3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
4 * by Larry Wall and others
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.
12 * Definitions etc. for regexp(3) routines.
14 * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
15 * not the System V one.
25 typedef struct regnode regnode;
27 struct reg_substr_data;
31 typedef struct regexp {
35 struct reg_substr_data *substrs;
36 char *precomp; /* pre-compilation regular expression */
37 struct reg_data *data; /* Additional data. */
38 char *subbeg; /* saved or original string
39 so \digit works forever. */
40 #ifdef PERL_OLD_COPY_ON_WRITE
41 SV *saved_copy; /* If non-NULL, SV which is COW from original */
43 U32 *offsets; /* offset annotations 20001228 MJD */
44 I32 sublen; /* Length of string pointed by subbeg */
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 */
50 U32 lastcloseparen; /* last paren matched */
51 U32 reganch; /* Internal use only +
52 Tainted information used by regexec? */
53 regnode program[1]; /* Unwarranted chumminess with compiler. */
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)
58 #define ROPT_ANCH_BOL 0x00001
59 #define ROPT_ANCH_MBOL 0x00002
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
69 #define ROPT_CANY_SEEN 0x00800
70 #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
72 /* 0xf800 of reganch is used by PMf_COMPILETIME */
74 #define ROPT_UTF8 0x10000
75 #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */
76 #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */
77 #define ROPT_TAINTED_SEEN 0x80000
78 #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
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
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)
93 #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
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))
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))
107 #ifdef PERL_OLD_COPY_ON_WRITE
108 #define RX_MATCH_COPY_FREE(rx) \
109 STMT_START {if (rx->saved_copy) { \
110 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
112 if (RX_MATCH_COPIED(rx)) { \
113 Safefree(rx->subbeg); \
114 RX_MATCH_COPIED_off(rx); \
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); \
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)))
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. */
135 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
137 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
138 #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re)
140 #define FBMcf_TAIL_DOLLAR 1
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)
146 #define FBMrf_MULTILINE 1
148 struct re_scream_pos_data_s;
150 /* an accepting state/position*/
151 struct _reg_trie_accepted {
155 typedef struct _reg_trie_accepted reg_trie_accepted;
158 /* structures for holding and saving the state maintained by regmatch() */
160 typedef I32 CHECKPOINT;
182 } regmatch_resume_states;
185 typedef struct regmatch_state {
187 /* these vars contain state that needs to be maintained
188 * across the main while loop ... */
190 regmatch_resume_states resume_state; /* where to jump to on return */
191 regnode *scan; /* Current node. */
192 regnode *next; /* Next node. */
193 bool minmod; /* the next "{n,m}" is a "{n,m}?" */
194 bool sw; /* the condition value in (?(cond)a|b) */
196 I32 unwind; /* savestack index of current unwind block */
197 struct regmatch_state *cc; /* current innermost curly state */
200 /* ... while the rest of these are local to an individual branch */
202 I32 n; /* no or next */
203 I32 ln; /* len or last */
207 reg_trie_accepted *accept_buff;
208 U32 accepted; /* how many accepting states we have seen */
214 CHECKPOINT cp; /* remember current savestack indexes */
216 struct regmatch_state *prev_eval; /* save cur_eval */
217 struct regmatch_slab *prev_slab;
223 CHECKPOINT cp; /* remember current savestack indexes */
224 struct regmatch_state *outercc; /* outer CURLYX state if any */
226 /* these contain the current curly state, and are accessed
227 * by subsequent WHILEMs */
228 int parenfloor;/* how far back to strip paren data */
229 int cur; /* how many instances of scan we've matched */
230 int min; /* the minimal number of scans to match */
231 int max; /* the maximal number of scans to match */
232 regnode * scan; /* the thing to match */
233 char * lastloc;/* where we started matching this scan */
237 CHECKPOINT cp; /* remember current savestack indexes */
239 struct regmatch_state *savecc;
240 char *lastloc; /* Detection of 0-len. */
247 I32 c1, c2; /* case fold search */
257 I32 c1, c2; /* case fold search */
261 } plus; /* and CURLYN/CURLY/STAR */
265 /* how many regmatch_state structs to allocate as a single slab.
