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 string to match */
51 I32 minlenret; /* mininum possible length of $& */
52 I32 prelen; /* length of precomp */
53 U32 nparens; /* number of parentheses */
54 U32 lastparen; /* last paren matched */
55 U32 lastcloseparen; /* last paren matched */
56 U32 reganch; /* Internal use only +
57 Tainted information used by regexec? */
58 HV *paren_names; /* Paren names */
59 const struct regexp_engine* engine;
60 regnode program[1]; /* Unwarranted chumminess with compiler. */
64 typedef struct re_scream_pos_data_s
66 char **scream_olds; /* match pos */
67 I32 *scream_pos; /* Internal iterator of scream. */
70 typedef struct regexp_engine {
71 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
72 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
73 char* strbeg, I32 minend, SV* screamer,
74 void* data, U32 flags);
75 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
76 char *strend, U32 flags,
77 struct re_scream_pos_data_s *data);
78 SV* (*checkstr) (pTHX_ regexp *prog);
79 void (*free) (pTHX_ struct regexp* r);
80 char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval);
82 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
86 #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
87 #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
88 #define ROPT_ANCH_BOL 0x00000001
89 #define ROPT_ANCH_MBOL 0x00000002
90 #define ROPT_ANCH_SBOL 0x00000004
91 #define ROPT_ANCH_GPOS 0x00000008
92 #define ROPT_SKIP 0x00000010
93 #define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */
94 #define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */
95 #define ROPT_GPOS_SEEN 0x00000080
96 #define ROPT_CHECK_ALL 0x00000100
97 #define ROPT_LOOKBEHIND_SEEN 0x00000200
98 #define ROPT_EVAL_SEEN 0x00000400
99 #define ROPT_CANY_SEEN 0x00000800
100 #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
101 #define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS)
103 /* 0xF800 of reganch is used by PMf_COMPILETIME */
105 #define ROPT_UTF8 0x00010000
106 #define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */
107 #define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */
108 #define ROPT_TAINTED_SEEN 0x00080000
109 #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
110 #define ROPT_VERBARG_SEEN 0x20000000
111 #define ROPT_CUTGROUP_SEEN 0x40000000
113 #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
114 #define RE_USE_INTUIT_ML 0x00200000
115 #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
116 #define REINT_AUTORITATIVE_ML 0x00800000
117 #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
118 #define REINT_ONCE_ML 0x02000000
119 #define RE_INTUIT_ONECHAR 0x04000000
120 #define RE_INTUIT_TAIL 0x08000000
123 #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
124 #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
125 #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
127 #define RX_HAS_CUTGROUP(prog) ((prog)->reganch & ROPT_CUTGROUP_SEEN)
128 #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
129 #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
130 #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
131 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
132 ? RX_MATCH_TAINTED_on(prog) \
133 : RX_MATCH_TAINTED_off(prog))
135 #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
136 #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
137 #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
138 #define RX_MATCH_COPIED_set(prog,t) ((t) \
139 ? RX_MATCH_COPIED_on(prog) \
140 : RX_MATCH_COPIED_off(prog))
142 #endif /* PLUGGABLE_RE_EXTENSION */
144 /* Stuff that needs to be included in the plugable extension goes below here */
146 #define RE_DEBUG_BIT 0x20000000
147 #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT)
148 #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT)
150 #ifdef PERL_OLD_COPY_ON_WRITE
151 #define RX_MATCH_COPY_FREE(rx) \
152 STMT_START {if (rx->saved_copy) { \
153 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
155 if (RX_MATCH_COPIED(rx)) { \
156 Safefree(rx->subbeg); \
157 RX_MATCH_COPIED_off(rx); \
160 #define RX_MATCH_COPY_FREE(rx) \
161 STMT_START {if (RX_MATCH_COPIED(rx)) { \
162 Safefree(rx->subbeg); \
163 RX_MATCH_COPIED_off(rx); \
167 #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
168 #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
169 #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
170 #define RX_MATCH_UTF8_set(prog, t) ((t) \
171 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
172 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
174 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
175 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
176 #define REXEC_SCREAM 0x04 /* use scream table. */
177 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
178 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
180 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
181 #define ReREFCNT_dec(re) CALLREGFREE(re)
183 #define FBMcf_TAIL_DOLLAR 1
184 #define FBMcf_TAIL_DOLLARM 2
185 #define FBMcf_TAIL_Z 4
186 #define FBMcf_TAIL_z 8
187 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
189 #define FBMrf_MULTILINE 1
191 struct re_scream_pos_data_s;
193 /* an accepting state/position*/
194 struct _reg_trie_accepted {
198 typedef struct _reg_trie_accepted reg_trie_accepted;
200 /* some basic information about the current match that is created by
201 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
213 /* structures for holding and saving the state maintained by regmatch() */
215 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
217 typedef I32 CHECKPOINT;
219 typedef struct regmatch_state {
220 int resume_state; /* where to jump to on return */
221 char *locinput; /* where to backtrack in string on failure */
225 /* this is a fake union member that matches the first element
226 * of each member that needs to store positive backtrack
229 struct regmatch_state *prev_yes_state;
233 /* this first element must match u.yes */
234 struct regmatch_state *prev_yes_state;
235 reg_trie_accepted *accept_buff;
236 U32 accepted; /* how many accepting states we have seen */
237 U16 *jump; /* positive offsets from me */
238 regnode *B; /* node following the trie */
239 regnode *me; /* Which node am I - needed for jump tries*/
243 /* this first element must match u.yes */
244 struct regmatch_state *prev_yes_state;
245 struct regmatch_state *prev_eval;
246 struct regmatch_state *prev_curlyx;
248 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
249 flip when transitioning between
250 inner and outer rexen */
251 CHECKPOINT cp; /* remember current savestack indexes */
253 regnode *B; /* the node following us */
254 U32 close_paren; /* which close bracket is our end */
258 /* this first element must match u.yes */
259 struct regmatch_state *prev_yes_state;
260 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
261 CHECKPOINT cp; /* remember current savestack index */
263 int parenfloor;/* how far back to strip paren data */
264 int min; /* the minimal number of A's to match */
265 int max; /* the maximal number of A's to match */
266 regnode *A, *B; /* the nodes corresponding to /A*B/ */
268 /* these two are modified by WHILEM */
269 int count; /* how many instances of A we've matched */
270 char *lastloc;/* where previous A matched (0-len detect) */
274 /* this first element must match u.yes */
275 struct regmatch_state *prev_yes_state;
276 struct regmatch_state *save_curlyx;
277 CHECKPOINT cp; /* remember current savestack indexes */
279 char *save_lastloc; /* previous curlyx.lastloc */
285 /* this first element must match u.yes */
286 struct regmatch_state *prev_yes_state;
288 regnode *next_branch; /* next branch node */
293 /* this first element must match u.yes */
294 struct regmatch_state *prev_yes_state;
295 I32 c1, c2; /* case fold search */
297 I32 alen; /* length of first-matched A string */
300 regnode *A, *B; /* the nodes corresponding to /A*B/ */
301 regnode *me; /* the curlym node */
307 I32 c1, c2; /* case fold search */
308 char *maxpos; /* highest possible point in string to match */
309 char *oldloc; /* the previous locinput */
311 int min, max; /* {m,n} */
312 regnode *A, *B; /* the nodes corresponding to /A*B/ */
313 } curly; /* and CURLYN/PLUS/STAR */
316 /* this first element must match u.yes */
317 struct regmatch_state *prev_yes_state;
319 I32 logical; /* saved copy of 'logical' var */
320 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
321 } ifmatch; /* and SUSPEND/UNLESSM */
324 /* this first element must match u.yes */
325 struct regmatch_state *prev_yes_state;
326 struct regmatch_state *prev_mark;
333 /* how many regmatch_state structs to allocate as a single slab.
334 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
335 * pointers, plus 1 for any mythical malloc overhead. */
337 #define PERL_REGMATCH_SLAB_SLOTS \
338 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
340 typedef struct regmatch_slab {
341 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
342 struct regmatch_slab *prev, *next;
345 #define PL_reg_flags PL_reg_state.re_state_reg_flags
346 #define PL_bostr PL_reg_state.re_state_bostr
347 #define PL_reginput PL_reg_state.re_state_reginput
348 #define PL_regeol PL_reg_state.re_state_regeol
349 #define PL_regstartp PL_reg_state.re_state_regstartp
350 #define PL_regendp PL_reg_state.re_state_regendp
351 #define PL_reglastparen PL_reg_state.re_state_reglastparen
352 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
353 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
354 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
355 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
356 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
357 #define PL_reg_magic PL_reg_state.re_state_reg_magic
358 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
359 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
360 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
361 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
362 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
363 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
364 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
365 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
366 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
367 #define PL_regsize PL_reg_state.re_state_regsize
368 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
369 #define PL_nrs PL_reg_state.re_state_nrs
371 struct re_save_state {
372 U32 re_state_reg_flags; /* from regexec.c */
373 char *re_state_bostr;
374 char *re_state_reginput; /* String-input pointer. */
375 char *re_state_regeol; /* End of input, for $ check. */
376 I32 *re_state_regstartp; /* Pointer to startp array. */
377 I32 *re_state_regendp; /* Ditto for endp. */
378 U32 *re_state_reglastparen; /* Similarly for lastparen. */
379 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
380 char **re_state_reg_start_tmp; /* from regexec.c */
381 U32 re_state_reg_start_tmpl; /* from regexec.c */
382 I32 re_state_reg_eval_set; /* from regexec.c */
383 bool re_state_reg_match_utf8; /* from regexec.c */
384 MAGIC *re_state_reg_magic; /* from regexec.c */
385 I32 re_state_reg_oldpos; /* from regexec.c */
386 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
387 PMOP *re_state_reg_curpm; /* from regexec.c */
388 char *re_state_reg_oldsaved; /* old saved substr during match */
389 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
390 I32 re_state_reg_maxiter; /* max wait until caching pos */
391 I32 re_state_reg_leftiter; /* wait until caching pos */
392 char *re_state_reg_poscache; /* cache of pos of WHILEM */
393 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
394 U32 re_state_regsize; /* from regexec.c */
395 char *re_state_reg_starttry; /* from regexec.c */
396 #ifdef PERL_OLD_COPY_ON_WRITE
397 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 */
401 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
402 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
405 * c-indentation-style: bsd
407 * indent-tabs-mode: t
410 * ex: set ts=8 sts=4 sw=4 noet: