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 HV *paren_names; /* Paren names */
58 const struct regexp_engine* engine;
59 regnode program[1]; /* Unwarranted chumminess with compiler. */
63 typedef struct re_scream_pos_data_s
65 char **scream_olds; /* match pos */
66 I32 *scream_pos; /* Internal iterator of scream. */
69 typedef struct regexp_engine {
70 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
71 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
72 char* strbeg, I32 minend, SV* screamer,
73 void* data, U32 flags);
74 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
75 char *strend, U32 flags,
76 struct re_scream_pos_data_s *data);
77 SV* (*checkstr) (pTHX_ regexp *prog);
78 void (*free) (pTHX_ struct regexp* r);
80 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
84 #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
85 #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
86 #define ROPT_ANCH_BOL 0x00000001
87 #define ROPT_ANCH_MBOL 0x00000002
88 #define ROPT_ANCH_SBOL 0x00000004
89 #define ROPT_ANCH_GPOS 0x00000008
90 #define ROPT_SKIP 0x00000010
91 #define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */
92 #define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */
93 #define ROPT_GPOS_SEEN 0x00000080
94 #define ROPT_CHECK_ALL 0x00000100
95 #define ROPT_LOOKBEHIND_SEEN 0x00000200
96 #define ROPT_EVAL_SEEN 0x00000400
97 #define ROPT_CANY_SEEN 0x00000800
98 #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
99 #define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS)
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 */
108 #define ROPT_RECURSE_SEEN 0x20000000
109 #define ROPT_VERBARG_SEEN 0x40000000
111 #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */
112 #define RE_USE_INTUIT_ML 0x00200000
113 #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */
114 #define REINT_AUTORITATIVE_ML 0x00800000
115 #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */
116 #define REINT_ONCE_ML 0x02000000
117 #define RE_INTUIT_ONECHAR 0x04000000
118 #define RE_INTUIT_TAIL 0x08000000
121 #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
122 #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
123 #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
125 #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
126 #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
127 #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
128 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
129 ? RX_MATCH_TAINTED_on(prog) \
130 : RX_MATCH_TAINTED_off(prog))
132 #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
133 #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
134 #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
135 #define RX_MATCH_COPIED_set(prog,t) ((t) \
136 ? RX_MATCH_COPIED_on(prog) \
137 : RX_MATCH_COPIED_off(prog))
139 #endif /* PLUGGABLE_RE_EXTENSION */
141 /* Stuff that needs to be included in the plugable extension goes below here */
143 #define RE_DEBUG_BIT 0x20000000
144 #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT)
145 #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT)
147 #ifdef PERL_OLD_COPY_ON_WRITE
148 #define RX_MATCH_COPY_FREE(rx) \
149 STMT_START {if (rx->saved_copy) { \
150 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
152 if (RX_MATCH_COPIED(rx)) { \
153 Safefree(rx->subbeg); \
154 RX_MATCH_COPIED_off(rx); \
157 #define RX_MATCH_COPY_FREE(rx) \
158 STMT_START {if (RX_MATCH_COPIED(rx)) { \
159 Safefree(rx->subbeg); \
160 RX_MATCH_COPIED_off(rx); \
164 #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
165 #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
166 #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
167 #define RX_MATCH_UTF8_set(prog, t) ((t) \
168 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
169 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
171 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
172 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
173 #define REXEC_SCREAM 0x04 /* use scream table. */
174 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
175 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
177 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
178 #define ReREFCNT_dec(re) CALLREGFREE(re)
180 #define FBMcf_TAIL_DOLLAR 1
181 #define FBMcf_TAIL_DOLLARM 2
182 #define FBMcf_TAIL_Z 4
183 #define FBMcf_TAIL_z 8
184 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
186 #define FBMrf_MULTILINE 1
188 struct re_scream_pos_data_s;
190 /* an accepting state/position*/
191 struct _reg_trie_accepted {
195 typedef struct _reg_trie_accepted reg_trie_accepted;
197 /* some basic information about the current match that is created by
198 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
210 /* structures for holding and saving the state maintained by regmatch() */
212 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
214 typedef I32 CHECKPOINT;
216 typedef struct regmatch_state {
217 int resume_state; /* where to jump to on return */
218 char *locinput; /* where to backtrack in string on failure */
222 /* this is a fake union member that matches the first element
223 * of each member that needs to store positive backtrack
226 struct regmatch_state *prev_yes_state;
230 reg_trie_accepted *accept_buff;
231 U32 accepted; /* how many accepting states we have seen */
232 U16 *jump; /* positive offsets from me */
233 regnode *B; /* node following the trie */
234 regnode *me; /* Which node am I - needed for jump tries*/
238 /* this first element must match u.yes */
239 struct regmatch_state *prev_yes_state;
240 struct regmatch_state *prev_eval;
