
418 | assert(SvTYPE(_rx_nparens) == SVt_REGEXP); \ | |
419 | &SvANY(_rx_nparens)->nparens; \ | |
4450b630 NC |
420 | })) |
421 | #else | |
422 | # define RX_EXTFLAGS(prog) RXp_EXTFLAGS((struct regexp *)SvANY(prog)) | |
423 | # define RX_ENGINE(prog) (((struct regexp *)SvANY(prog))->engine) | |
424 | # define RX_SUBBEG(prog) (((struct regexp *)SvANY(prog))->subbeg) | |
425 | # define RX_OFFS(prog) (((struct regexp *)SvANY(prog))->offs) | |
426 | # define RX_NPARENS(prog) (((struct regexp *)SvANY(prog))->nparens) | |
427 | #endif | |
288b8c02 | 428 | #define RX_SUBLEN(prog) (((struct regexp *)SvANY(prog))->sublen) |
288b8c02 NC |
429 | #define RX_MINLEN(prog) (((struct regexp *)SvANY(prog))->minlen) |
430 | #define RX_MINLENRET(prog) (((struct regexp *)SvANY(prog))->minlenret) | |
431 | #define RX_GOFS(prog) (((struct regexp *)SvANY(prog))->gofs) | |
432 | #define RX_LASTPAREN(prog) (((struct regexp *)SvANY(prog))->lastparen) | |
433 | #define RX_LASTCLOSEPAREN(prog) (((struct regexp *)SvANY(prog))->lastcloseparen) | |
434 | #define RX_SEEN_EVALS(prog) (((struct regexp *)SvANY(prog))->seen_evals) | |
bdd9a1b1 | 435 | #define RX_SAVED_COPY(prog) (((struct regexp *)SvANY(prog))->saved_copy) |
220fc49f | 436 | |
be8e71aa YO |
437 | #endif /* PLUGGABLE_RE_EXTENSION */ |
438 | ||
439 | /* Stuff that needs to be included in the plugable extension goes below here */ | |
440 | ||
f8c7b90f | 441 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 | 442 | #define RX_MATCH_COPY_FREE(rx) \ |
bdd9a1b1 NC |
443 | STMT_START {if (RX_SAVED_COPY(rx)) { \ |
444 | SV_CHECK_THINKFIRST_COW_DROP(RX_SAVED_COPY(rx)); \ | |
ed252734 NC |
445 | } \ |
446 | if (RX_MATCH_COPIED(rx)) { \ | |
07bc277f | 447 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 NC |
448 | RX_MATCH_COPIED_off(rx); \ |
449 | }} STMT_END | |
450 | #else | |
451 | #define RX_MATCH_COPY_FREE(rx) \ | |
452 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ | |
07bc277f | 453 | Safefree(RX_SUBBEG(rx)); \ |
ed252734 NC |
454 | RX_MATCH_COPIED_off(rx); \ |
455 | }} STMT_END | |
456 | #endif | |
457 | ||
07bc277f NC |
458 | #define RXp_MATCH_UTF8(prog) (RXp_EXTFLAGS(prog) & RXf_MATCH_UTF8) |
459 | #define RX_MATCH_UTF8(prog) (RX_EXTFLAGS(prog) & RXf_MATCH_UTF8) | |
460 | #define RX_MATCH_UTF8_on(prog) (RX_EXTFLAGS(prog) |= RXf_MATCH_UTF8) | |
461 | #define RX_MATCH_UTF8_off(prog) (RX_EXTFLAGS(prog) &= ~RXf_MATCH_UTF8) | |
a30b2f1f RGS |
462 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
463 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
464 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
efd26800 NC |
465 | |
466 | /* Whether the pattern stored at RX_WRAPPED is in UTF-8 */ | |
8f6ae13c | 467 | #define RX_UTF8(prog) SvUTF8(prog) |
7948fc08 | 468 | |
cad2e5aa JH |
469 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
470 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
471 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
472 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 473 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 | 474 | |
7fe44985 | 475 | #if defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) |
288b8c02 NC |
476 | # define ReREFCNT_inc(re) \ |
477 | ({ \ | |
478 | /* This is here to generate a casting warning if incorrect. */ \ | |
7948fc08 RU |
479 | REGEXP *const _rerefcnt_inc = (re); \ |
480 | assert(SvTYPE(_rerefcnt_inc) == SVt_REGEXP); \ | |
481 | SvREFCNT_inc(_rerefcnt_inc); \ | |
482 | _rerefcnt_inc; \ | |
288b8c02 NC |
483 | }) |
484 | # define ReREFCNT_dec(re) \ | |
485 | ({ \ | |
486 | /* This is here to generate a casting warning if incorrect. */ \ | |
7948fc08 RU |
487 | REGEXP *const _rerefcnt_dec = (re); \ |
488 | SvREFCNT_dec(_rerefcnt_dec); \ | |
288b8c02 NC |
489 | }) |
490 | #else | |
491 | # define ReREFCNT_dec(re) SvREFCNT_dec(re) | |
b7e0bd9e | 492 | # define ReREFCNT_inc(re) ((REGEXP *) SvREFCNT_inc(re)) |
288b8c02 NC |
493 | #endif |
494 | ||
495 | /* FIXME for plugins. */ | |
cf93c79d IZ |
496 | |
497 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
498 | #define FBMcf_TAIL_DOLLARM 2 |
499 | #define FBMcf_TAIL_Z 4 | |
500 | #define FBMcf_TAIL_z 8 | |
501 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
502 | |
503 | #define FBMrf_MULTILINE 1 | |
cad2e5aa | 504 | |
3b0527fe DM |
505 | /* some basic information about the current match that is created by |
506 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ | |
507 | ||
508 | typedef struct { | |
288b8c02 | 509 | REGEXP *prog; |
3b0527fe DM |
510 | char *bol; |
511 | char *till; | |
512 | SV *sv; | |
513 | char *ganch; | |
24b23f37 | 514 | char *cutpoint; |
3b0527fe DM |
515 | } regmatch_info; |
516 | ||
5d9a96ca DM |
517 | |
518 | /* structures for holding and saving the state maintained by regmatch() */ | |
519 | ||
d56b3014 | 520 | #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH |
4b196cd4 | 521 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000 |
d56b3014 | 522 | #endif |
6bda09f9 | 523 | |
5d9a96ca DM |
524 | typedef I32 CHECKPOINT; |
525 | ||
a0374537 | 526 | typedef struct regmatch_state { |
40a82448 | 527 | int resume_state; /* where to jump to on return */ |
24d3c4a9 | 528 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 DM |
529 | |
530 | union { | |
77cb431f DM |
531 | |
532 | /* this is a fake union member that matches the first element | |
533 | * of each member that needs to store positive backtrack | |
534 | * information */ | |
535 | struct { | |
536 | struct regmatch_state *prev_yes_state; | |
537 | } yes; | |
538 | ||
fae667d5 YO |
539 | /* branchlike members */ |
540 | /* this is a fake union member that matches the first elements | |
541 | * of each member that needs to behave like a branch */ | |
542 | struct { | |
5d458dd8 YO |
543 | /* this first element must match u.yes */ |
544 | struct regmatch_state *prev_yes_state; | |
fae667d5 YO |
545 | U32 lastparen; |
546 | CHECKPOINT cp; | |
547 | ||
548 | } branchlike; | |
549 | ||
550 | struct { | |
551 | /* the first elements must match u.branchlike */ | |
552 | struct regmatch_state *prev_yes_state; | |
553 | U32 lastparen; | |
554 | CHECKPOINT cp; | |
555 | ||
556 | regnode *next_branch; /* next branch node */ | |
557 | } branch; | |
558 | ||
559 | struct { | |
560 | /* the first elements must match u.branchlike */ | |
561 | struct regmatch_state *prev_yes_state; | |
562 | U32 lastparen; | |
563 | CHECKPOINT cp; | |
564 | ||
2e64971a | 565 | U32 accepted; /* how many accepting states left */ |
7f69552c | 566 | U16 *jump; /* positive offsets from me */ |
166ba7cd | 567 | regnode *B; /* node following the trie */ |
7f69552c | 568 | regnode *me; /* Which node am I - needed for jump tries*/ |
2e64971a DM |
569 | U8 *firstpos;/* pos in string of first trie match */ |
570 | U32 firstchars;/* len in chars of firstpos from start */ | |
571 | U16 nextword;/* next word to try */ | |
572 | U16 topword; /* longest accepted word */ | |
573 | bool longfold;/* saw a fold with a 1->n char mapping */ | |
e822a8b4 DM |
574 | } trie; |
575 | ||
fae667d5 YO |
576 | /* special types - these members are used to store state for special |
577 | regops like eval, if/then, lookaround and the markpoint state */ | |
e822a8b4 | 578 | struct { |
77cb431f DM |
579 | /* this first element must match u.yes */ |
580 | struct regmatch_state *prev_yes_state; | |
faec1544 DM |
581 | struct regmatch_state *prev_eval; |
582 | struct regmatch_state *prev_curlyx; | |
288b8c02 | 583 | REGEXP *prev_rex; |
faec1544 DM |
584 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
585 | flip when transitioning between | |
586 | inner and outer rexen */ | |
aa283a38 DM |
587 | CHECKPOINT cp; /* remember current savestack indexes */ |
588 | CHECKPOINT lastcp; | |
6bda09f9 | 589 | U32 close_paren; /* which close bracket is our end */ |
1b5c067c | 590 | regnode *B; /* the node following us */ |
e822a8b4 DM |
591 | } eval; |
592 | ||
593 | struct { | |
c476f425 | 594 | /* this first element must match u.yes */ |
262b90c4 | 595 | struct regmatch_state *prev_yes_state; |
fae667d5 YO |
596 | I32 wanted; |
597 | I32 logical; /* saved copy of 'logical' var */ | |
598 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ | |
599 | } ifmatch; /* and SUSPEND/UNLESSM */ | |
600 | ||
601 | struct { | |
602 | /* this first element must match u.yes */ | |
603 | struct regmatch_state *prev_yes_state; | |
604 | struct regmatch_state *prev_mark; | |
605 | SV* mark_name; | |
606 | char *mark_loc; | |
607 | } mark; | |
608 | ||
609 | struct { | |
610 | int val; | |
611 | } keeper; | |
612 | ||
613 | /* quantifiers - these members are used for storing state for | |
614 | for the regops used to implement quantifiers */ | |
615 | struct { | |
616 | /* this first element must match u.yes */ | |
617 | struct regmatch_state *prev_yes_state; | |
c476f425 | 618 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
d02d6d97 DM |
619 | regnode *me; /* the CURLYX node */ |
620 | regnode *B; /* the B node in /A*B/ */ | |
c476f425 DM |
621 | CHECKPOINT cp; /* remember current savestack index */ |
622 | bool minmod; | |
a0374537 | 623 | int parenfloor;/* how far back to strip paren data */ |
c476f425 DM |
624 | |
625 | /* these two are modified by WHILEM */ | |
626 | int count; /* how many instances of A we've matched */ | |
627 | char *lastloc;/* where previous A matched (0-len detect) */ | |
e822a8b4 DM |
628 | } curlyx; |
629 | ||
630 | struct { | |
c476f425 | 631 | /* this first element must match u.yes */ |
262b90c4 | 632 | struct regmatch_state *prev_yes_state; |
c476f425 DM |
633 | struct regmatch_state *save_curlyx; |
634 | CHECKPOINT cp; /* remember current savestack indexes */ | |
635 | CHECKPOINT lastcp; | |
636 | char *save_lastloc; /* previous curlyx.lastloc */ | |
637 | I32 cache_offset; | |
638 | I32 cache_mask; | |
e822a8b4 DM |
639 | } whilem; |
640 | ||
641 | struct { | |
5d458dd8 YO |
642 | /* this first element must match u.yes */ |
643 | struct regmatch_state *prev_yes_state; | |
e822a8b4 | 644 | I32 c1, c2; /* case fold search */ |
40a82448 DM |
645 | CHECKPOINT cp; |
646 | I32 alen; /* length of first-matched A string */ | |
647 | I32 count; | |
0cadcf80 | 648 | bool minmod; |
40a82448 DM |
649 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
650 | regnode *me; /* the curlym node */ | |
e822a8b4 DM |
651 | } curlym; |
652 | ||
653 | struct { | |
3b6647e0 | 654 | U32 paren; |
c255a977 | 655 | CHECKPOINT cp; |
e822a8b4 | 656 | I32 c1, c2; /* case fold search */ |
c255a977 DM |
657 | char *maxpos; /* highest possible point in string to match */ |
658 | char *oldloc; /* the previous locinput */ | |
e822a8b4 | 659 | int count; |
c255a977 DM |
660 | int min, max; /* {m,n} */ |
661 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ | |
662 | } curly; /* and CURLYN/PLUS/STAR */ | |
dad79028 | 663 | |
d8319b27 | 664 | } u; |
5d9a96ca DM |
665 | } regmatch_state; |
666 | ||
667 | /* how many regmatch_state structs to allocate as a single slab. | |
668 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev | |
669 | * pointers, plus 1 for any mythical malloc overhead. */ | |
670 | ||
671 | #define PERL_REGMATCH_SLAB_SLOTS \ | |
672 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) | |
673 | ||
674 | typedef struct regmatch_slab { | |
675 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; | |
676 | struct regmatch_slab *prev, *next; | |
677 | } regmatch_slab; | |
1ade1aa1 | 678 | |
46ab3289 NC |
679 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
680 | #define PL_bostr PL_reg_state.re_state_bostr | |
681 | #define PL_reginput PL_reg_state.re_state_reginput | |
46ab3289 | 682 | #define PL_regeol PL_reg_state.re_state_regeol |
f0ab9afb | 683 | #define PL_regoffs PL_reg_state.re_state_regoffs |
46ab3289 NC |
684 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
685 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen | |
46ab3289 NC |
686 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
687 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl | |
688 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set | |
46ab3289 NC |
689 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
690 | #define PL_reg_magic PL_reg_state.re_state_reg_magic | |
691 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos | |
692 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm | |
693 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm | |
694 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved | |
695 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen | |
696 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter | |
697 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter | |
698 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache | |
699 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size | |
700 | #define PL_regsize PL_reg_state.re_state_regsize | |
701 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry | |
702 | #define PL_nrs PL_reg_state.re_state_nrs | |
703 | ||
1ade1aa1 NC |
704 | struct re_save_state { |
705 | U32 re_state_reg_flags; /* from regexec.c */ | |
1b5c067c NC |
706 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
707 | I32 re_state_reg_eval_set; /* from regexec.c */ | |
708 | bool re_state_reg_match_utf8; /* from regexec.c */ | |
1ade1aa1 NC |
709 | char *re_state_bostr; |
710 | char *re_state_reginput; /* String-input pointer. */ | |
1ade1aa1 | 711 | char *re_state_regeol; /* End of input, for $ check. */ |
f0ab9afb | 712 | regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */ |
1ade1aa1 NC |
713 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
714 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ | |
1ade1aa1 | 715 | char **re_state_reg_start_tmp; /* from regexec.c */ |
1ade1aa1 | 716 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
1ade1aa1 NC |
717 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
718 | PMOP *re_state_reg_curpm; /* from regexec.c */ | |
719 | char *re_state_reg_oldsaved; /* old saved substr during match */ | |
720 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ | |
1b5c067c NC |
721 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
722 | I32 re_state_reg_oldpos; /* from regexec.c */ | |
1ade1aa1 NC |
723 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
724 | I32 re_state_reg_leftiter; /* wait until caching pos */ | |
3b6647e0 | 725 | U32 re_state_regsize; /* from regexec.c */ |
1b5c067c | 726 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
1ade1aa1 NC |
727 | char *re_state_reg_starttry; /* from regexec.c */ |
728 | #ifdef PERL_OLD_COPY_ON_WRITE | |
46ab3289 | 729 | 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 |
730 | #endif |
731 | }; | |
732 | ||
733 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ | |
734 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) | |
cde0cee5 | 735 | |
1ade1aa1 NC |
736 | /* |
737 | * Local variables: | |
738 | * c-indentation-style: bsd | |
739 | * c-basic-offset: 4 | |
740 | * indent-tabs-mode: t | |
741 | * End: | |
742 | * | |
743 | * ex: set ts=8 sts=4 sw=4 noet: | |
744 | */ |