Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* regexp.h |
d6376244 | 2 | * |
4bb101f2 JH |
3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
4 | * 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 | */ | |
17 | ||
378cc40b | 18 | |
c277df42 IZ |
19 | struct regnode { |
20 | U8 flags; | |
21 | U8 type; | |
22 | U16 next_off; | |
23 | }; | |
24 | ||
25 | typedef struct regnode regnode; | |
26 | ||
cad2e5aa | 27 | struct reg_substr_data; |
2779dcf1 | 28 | |
0ee3c8d0 JH |
29 | struct reg_data; |
30 | ||
378cc40b | 31 | typedef struct regexp { |
cf93c79d IZ |
32 | I32 *startp; |
33 | I32 *endp; | |
c277df42 | 34 | regnode *regstclass; |
2779dcf1 | 35 | struct reg_substr_data *substrs; |
cf93c79d | 36 | char *precomp; /* pre-compilation regular expression */ |
c277df42 | 37 | struct reg_data *data; /* Additional data. */ |
cf93c79d IZ |
38 | char *subbeg; /* saved or original string |
39 | so \digit works forever. */ | |
f8c7b90f | 40 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
41 | SV *saved_copy; /* If non-NULL, SV which is COW from original */ |
42 | #endif | |
fac92740 | 43 | U32 *offsets; /* offset annotations 20001228 MJD */ |
cf93c79d IZ |
44 | I32 sublen; /* Length of string pointed by subbeg */ |
45 | I32 refcnt; | |
46 | I32 minlen; /* mininum possible length of $& */ | |
47 | I32 prelen; /* length of precomp */ | |
48 | U32 nparens; /* number of parentheses */ | |
49 | U32 lastparen; /* last paren matched */ | |
a01268b5 | 50 | U32 lastcloseparen; /* last paren matched */ |
cf93c79d IZ |
51 | U32 reganch; /* Internal use only + |
52 | Tainted information used by regexec? */ | |
c277df42 | 53 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
378cc40b LW |
54 | } regexp; |
55 | ||
cad2e5aa JH |
56 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
57 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) | |
a0ed51b3 LW |
58 | #define ROPT_ANCH_BOL 0x00001 |
59 | #define ROPT_ANCH_MBOL 0x00002 | |
cad2e5aa JH |
60 | #define ROPT_ANCH_SBOL 0x00004 |
61 | #define ROPT_ANCH_GPOS 0x00008 | |
62 | #define ROPT_SKIP 0x00010 | |
63 | #define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */ | |
64 | #define ROPT_NOSCAN 0x00040 /* Check-string always at start. */ | |
65 | #define ROPT_GPOS_SEEN 0x00080 | |
66 | #define ROPT_CHECK_ALL 0x00100 | |
67 | #define ROPT_LOOKBEHIND_SEEN 0x00200 | |
68 | #define ROPT_EVAL_SEEN 0x00400 | |
cce850e4 JH |
69 | #define ROPT_CANY_SEEN 0x00800 |
70 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ | |
a0ed51b3 | 71 | |
8782bef2 | 72 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
c277df42 | 73 | |
a0ed51b3 LW |
74 | #define ROPT_UTF8 0x10000 |
75 | #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ | |
cf93c79d | 76 | #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ |
bc9a0d2c | 77 | #define ROPT_TAINTED_SEEN 0x80000 |
a30b2f1f | 78 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
a0ed51b3 | 79 | |
cad2e5aa JH |
80 | #define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */ |
81 | #define RE_USE_INTUIT_ML 0x0200000 | |
82 | #define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */ | |
83 | #define REINT_AUTORITATIVE_ML 0x0800000 | |
84 | #define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */ | |
85 | #define REINT_ONCE_ML 0x2000000 | |
86 | #define RE_INTUIT_ONECHAR 0x4000000 | |
87 | #define RE_INTUIT_TAIL 0x8000000 | |
88 | ||
89 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) | |
90 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) | |
91 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) | |
92 | ||
c277df42 | 93 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
72311751 GS |
94 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
95 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) | |
96 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ | |
97 | ? RX_MATCH_TAINTED_on(prog) \ | |
98 | : RX_MATCH_TAINTED_off(prog)) | |
c277df42 | 99 | |
cf93c79d IZ |
100 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
101 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) | |
102 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) | |
103 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ | |
104 | ? RX_MATCH_COPIED_on(prog) \ | |
105 | : RX_MATCH_COPIED_off(prog)) | |
106 | ||
f8c7b90f | 107 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
108 | #define RX_MATCH_COPY_FREE(rx) \ |
109 | STMT_START {if (rx->saved_copy) { \ | |
110 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ | |
111 | } \ | |
112 | if (RX_MATCH_COPIED(rx)) { \ | |
113 | Safefree(rx->subbeg); \ | |
114 | RX_MATCH_COPIED_off(rx); \ | |
115 | }} STMT_END | |
116 | #else | |
117 | #define RX_MATCH_COPY_FREE(rx) \ | |
118 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ | |
119 | Safefree(rx->subbeg); \ | |
120 | RX_MATCH_COPIED_off(rx); \ | |
121 | }} STMT_END | |
122 | #endif | |
123 | ||
a30b2f1f RGS |
124 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
125 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) | |
126 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) | |
127 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ | |
128 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ | |
129 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) | |
130 | ||
cad2e5aa JH |
131 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
132 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ | |
133 | #define REXEC_SCREAM 0x04 /* use scream table. */ | |
134 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ | |
cf93c79d | 135 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
c277df42 | 136 | |
155aba94 | 137 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
cad2e5aa | 138 | #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) |
cf93c79d IZ |
139 | |
140 | #define FBMcf_TAIL_DOLLAR 1 | |
cad2e5aa JH |
141 | #define FBMcf_TAIL_DOLLARM 2 |
142 | #define FBMcf_TAIL_Z 4 | |
143 | #define FBMcf_TAIL_z 8 | |
144 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) | |
cf93c79d IZ |
145 | |
146 | #define FBMrf_MULTILINE 1 | |
cad2e5aa JH |
147 | |
148 | struct re_scream_pos_data_s; | |
5d9a96ca DM |
149 | |
150 | /* an accepting state/position*/ | |
151 | struct _reg_trie_accepted { | |
152 | U8 *endpos; | |
153 | U16 wordnum; | |
154 | }; | |
155 | typedef struct _reg_trie_accepted reg_trie_accepted; | |
156 | ||
3b0527fe DM |
157 | /* some basic information about the current match that is created by |
158 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ | |
159 | ||
160 | typedef struct { | |
161 | regexp *prog; | |
162 | char *bol; | |
163 | char *till; | |
164 | SV *sv; | |
165 | char *ganch; | |
166 | } regmatch_info; | |
167 | ||
5d9a96ca DM |
168 | |
169 | /* structures for holding and saving the state maintained by regmatch() */ | |
170 | ||
171 | typedef I32 CHECKPOINT; | |
172 | ||
5d9a96ca DM |
173 | typedef enum { |
174 | resume_TRIE1, | |
175 | resume_TRIE2, | |
aa283a38 | 176 | resume_EVAL, |
5d9a96ca DM |
177 | resume_CURLYX, |
178 | resume_WHILEM1, | |
179 | resume_WHILEM2, | |
180 | resume_WHILEM3, | |
181 | resume_WHILEM4, | |
182 | resume_WHILEM5, | |
183 | resume_WHILEM6, | |
184 | resume_CURLYM1, | |
185 | resume_CURLYM2, | |
186 | resume_CURLYM3, | |
5d9a96ca DM |
187 | resume_IFMATCH, |
188 | resume_PLUS1, | |
189 | resume_PLUS2, | |
190 | resume_PLUS3, | |
aa283a38 | 191 | resume_PLUS4 |
5d9a96ca DM |
192 | } regmatch_resume_states; |
193 | ||
194 | ||
a0374537 | 195 | typedef struct regmatch_state { |
5d9a96ca DM |
196 | |
197 | /* these vars contain state that needs to be maintained | |
198 | * across the main while loop ... */ | |
199 | ||
200 | regmatch_resume_states resume_state; /* where to jump to on return */ | |
201 | regnode *scan; /* Current node. */ | |
202 | regnode *next; /* Next node. */ | |
aa283a38 | 203 | bool minmod; /* the next "{n,m}" is a "{n,m}?" */ |
5d9a96ca DM |
204 | bool sw; /* the condition value in (?(cond)a|b) */ |
205 | int logical; | |
206 | I32 unwind; /* savestack index of current unwind block */ | |
a0374537 | 207 | struct regmatch_state *cc; /* current innermost curly state */ |
5d9a96ca DM |
208 | char *locinput; |
209 | ||
e822a8b4 | 210 | /* ... while the rest of these are local to an individual branch */ |
5d9a96ca DM |
211 | |
212 | I32 n; /* no or next */ | |
213 | I32 ln; /* len or last */ | |
e822a8b4 DM |
214 | |
215 | union { | |
77cb431f DM |
216 | |
217 | /* this is a fake union member that matches the first element | |
218 | * of each member that needs to store positive backtrack | |
219 | * information */ | |
220 | struct { | |
221 | struct regmatch_state *prev_yes_state; | |
222 | } yes; | |
223 | ||
e822a8b4 DM |
224 | struct { |
225 | reg_trie_accepted *accept_buff; | |
226 | U32 accepted; /* how many accepting states we have seen */ | |
227 | } trie; | |
228 | ||
229 | struct { | |
77cb431f DM |
230 | /* this first element must match u.yes */ |
231 | struct regmatch_state *prev_yes_state; | |
aa283a38 DM |
232 | regexp *prev_rex; |
233 | int toggleutf; | |
234 | CHECKPOINT cp; /* remember current savestack indexes */ | |
235 | CHECKPOINT lastcp; | |
e822a8b4 DM |
236 | } eval; |
237 | ||
238 | struct { | |
239 | CHECKPOINT cp; /* remember current savestack indexes */ | |
a0374537 DM |
240 | struct regmatch_state *outercc; /* outer CURLYX state if any */ |
241 | ||
242 | /* these contain the current curly state, and are accessed | |
243 | * by subsequent WHILEMs */ | |
244 | int parenfloor;/* how far back to strip paren data */ | |
245 | int cur; /* how many instances of scan we've matched */ | |
246 | int min; /* the minimal number of scans to match */ | |
247 | int max; /* the maximal number of scans to match */ | |
248 | regnode * scan; /* the thing to match */ | |
249 | char * lastloc;/* where we started matching this scan */ | |
e822a8b4 DM |
250 | } curlyx; |
251 | ||
252 | struct { | |
253 | CHECKPOINT cp; /* remember current savestack indexes */ | |
254 | CHECKPOINT lastcp; | |
a0374537 | 255 | struct regmatch_state *savecc; |
e822a8b4 DM |
256 | char *lastloc; /* Detection of 0-len. */ |
257 | I32 cache_offset; | |
258 | I32 cache_bit; | |
259 | } whilem; | |
260 | ||
261 | struct { | |
dad79028 DM |
262 | /* this first element must match u.yes */ |
263 | struct regmatch_state *prev_yes_state; | |
e822a8b4 DM |
264 | I32 paren; |
265 | I32 c1, c2; /* case fold search */ | |
266 | CHECKPOINT lastcp; | |
267 | I32 l; | |
268 | I32 matches; | |
269 | I32 maxwanted; | |
0cadcf80 | 270 | bool minmod; |
e822a8b4 DM |
271 | } curlym; |
272 | ||
273 | struct { | |
274 | I32 paren; | |
275 | CHECKPOINT lastcp; | |
276 | I32 c1, c2; /* case fold search */ | |
277 | char *e; | |
278 | char *old; | |
279 | int count; | |
280 | } plus; /* and CURLYN/CURLY/STAR */ | |
dad79028 DM |
281 | |
282 | struct { | |
283 | /* this first element must match u.yes */ | |
284 | struct regmatch_state *prev_yes_state; | |
285 | I32 wanted; | |
286 | } ifmatch; /* and SUSPEND/UNLESSM */ | |
d8319b27 | 287 | } u; |
5d9a96ca DM |
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 NC |
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 NC |
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 NC |
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 NC |
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 NC |
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 NC |
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 NC |
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 NC |
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 NC |
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 | */ |