266 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
267 * pointers, plus 1 for any mythical malloc overhead. */
269 #define PERL_REGMATCH_SLAB_SLOTS \
270 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
272 typedef struct regmatch_slab {
273 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
274 struct regmatch_slab *prev, *next;
277 #define PL_reg_flags PL_reg_state.re_state_reg_flags
278 #define PL_bostr PL_reg_state.re_state_bostr
279 #define PL_reginput PL_reg_state.re_state_reginput
280 #define PL_regbol PL_reg_state.re_state_regbol
281 #define PL_regeol PL_reg_state.re_state_regeol
282 #define PL_regstartp PL_reg_state.re_state_regstartp
283 #define PL_regendp PL_reg_state.re_state_regendp
284 #define PL_reglastparen PL_reg_state.re_state_reglastparen
285 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
286 #define PL_regtill PL_reg_state.re_state_regtill
287 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
288 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
289 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
290 #define PL_regnarrate PL_reg_state.re_state_regnarrate
291 #define PL_regindent PL_reg_state.re_state_regindent
292 #define PL_reg_re PL_reg_state.re_state_reg_re
293 #define PL_reg_ganch PL_reg_state.re_state_reg_ganch
294 #define PL_reg_sv PL_reg_state.re_state_reg_sv
295 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
296 #define PL_reg_magic PL_reg_state.re_state_reg_magic
297 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
298 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
299 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
300 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
301 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
302 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
303 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
304 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
305 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
306 #define PL_regsize PL_reg_state.re_state_regsize
307 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
308 #define PL_nrs PL_reg_state.re_state_nrs
310 struct re_save_state {
311 U32 re_state_reg_flags; /* from regexec.c */
312 char *re_state_bostr;
313 char *re_state_reginput; /* String-input pointer. */
314 char *re_state_regbol; /* Beginning of input, for ^ check. */
315 char *re_state_regeol; /* End of input, for $ check. */
316 I32 *re_state_regstartp; /* Pointer to startp array. */
317 I32 *re_state_regendp; /* Ditto for endp. */
318 U32 *re_state_reglastparen; /* Similarly for lastparen. */
319 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
320 char *re_state_regtill; /* How far we are required to go. */
321 char **re_state_reg_start_tmp; /* from regexec.c */
322 U32 re_state_reg_start_tmpl; /* from regexec.c */
323 I32 re_state_reg_eval_set; /* from regexec.c */
324 I32 re_state_regnarrate; /* from regexec.c */
325 int re_state_regindent; /* from regexec.c */
326 regexp *re_state_reg_re; /* from regexec.c */
327 char *re_state_reg_ganch; /* from regexec.c */
328 SV *re_state_reg_sv; /* from regexec.c */
329 bool re_state_reg_match_utf8; /* from regexec.c */
330 MAGIC *re_state_reg_magic; /* from regexec.c */
331 I32 re_state_reg_oldpos; /* from regexec.c */
332 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
333 PMOP *re_state_reg_curpm; /* from regexec.c */
334 char *re_state_reg_oldsaved; /* old saved substr during match */
335 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
336 I32 re_state_reg_maxiter; /* max wait until caching pos */
337 I32 re_state_reg_leftiter; /* wait until caching pos */
338 char *re_state_reg_poscache; /* cache of pos of WHILEM */
339 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
340 I32 re_state_regsize; /* from regexec.c */
341 char *re_state_reg_starttry; /* from regexec.c */
342 #ifdef PERL_OLD_COPY_ON_WRITE
343 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 */
347 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
348 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
351 * c-indentation-style: bsd
353 * indent-tabs-mode: t
356 * ex: set ts=8 sts=4 sw=4 noet: