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.
17 #ifndef PLUGGABLE_RE_EXTENSION
18 /* we don't want to include this stuff if we are inside Nicholas'
19 * pluggable regex engine code */
27 typedef struct regnode regnode;
29 struct reg_substr_data;
35 typedef struct regexp {
39 struct reg_substr_data *substrs;
40 char *precomp; /* pre-compilation regular expression */
41 struct reg_data *data; /* Additional data. */
42 char *subbeg; /* saved or original string
43 so \digit works forever. */
44 #ifdef PERL_OLD_COPY_ON_WRITE
45 SV *saved_copy; /* If non-NULL, SV which is COW from original */
47 U32 *offsets; /* offset annotations 20001228 MJD */
48 I32 sublen; /* Length of string pointed by subbeg */
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 */
54 U32 lastcloseparen; /* last paren matched */
55 U32 reganch; /* Internal use only +
56 Tainted information used by regexec? */
57 const struct regexp_engine* engine;
58 regnode program[1]; /* Unwarranted chumminess with compiler. */
62 typedef struct re_scream_pos_data_s
64 char **scream_olds; /* match pos */
65 I32 *scream_pos; /* Internal iterator of scream. */
68 typedef struct regexp_engine {
69 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
70 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
71 char* strbeg, I32 minend, SV* screamer,
72 void* data, U32 flags);
73 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
74 char *strend, U32 flags,
75 struct re_scream_pos_data_s *data);
76 SV* (*checkstr) (pTHX_ regexp *prog);
77 void (*free) (pTHX_ struct regexp* r);
79 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
83 #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
84 #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
85 #define ROPT_ANCH_BOL 0x00000001
86 #define ROPT_ANCH_MBOL 0x00000002
87 #define ROPT_ANCH_SBOL 0x00000004
88 #define ROPT_ANCH_GPOS 0x00000008
89 #define ROPT_SKIP 0x00000010
90 #define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */
91 #define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */
92 #define ROPT_GPOS_SEEN 0x00000080
93 #define ROPT_CHECK_ALL 0x00000100
94 #define ROPT_LOOKBEHIND_SEEN 0x00000200
95 #define ROPT_EVAL_SEEN 0x00000400
96 #define ROPT_CANY_SEEN 0x00000800
97 #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
98 #define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS)
100 /* 0xf800 of reganch is used by PMf_COMPILETIME */
102 #define ROPT_UTF8 0x00010000
103 #define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */
104 #define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */
105 #define ROPT_TAINTED_SEEN 0x00080000
106 #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
108 #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
109 #define RE_USE_INTUIT_ML 0x00200000
110 #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
111 #define REINT_AUTORITATIVE_ML 0x00800000
112 #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
113 #define REINT_ONCE_ML 0x02000000
114 #define RE_INTUIT_ONECHAR 0x04000000
115 #define RE_INTUIT_TAIL 0x08000000
118 #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
119 #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
120 #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
122 #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
123 #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
124 #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
125 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
126 ? RX_MATCH_TAINTED_on(prog) \
127 : RX_MATCH_TAINTED_off(prog))
129 #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
130 #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
131 #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
132 #define RX_MATCH_COPIED_set(prog,t) ((t) \
133 ? RX_MATCH_COPIED_on(prog) \
134 : RX_MATCH_COPIED_off(prog))
136 #endif /* PLUGGABLE_RE_EXTENSION */
138 /* Stuff that needs to be included in the plugable extension goes below here */
140 #define RE_DEBUG_BIT 0x20000000
141 #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT)
142 #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT)
144 #ifdef PERL_OLD_COPY_ON_WRITE
145 #define RX_MATCH_COPY_FREE(rx) \
146 STMT_START {if (rx->saved_copy) { \
147 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
149 if (RX_MATCH_COPIED(rx)) { \
150 Safefree(rx->subbeg); \
151 RX_MATCH_COPIED_off(rx); \
154 #define RX_MATCH_COPY_FREE(rx) \
155 STMT_START {if (RX_MATCH_COPIED(rx)) { \
156 Safefree(rx->subbeg); \
157 RX_MATCH_COPIED_off(rx); \
161 #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
162 #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
163 #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
164 #define RX_MATCH_UTF8_set(prog, t) ((t) \
165 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
166 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
168 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
169 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
170 #define REXEC_SCREAM 0x04 /* use scream table. */
171 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
172 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
174 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
175 #define ReREFCNT_dec(re) CALLREGFREE(re)
177 #define FBMcf_TAIL_DOLLAR 1
178 #define FBMcf_TAIL_DOLLARM 2
179 #define FBMcf_TAIL_Z 4
180 #define FBMcf_TAIL_z 8
181 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
183 #define FBMrf_MULTILINE 1
185 struct re_scream_pos_data_s;
187 /* an accepting state/position*/
188 struct _reg_trie_accepted {
192 typedef struct _reg_trie_accepted reg_trie_accepted;
194 /* some basic information about the current match that is created by
195 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
206 /* structures for holding and saving the state maintained by regmatch() */
208 typedef I32 CHECKPOINT;
210 typedef struct regmatch_state {
212 /* these vars contain state that needs to be maintained
213 * across the main while loop ... */
215 int resume_state; /* where to jump to on return */
216 regnode *scan; /* Current node. */
217 regnode *next; /* Next node. */
218 bool minmod; /* the next "{n,m}" is a "{n,m}?" */
219 bool sw; /* the condition value in (?(cond)a|b) */
221 I32 unwind; /* savestack index of current unwind block */
224 /* ... while the rest of these are local to an individual branch */
226 I32 n; /* no or next */
227 I32 ln; /* len or last */
231 /* this is a fake union member that matches the first element
232 * of each member that needs to store positive backtrack
235 struct regmatch_state *prev_yes_state;
239 reg_trie_accepted *accept_buff;
240 U32 accepted; /* how many accepting states we have seen */
241 U16 *jump; /* negative offsets from B */
242 regnode *B; /* node following the trie */
243 regnode *me; /* only needed for debugging */
247 /* this first element must match u.yes */
248 struct regmatch_state *prev_yes_state;
249 struct regmatch_state *prev_eval;
250 struct regmatch_state *prev_curlyx;
252 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
253 flip when transitioning between
254 inner and outer rexen */
255 CHECKPOINT cp; /* remember current savestack indexes */
257 regnode *B; /* the node following us */
261 CHECKPOINT cp; /* remember current savestack indexes */
262 struct regmatch_state *outercc; /* outer CURLYX state if any */
264 /* these contain the current curly state, and are accessed
265 * by subsequent WHILEMs */
266 int parenfloor;/* how far back to strip paren data */
267 int cur; /* how many instances of scan we've matched */
268 int min; /* the minimal number of scans to match */
269 int max; /* the maximal number of scans to match */
270 regnode * scan; /* the thing to match */
271 char * lastloc;/* where we started matching this scan */
275 CHECKPOINT cp; /* remember current savestack indexes */
277 struct regmatch_state *savecc;
278 char *lastloc; /* Detection of 0-len. */
285 regnode *next_branch; /* next branch node */
290 /* this first element must match u.yes */
291 struct regmatch_state *prev_yes_state;
292 I32 c1, c2; /* case fold search */
294 I32 alen; /* length of first-matched A string */
297 regnode *A, *B; /* the nodes corresponding to /A*B/ */
298 regnode *me; /* the curlym node */
304 I32 c1, c2; /* case fold search */
305 char *maxpos; /* highest possible point in string to match */
306 char *oldloc; /* the previous locinput */
308 int min, max; /* {m,n} */
309 regnode *A, *B; /* the nodes corresponding to /A*B/ */
310 } curly; /* and CURLYN/PLUS/STAR */
313 /* this first element must match u.yes */
314 struct regmatch_state *prev_yes_state;
316 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
317 } ifmatch; /* and SUSPEND/UNLESSM */
321 /* how many regmatch_state structs to allocate as a single slab.
322 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
323 * pointers, plus 1 for any mythical malloc overhead. */
325 #define PERL_REGMATCH_SLAB_SLOTS \
326 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
328 typedef struct regmatch_slab {
329 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
330 struct regmatch_slab *prev, *next;
333 #define PL_reg_flags PL_reg_state.re_state_reg_flags
334 #define PL_bostr PL_reg_state.re_state_bostr
335 #define PL_reginput PL_reg_state.re_state_reginput
336 #define PL_regeol PL_reg_state.re_state_regeol
337 #define PL_regstartp PL_reg_state.re_state_regstartp
338 #define PL_regendp PL_reg_state.re_state_regendp
339 #define PL_reglastparen PL_reg_state.re_state_reglastparen
340 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
341 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
342 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
343 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
344 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
345 #define PL_reg_magic PL_reg_state.re_state_reg_magic
346 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
347 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
348 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
349 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
350 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
351 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
352 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
353 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
354 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
355 #define PL_regsize PL_reg_state.re_state_regsize
356 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
357 #define PL_nrs PL_reg_state.re_state_nrs
359 struct re_save_state {
360 U32 re_state_reg_flags; /* from regexec.c */
361 char *re_state_bostr;
362 char *re_state_reginput; /* String-input pointer. */
363 char *re_state_regeol; /* End of input, for $ check. */
364 I32 *re_state_regstartp; /* Pointer to startp array. */
365 I32 *re_state_regendp; /* Ditto for endp. */
366 U32 *re_state_reglastparen; /* Similarly for lastparen. */
367 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
368 char **re_state_reg_start_tmp; /* from regexec.c */
369 U32 re_state_reg_start_tmpl; /* from regexec.c */
370 I32 re_state_reg_eval_set; /* from regexec.c */
371 bool re_state_reg_match_utf8; /* from regexec.c */
372 MAGIC *re_state_reg_magic; /* from regexec.c */
373 I32 re_state_reg_oldpos; /* from regexec.c */
374 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
375 PMOP *re_state_reg_curpm; /* from regexec.c */
376 char *re_state_reg_oldsaved; /* old saved substr during match */
377 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
378 I32 re_state_reg_maxiter; /* max wait until caching pos */
379 I32 re_state_reg_leftiter; /* wait until caching pos */
380 char *re_state_reg_poscache; /* cache of pos of WHILEM */
381 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
382 I32 re_state_regsize; /* from regexec.c */
383 char *re_state_reg_starttry; /* from regexec.c */
384 #ifdef PERL_OLD_COPY_ON_WRITE
385 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 */
389 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
390 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
393 * c-indentation-style: bsd
395 * indent-tabs-mode: t
398 * ex: set ts=8 sts=4 sw=4 noet: