Commit | Line | Data |
a0d0e21e |
1 | /* regexp.h |
d6376244 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
4 | * 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 |
18 | /* we don't want to include this stuff if we are inside Nicholas' |
19 | * pluggable regex engine code */ |
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; |
34 | |
378cc40b |
35 | typedef struct regexp { |
cf93c79d |
36 | I32 *startp; |
37 | I32 *endp; |
c277df42 |
38 | regnode *regstclass; |
2779dcf1 |
39 | struct reg_substr_data *substrs; |
cf93c79d |
40 | char *precomp; /* pre-compilation regular expression */ |
c277df42 |
41 | struct reg_data *data; /* Additional data. */ |
cf93c79d |
42 | char *subbeg; /* saved or original string |
43 | so \digit works forever. */ |
f8c7b90f |
44 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 |
45 | SV *saved_copy; /* If non-NULL, SV which is COW from original */ |
46 | #endif |
fac92740 |
47 | U32 *offsets; /* offset annotations 20001228 MJD */ |
cf93c79d |
48 | I32 sublen; /* Length of string pointed by subbeg */ |
49 | I32 refcnt; |
50 | I32 minlen; /* mininum possible length of $& */ |
51 | I32 prelen; /* length of precomp */ |
52 | U32 nparens; /* number of parentheses */ |
53 | U32 lastparen; /* last paren matched */ |
a01268b5 |
54 | U32 lastcloseparen; /* last paren matched */ |
cf93c79d |
55 | U32 reganch; /* Internal use only + |
56 | Tainted information used by regexec? */ |
81714fb9 |
57 | HV *paren_names; /* Paren names */ |
58 | const struct regexp_engine* engine; |
c277df42 |
59 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
378cc40b |
60 | } regexp; |
61 | |
f9f4320a |
62 | |
63 | typedef struct re_scream_pos_data_s |
64 | { |
65 | char **scream_olds; /* match pos */ |
66 | I32 *scream_pos; /* Internal iterator of scream. */ |
67 | } re_scream_pos_data; |
68 | |
69 | typedef struct regexp_engine { |
f2f78491 |
70 | regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm); |
71 | I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend, |
f9f4320a |
72 | char* strbeg, I32 minend, SV* screamer, |
73 | void* data, U32 flags); |
f2f78491 |
74 | char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos, |
75 | char *strend, U32 flags, |
76 | struct re_scream_pos_data_s *data); |
77 | SV* (*checkstr) (pTHX_ regexp *prog); |
78 | void (*free) (pTHX_ struct regexp* r); |
79 | #ifdef USE_ITHREADS |
80 | regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param); |
f9f4320a |
81 | #endif |
82 | } regexp_engine; |
83 | |
cad2e5aa |
84 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
85 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) |
f2278c82 |
86 | #define ROPT_ANCH_BOL 0x00000001 |
87 | #define ROPT_ANCH_MBOL 0x00000002 |
88 | #define ROPT_ANCH_SBOL 0x00000004 |
89 | #define ROPT_ANCH_GPOS 0x00000008 |
90 | #define ROPT_SKIP 0x00000010 |
91 | #define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */ |
92 | #define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */ |
93 | #define ROPT_GPOS_SEEN 0x00000080 |
94 | #define ROPT_CHECK_ALL 0x00000100 |
95 | #define ROPT_LOOKBEHIND_SEEN 0x00000200 |
96 | #define ROPT_EVAL_SEEN 0x00000400 |
97 | #define ROPT_CANY_SEEN 0x00000800 |
cce850e4 |
98 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ |
f9f4320a |
99 | #define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS) |
6bda09f9 |
100 | #define ROPT_RECURSE_SEEN 0x00001000 |
a0ed51b3 |
101 | |
8782bef2 |
102 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
c277df42 |
103 | |
f2278c82 |
104 | #define ROPT_UTF8 0x00010000 |
105 | #define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */ |
106 | #define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */ |
107 | #define ROPT_TAINTED_SEEN 0x00080000 |
a30b2f1f |
108 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
a0ed51b3 |
109 | |
f2278c82 |
110 | #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */ |
111 | #define RE_USE_INTUIT_ML 0x00200000 |
112 | #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */ |
be8e71aa |
113 | #define REINT_AUTORITATIVE_ML 0x00800000 |
f2278c82 |
114 | #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */ |
115 | #define REINT_ONCE_ML 0x02000000 |
116 | #define RE_INTUIT_ONECHAR 0x04000000 |
117 | #define RE_INTUIT_TAIL 0x08000000 |
118 | |
cad2e5aa |
119 | |
120 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) |
121 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) |
122 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) |
123 | |
c277df42 |
124 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
72311751 |
125 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
126 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) |
127 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
128 | ? RX_MATCH_TAINTED_on(prog) \ |
129 | : RX_MATCH_TAINTED_off(prog)) |
c277df42 |
130 | |
cf93c79d |
131 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
132 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) |
133 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) |
134 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
135 | ? RX_MATCH_COPIED_on(prog) \ |
136 | : RX_MATCH_COPIED_off(prog)) |
f9f4320a |
137 | |
be8e71aa |
138 | #endif /* PLUGGABLE_RE_EXTENSION */ |
139 | |
140 | /* Stuff that needs to be included in the plugable extension goes below here */ |
141 | |
142 | #define RE_DEBUG_BIT 0x20000000 |
143 | #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT) |
144 | #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT) |
cf93c79d |
145 | |
f8c7b90f |
146 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 |
147 | #define RX_MATCH_COPY_FREE(rx) \ |
148 | STMT_START {if (rx->saved_copy) { \ |
149 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ |
150 | } \ |
151 | if (RX_MATCH_COPIED(rx)) { \ |
152 | Safefree(rx->subbeg); \ |
153 | RX_MATCH_COPIED_off(rx); \ |
154 | }} STMT_END |
155 | #else |
156 | #define RX_MATCH_COPY_FREE(rx) \ |
157 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
158 | Safefree(rx->subbeg); \ |
159 | RX_MATCH_COPIED_off(rx); \ |
160 | }} STMT_END |
161 | #endif |
162 | |
a30b2f1f |
163 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
164 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) |
165 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) |
166 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
167 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
168 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
169 | |
cad2e5aa |
170 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
171 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
172 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
173 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
cf93c79d |
174 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 |
175 | |
155aba94 |
176 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
f9f4320a |
177 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
cf93c79d |
178 | |
179 | #define FBMcf_TAIL_DOLLAR 1 |
cad2e5aa |
180 | #define FBMcf_TAIL_DOLLARM 2 |
181 | #define FBMcf_TAIL_Z 4 |
182 | #define FBMcf_TAIL_z 8 |
183 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
cf93c79d |
184 | |
185 | #define FBMrf_MULTILINE 1 |
cad2e5aa |
186 | |
187 | struct re_scream_pos_data_s; |
5d9a96ca |
188 | |
189 | /* an accepting state/position*/ |
190 | struct _reg_trie_accepted { |
191 | U8 *endpos; |
192 | U16 wordnum; |
193 | }; |
194 | typedef struct _reg_trie_accepted reg_trie_accepted; |
195 | |
3b0527fe |
196 | /* some basic information about the current match that is created by |
197 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ |
198 | |
199 | typedef struct { |
200 | regexp *prog; |
201 | char *bol; |
202 | char *till; |
203 | SV *sv; |
204 | char *ganch; |
24b23f37 |
205 | char *cutpoint; |
3b0527fe |
206 | } regmatch_info; |
207 | |
5d9a96ca |
208 | |
209 | /* structures for holding and saving the state maintained by regmatch() */ |
210 | |
6bda09f9 |
211 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50 |
212 | |
5d9a96ca |
213 | typedef I32 CHECKPOINT; |
214 | |
a0374537 |
215 | typedef struct regmatch_state { |
40a82448 |
216 | int resume_state; /* where to jump to on return */ |
24d3c4a9 |
217 | char *locinput; /* where to backtrack in string on failure */ |
e822a8b4 |
218 | |
219 | union { |
77cb431f |
220 | |
221 | /* this is a fake union member that matches the first element |
222 | * of each member that needs to store positive backtrack |
223 | * information */ |
224 | struct { |
225 | struct regmatch_state *prev_yes_state; |
226 | } yes; |
227 | |
e822a8b4 |
228 | struct { |
229 | reg_trie_accepted *accept_buff; |
166ba7cd |
230 | U32 accepted; /* how many accepting states we have seen */ |
7f69552c |
231 | U16 *jump; /* positive offsets from me */ |
166ba7cd |
232 | regnode *B; /* node following the trie */ |
7f69552c |
233 | regnode *me; /* Which node am I - needed for jump tries*/ |
e822a8b4 |
234 | } trie; |
235 | |
236 | struct { |
77cb431f |
237 | /* this first element must match u.yes */ |
238 | struct regmatch_state *prev_yes_state; |
faec1544 |
239 | struct regmatch_state *prev_eval; |
240 | struct regmatch_state *prev_curlyx; |
aa283a38 |
241 | regexp *prev_rex; |
faec1544 |
242 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
243 | flip when transitioning between |
244 | inner and outer rexen */ |
aa283a38 |
245 | CHECKPOINT cp; /* remember current savestack indexes */ |
246 | CHECKPOINT lastcp; |
40a82448 |
247 | regnode *B; /* the node following us */ |
6bda09f9 |
248 | U32 close_paren; /* which close bracket is our end */ |
e822a8b4 |
249 | } eval; |
250 | |
251 | struct { |
c476f425 |
252 | /* this first element must match u.yes */ |
262b90c4 |
253 | struct regmatch_state *prev_yes_state; |
c476f425 |
254 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
255 | CHECKPOINT cp; /* remember current savestack index */ |
256 | bool minmod; |
a0374537 |
257 | int parenfloor;/* how far back to strip paren data */ |
c476f425 |
258 | int min; /* the minimal number of A's to match */ |
259 | int max; /* the maximal number of A's to match */ |
260 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
261 | |
262 | /* these two are modified by WHILEM */ |
263 | int count; /* how many instances of A we've matched */ |
264 | char *lastloc;/* where previous A matched (0-len detect) */ |
e822a8b4 |
265 | } curlyx; |
266 | |
267 | struct { |
c476f425 |
268 | /* this first element must match u.yes */ |
262b90c4 |
269 | struct regmatch_state *prev_yes_state; |
c476f425 |
270 | struct regmatch_state *save_curlyx; |
271 | CHECKPOINT cp; /* remember current savestack indexes */ |
272 | CHECKPOINT lastcp; |
273 | char *save_lastloc; /* previous curlyx.lastloc */ |
274 | I32 cache_offset; |
275 | I32 cache_mask; |
e822a8b4 |
276 | } whilem; |
277 | |
278 | struct { |
40a82448 |
279 | I32 lastparen; |
280 | regnode *next_branch; /* next branch node */ |
281 | CHECKPOINT cp; |
282 | } branch; |
283 | |
284 | struct { |
dad79028 |
285 | /* this first element must match u.yes */ |
286 | struct regmatch_state *prev_yes_state; |
e822a8b4 |
287 | I32 c1, c2; /* case fold search */ |
40a82448 |
288 | CHECKPOINT cp; |
289 | I32 alen; /* length of first-matched A string */ |
290 | I32 count; |
0cadcf80 |
291 | bool minmod; |
40a82448 |
292 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
293 | regnode *me; /* the curlym node */ |
e822a8b4 |
294 | } curlym; |
295 | |
296 | struct { |
297 | I32 paren; |
c255a977 |
298 | CHECKPOINT cp; |
e822a8b4 |
299 | I32 c1, c2; /* case fold search */ |
c255a977 |
300 | char *maxpos; /* highest possible point in string to match */ |
301 | char *oldloc; /* the previous locinput */ |
e822a8b4 |
302 | int count; |
c255a977 |
303 | int min, max; /* {m,n} */ |
304 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
305 | } curly; /* and CURLYN/PLUS/STAR */ |
dad79028 |
306 | |
307 | struct { |
308 | /* this first element must match u.yes */ |
309 | struct regmatch_state *prev_yes_state; |
310 | I32 wanted; |
24d3c4a9 |
311 | I32 logical; /* saved copy of 'logical' var */ |
40a82448 |
312 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ |
dad79028 |
313 | } ifmatch; /* and SUSPEND/UNLESSM */ |
d8319b27 |
314 | } u; |
5d9a96ca |
315 | } regmatch_state; |
316 | |
317 | /* how many regmatch_state structs to allocate as a single slab. |
318 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
319 | * pointers, plus 1 for any mythical malloc overhead. */ |
320 | |
321 | #define PERL_REGMATCH_SLAB_SLOTS \ |
322 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
323 | |
324 | typedef struct regmatch_slab { |
325 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
326 | struct regmatch_slab *prev, *next; |
327 | } regmatch_slab; |
1ade1aa1 |
328 | |
46ab3289 |
329 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
330 | #define PL_bostr PL_reg_state.re_state_bostr |
331 | #define PL_reginput PL_reg_state.re_state_reginput |
46ab3289 |
332 | #define PL_regeol PL_reg_state.re_state_regeol |
333 | #define PL_regstartp PL_reg_state.re_state_regstartp |
334 | #define PL_regendp PL_reg_state.re_state_regendp |
335 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
336 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen |
46ab3289 |
337 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
338 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl |
339 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set |
46ab3289 |
340 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
341 | #define PL_reg_magic PL_reg_state.re_state_reg_magic |
342 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos |
343 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm |
344 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm |
345 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved |
346 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen |
347 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter |
348 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter |
349 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache |
350 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size |
351 | #define PL_regsize PL_reg_state.re_state_regsize |
352 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry |
353 | #define PL_nrs PL_reg_state.re_state_nrs |
354 | |
1ade1aa1 |
355 | struct re_save_state { |
356 | U32 re_state_reg_flags; /* from regexec.c */ |
357 | char *re_state_bostr; |
358 | char *re_state_reginput; /* String-input pointer. */ |
1ade1aa1 |
359 | char *re_state_regeol; /* End of input, for $ check. */ |
360 | I32 *re_state_regstartp; /* Pointer to startp array. */ |
361 | I32 *re_state_regendp; /* Ditto for endp. */ |
362 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
363 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ |
1ade1aa1 |
364 | char **re_state_reg_start_tmp; /* from regexec.c */ |
365 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
366 | I32 re_state_reg_eval_set; /* from regexec.c */ |
1ade1aa1 |
367 | bool re_state_reg_match_utf8; /* from regexec.c */ |
368 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
369 | I32 re_state_reg_oldpos; /* from regexec.c */ |
370 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
371 | PMOP *re_state_reg_curpm; /* from regexec.c */ |
372 | char *re_state_reg_oldsaved; /* old saved substr during match */ |
373 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ |
374 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
375 | I32 re_state_reg_leftiter; /* wait until caching pos */ |
376 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
377 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
378 | I32 re_state_regsize; /* from regexec.c */ |
379 | char *re_state_reg_starttry; /* from regexec.c */ |
380 | #ifdef PERL_OLD_COPY_ON_WRITE |
46ab3289 |
381 | 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 |
382 | #endif |
383 | }; |
384 | |
385 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ |
386 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) |
387 | /* |
388 | * Local variables: |
389 | * c-indentation-style: bsd |
390 | * c-basic-offset: 4 |
391 | * indent-tabs-mode: t |
392 | * End: |
393 | * |
394 | * ex: set ts=8 sts=4 sw=4 noet: |
395 | */ |