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)
99 #define ROPT_RECURSE_SEEN 0x00001000
101 /* 0xf800 of reganch is used by PMf_COMPILETIME */
103 #define ROPT_UTF8 0x00010000
104 #define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */
105 #define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */
106 #define ROPT_TAINTED_SEEN 0x00080000
107 #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
109 #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
110 #define RE_USE_INTUIT_ML 0x00200000
111 #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
112 #define REINT_AUTORITATIVE_ML 0x00800000
113 #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
114 #define REINT_ONCE_ML 0x02000000
115 #define RE_INTUIT_ONECHAR 0x04000000
116 #define RE_INTUIT_TAIL 0x08000000
119 #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
120 #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
121 #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
123 #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
124 #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
125 #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
126 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
127 ? RX_MATCH_TAINTED_on(prog) \
128 : RX_MATCH_TAINTED_off(prog))
130 #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
131 #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
132 #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
133 #define RX_MATCH_COPIED_set(prog,t) ((t) \
134 ? RX_MATCH_COPIED_on(prog) \
135 : RX_MATCH_COPIED_off(prog))
137 #endif /* PLUGGABLE_RE_EXTENSION */
139 /* Stuff that needs to be included in the plugable extension goes below here */
141 #define RE_DEBUG_BIT 0x20000000
142 #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT)
143 #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT)
145 #ifdef PERL_OLD_COPY_ON_WRITE
146 #define RX_MATCH_COPY_FREE(rx) \
147 STMT_START {if (rx->saved_copy) { \
148 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
150 if (RX_MATCH_COPIED(rx)) { \
151 Safefree(rx->subbeg); \
152 RX_MATCH_COPIED_off(rx); \
155 #define RX_MATCH_COPY_FREE(rx) \
156 STMT_START {if (RX_MATCH_COPIED(rx)) { \
157 Safefree(rx->subbeg); \
158 RX_MATCH_COPIED_off(rx); \
162 #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
163 #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
164 #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
165 #define RX_MATCH_UTF8_set(prog, t) ((t) \
166 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
167 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
169 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
170 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
171 #define REXEC_SCREAM 0x04 /* use scream table. */
172 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
173 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
175 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
176 #define ReREFCNT_dec(re) CALLREGFREE(re)
178 #define FBMcf_TAIL_DOLLAR 1
179 #define FBMcf_TAIL_DOLLARM 2
180 #define FBMcf_TAIL_Z 4
181 #define FBMcf_TAIL_z 8
182 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
184 #define FBMrf_MULTILINE 1
186 struct re_scream_pos_data_s;
188 /* an accepting state/position*/
189 struct _reg_trie_accepted {
193 typedef struct _reg_trie_accepted reg_trie_accepted;
195 /* some basic information about the current match that is created by
196 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
207 /* structures for holding and saving the state maintained by regmatch() */
209 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
211 typedef I32 CHECKPOINT;
213 typedef struct regmatch_state {
214 int resume_state; /* where to jump to on return */
215 char *locinput; /* where to backtrack in string on failure */
219 /* this is a fake union member that matches the first element
220 * of each member that needs to store positive backtrack
223 struct regmatch_state *prev_yes_state;
227 reg_trie_accepted *accept_buff;
228 U32 accepted; /* how many accepting states we have seen */
229 U16 *jump; /* negative offsets from B */
230 regnode *B; /* node following the trie */
231 regnode *me; /* only needed for debugging */
235 /* this first element must match u.yes */
236 struct regmatch_state *prev_yes_state;
237 struct regmatch_state *prev_eval;
238 struct regmatch_state *prev_curlyx;
240 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
241 flip when transitioning between
242 inner and outer rexen */
243 CHECKPOINT cp; /* remember current savestack indexes */
245 regnode *B; /* the node following us */
246 U32 close_paren; /* which close bracket is our end */
250 /* this first element must match u.yes */
251 struct regmatch_state *prev_yes_state;
252 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
253 CHECKPOINT cp; /* remember current savestack index */
255 int parenfloor;/* how far back to strip paren data */