241 struct regmatch_state *prev_curlyx;
243 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
244 flip when transitioning between
245 inner and outer rexen */
246 CHECKPOINT cp; /* remember current savestack indexes */
248 regnode *B; /* the node following us */
249 U32 close_paren; /* which close bracket is our end */
253 /* this first element must match u.yes */
254 struct regmatch_state *prev_yes_state;
255 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
256 CHECKPOINT cp; /* remember current savestack index */
258 int parenfloor;/* how far back to strip paren data */
259 int min; /* the minimal number of A's to match */
260 int max; /* the maximal number of A's to match */
261 regnode *A, *B; /* the nodes corresponding to /A*B/ */
263 /* these two are modified by WHILEM */
264 int count; /* how many instances of A we've matched */
265 char *lastloc;/* where previous A matched (0-len detect) */
269 /* this first element must match u.yes */
270 struct regmatch_state *prev_yes_state;
271 struct regmatch_state *save_curlyx;
272 CHECKPOINT cp; /* remember current savestack indexes */
274 char *save_lastloc; /* previous curlyx.lastloc */
281 regnode *next_branch; /* next branch node */
286 /* this first element must match u.yes */
287 struct regmatch_state *prev_yes_state;
288 I32 c1, c2; /* case fold search */
290 I32 alen; /* length of first-matched A string */
293 regnode *A, *B; /* the nodes corresponding to /A*B/ */
294 regnode *me; /* the curlym node */
300 I32 c1, c2; /* case fold search */
301 char *maxpos; /* highest possible point in string to match */
302 char *oldloc; /* the previous locinput */
304 int min, max; /* {m,n} */
305 regnode *A, *B; /* the nodes corresponding to /A*B/ */
306 } curly; /* and CURLYN/PLUS/STAR */
309 /* this first element must match u.yes */
310 struct regmatch_state *prev_yes_state;
312 I32 logical; /* saved copy of 'logical' var */
313 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
314 } ifmatch; /* and SUSPEND/UNLESSM */
317 /* this first element must match u.yes */
318 struct regmatch_state *prev_yes_state;
319 struct regmatch_state *prev_mark;
326 /* how many regmatch_state structs to allocate as a single slab.
327 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
328 * pointers, plus 1 for any mythical malloc overhead. */
330 #define PERL_REGMATCH_SLAB_SLOTS \
331 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
333 typedef struct regmatch_slab {
334 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
335 struct regmatch_slab *prev, *next;
338 #define PL_reg_flags PL_reg_state.re_state_reg_flags
339 #define PL_bostr PL_reg_state.re_state_bostr
340 #define PL_reginput PL_reg_state.re_state_reginput
341 #define PL_regeol PL_reg_state.re_state_regeol
342 #define PL_regstartp PL_reg_state.re_state_regstartp
343 #define PL_regendp PL_reg_state.re_state_regendp
344 #define PL_reglastparen PL_reg_state.re_state_reglastparen
345 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
346 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
347 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
348 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
349 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
350 #define PL_reg_magic PL_reg_state.re_state_reg_magic
351 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
352 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
353 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
354 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
355 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
356 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
357 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
358 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
359 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
360 #define PL_regsize PL_reg_state.re_state_regsize
361 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
362 #define PL_nrs PL_reg_state.re_state_nrs
364 struct re_save_state {
365 U32 re_state_reg_flags; /* from regexec.c */
366 char *re_state_bostr;
367 char *re_state_reginput; /* String-input pointer. */
368 char *re_state_regeol; /* End of input, for $ check. */
369 I32 *re_state_regstartp; /* Pointer to startp array. */
370 I32 *re_state_regendp; /* Ditto for endp. */
371 U32 *re_state_reglastparen; /* Similarly for lastparen. */
372 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
373 char **re_state_reg_start_tmp; /* from regexec.c */
374 U32 re_state_reg_start_tmpl; /* from regexec.c */
375 I32 re_state_reg_eval_set; /* from regexec.c */
376 bool re_state_reg_match_utf8; /* from regexec.c */
377 MAGIC *re_state_reg_magic; /* from regexec.c */
378 I32 re_state_reg_oldpos; /* from regexec.c */
379 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
380 PMOP *re_state_reg_curpm; /* from regexec.c */
381 char *re_state_reg_oldsaved; /* old saved substr during match */
382 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
383 I32 re_state_reg_maxiter; /* max wait until caching pos */
384 I32 re_state_reg_leftiter; /* wait until caching pos */
385 char *re_state_reg_poscache; /* cache of pos of WHILEM */
386 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
387 I32 re_state_regsize; /* from regexec.c */
388 char *re_state_reg_starttry; /* from regexec.c */
389 #ifdef PERL_OLD_COPY_ON_WRITE
390 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 */
394 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
395 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
398 * c-indentation-style: bsd
400 * indent-tabs-mode: t
403 * ex: set ts=8 sts=4 sw=4 noet: