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