256 int min; /* the minimal number of A's to match */
257 int max; /* the maximal number of A's to match */
258 regnode *A, *B; /* the nodes corresponding to /A*B/ */
260 /* these two are modified by WHILEM */
261 int count; /* how many instances of A we've matched */
262 char *lastloc;/* where previous A matched (0-len detect) */
266 /* this first element must match u.yes */
267 struct regmatch_state *prev_yes_state;
268 struct regmatch_state *save_curlyx;
269 CHECKPOINT cp; /* remember current savestack indexes */
271 char *save_lastloc; /* previous curlyx.lastloc */
278 regnode *next_branch; /* next branch node */
283 /* this first element must match u.yes */
284 struct regmatch_state *prev_yes_state;
285 I32 c1, c2; /* case fold search */
287 I32 alen; /* length of first-matched A string */
290 regnode *A, *B; /* the nodes corresponding to /A*B/ */
291 regnode *me; /* the curlym node */
297 I32 c1, c2; /* case fold search */
298 char *maxpos; /* highest possible point in string to match */
299 char *oldloc; /* the previous locinput */
301 int min, max; /* {m,n} */
302 regnode *A, *B; /* the nodes corresponding to /A*B/ */
303 } curly; /* and CURLYN/PLUS/STAR */
306 /* this first element must match u.yes */
307 struct regmatch_state *prev_yes_state;
309 I32 logical; /* saved copy of 'logical' var */
310 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
311 } ifmatch; /* and SUSPEND/UNLESSM */
315 /* how many regmatch_state structs to allocate as a single slab.
316 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
317 * pointers, plus 1 for any mythical malloc overhead. */
319 #define PERL_REGMATCH_SLAB_SLOTS \
320 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
322 typedef struct regmatch_slab {
323 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
324 struct regmatch_slab *prev, *next;
327 #define PL_reg_flags PL_reg_state.re_state_reg_flags
328 #define PL_bostr PL_reg_state.re_state_bostr
329 #define PL_reginput PL_reg_state.re_state_reginput
330 #define PL_regeol PL_reg_state.re_state_regeol
331 #define PL_regstartp PL_reg_state.re_state_regstartp
332 #define PL_regendp PL_reg_state.re_state_regendp
333 #define PL_reglastparen PL_reg_state.re_state_reglastparen
334 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
335 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
336 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
337 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
338 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
339 #define PL_reg_magic PL_reg_state.re_state_reg_magic
340 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
341 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
342 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
343 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
344 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
345 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
346 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
347 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
348 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
349 #define PL_regsize PL_reg_state.re_state_regsize
350 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
351 #define PL_nrs PL_reg_state.re_state_nrs
353 struct re_save_state {
354 U32 re_state_reg_flags; /* from regexec.c */
355 char *re_state_bostr;
356 char *re_state_reginput; /* String-input pointer. */
357 char *re_state_regeol; /* End of input, for $ check. */
358 I32 *re_state_regstartp; /* Pointer to startp array. */
359 I32 *re_state_regendp; /* Ditto for endp. */
360 U32 *re_state_reglastparen; /* Similarly for lastparen. */
361 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
362 char **re_state_reg_start_tmp; /* from regexec.c */
363 U32 re_state_reg_start_tmpl; /* from regexec.c */
364 I32 re_state_reg_eval_set; /* from regexec.c */
365 bool re_state_reg_match_utf8; /* from regexec.c */
366 MAGIC *re_state_reg_magic; /* from regexec.c */
367 I32 re_state_reg_oldpos; /* from regexec.c */
368 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
369 PMOP *re_state_reg_curpm; /* from regexec.c */
370 char *re_state_reg_oldsaved; /* old saved substr during match */
371 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
372 I32 re_state_reg_maxiter; /* max wait until caching pos */
373 I32 re_state_reg_leftiter; /* wait until caching pos */
374 char *re_state_reg_poscache; /* cache of pos of WHILEM */
375 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
376 I32 re_state_regsize; /* from regexec.c */
377 char *re_state_reg_starttry; /* from regexec.c */
378 #ifdef PERL_OLD_COPY_ON_WRITE
379 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 */
383 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
384 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
387 * c-indentation-style: bsd
389 * indent-tabs-mode: t
392 * ex: set ts=8 sts=4 sw=4 noet: