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_paren_ofs {
40 #ifdef PERL_OLD_COPY_ON_WRITE
41 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
46 typedef struct regexp {
48 const struct regexp_engine* engine; /* what created this regexp? */
49 I32 refcnt; /* Refcount of this regexp */
51 /* The original string as passed to the compilation routine */
52 char *precomp; /* pre-compilation regular expression */
53 I32 prelen; /* length of precomp */
55 /* Used for generic optimisations by the perl core.
56 All engines are expected to provide this information. */
57 U32 extflags; /* Flags used both externally and internally */
58 I32 minlen; /* mininum possible length of string to match */
59 I32 minlenret; /* mininum possible length of $& */
60 U32 gofs; /* chars left of pos that we search from */
61 U32 nparens; /* number of capture buffers */
62 HV *paren_names; /* Optional hash of paren names */
63 struct reg_substr_data *substrs; /* substring data about strings that must appear
64 in the final match, used for optimisations */
66 /* Data about the last/current match. Used by the core and therefore
67 must be populated by all engines. */
68 char *subbeg; /* saved or original string
69 so \digit works forever. */
70 I32 sublen; /* Length of string pointed by subbeg */
71 I32 *startp; /* Array of offsets from start of string (@-) */
72 I32 *endp; /* Array of offsets from start of string (@+) */
74 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
75 U32 lastparen; /* last open paren matched */
76 U32 lastcloseparen; /* last close paren matched */
78 /* Perl Regex Engine specific data. Other engines shouldn't need
79 to touch this. Should be refactored out into a different structure
80 and accessed via the *pprivate field. (except intflags) */
81 U32 intflags; /* Internal flags */
82 void *pprivate; /* Data private to the regex engine which
83 created this object. Perl will never mess with
84 this member at all. */
88 typedef struct regexp_internal {
89 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
90 U32 *offsets; /* offset annotations 20001228 MJD
91 data about mapping the program to the
93 regnode *regstclass; /* Optional startclass as identified or constructed
95 struct reg_data *data; /* Additional miscellaneous data used by the program.
96 Used to make it easier to clone and free arbitrary
97 data that the regops need. Often the ARG field of
98 a regop is an index into this structure */
99 regnode program[1]; /* Unwarranted chumminess with compiler. */
102 #define RXi_SET(x,y) (x)->pprivate = (void*)(y)
103 #define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
104 #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
106 typedef struct re_scream_pos_data_s
108 char **scream_olds; /* match pos */
109 I32 *scream_pos; /* Internal iterator of scream. */
110 } re_scream_pos_data;
112 typedef struct regexp_engine {
113 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
114 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
115 char* strbeg, I32 minend, SV* screamer,
116 void* data, U32 flags);
117 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
118 char *strend, U32 flags,
119 struct re_scream_pos_data_s *data);
120 SV* (*checkstr) (pTHX_ regexp *prog);
121 void (*free) (pTHX_ struct regexp* r);
122 char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval);
124 regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
129 * Flags stored in regexp->intflags
130 * These are used only internally to the regexp engine
132 #define PREGf_SKIP 0x00000001
133 #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
134 #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
135 #define PREGf_VERBARG_SEEN 0x00000008
136 #define PREGf_CUTGROUP_SEEN 0x00000010
139 /* Flags stored in regexp->extflags
140 * These are used by code external to the regexp engine
143 /* Anchor and GPOS related stuff */
144 #define RXf_ANCH_BOL 0x00000001
145 #define RXf_ANCH_MBOL 0x00000002
146 #define RXf_ANCH_SBOL 0x00000004
147 #define RXf_ANCH_GPOS 0x00000008
148 #define RXf_GPOS_SEEN 0x00000010
149 #define RXf_GPOS_FLOAT 0x00000020
151 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
152 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
153 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
155 * 0xF800 of extflags is used by PMf_COMPILETIME
156 * These are the regex equivelent of the PMf_xyz stuff defined
159 #define RXf_PMf_LOCALE 0x00000800
160 #define RXf_PMf_MULTILINE 0x00001000
161 #define RXf_PMf_SINGLELINE 0x00002000
162 #define RXf_PMf_FOLD 0x00004000
163 #define RXf_PMf_EXTENDED 0x00008000
164 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
166 /* What we have seen */
168 #define RXf_LOOKBEHIND_SEEN 0x00020000
169 #define RXf_EVAL_SEEN 0x00040000
170 #define RXf_CANY_SEEN 0x00080000
173 #define RXf_NOSCAN 0x00100000
174 #define RXf_CHECK_ALL 0x00200000
177 #define RXf_UTF8 0x00400000
178 #define RXf_MATCH_UTF8 0x00800000
181 #define RXf_USE_INTUIT_NOML 0x01000000
182 #define RXf_USE_INTUIT_ML 0x02000000
183 #define RXf_INTUIT_TAIL 0x04000000
185 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
187 /* Copy and tainted info */
188 #define RXf_COPY_DONE 0x10000000
189 #define RXf_TAINTED_SEEN 0x20000000
193 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
194 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
195 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
196 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
197 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
198 ? RX_MATCH_TAINTED_on(prog) \
199 : RX_MATCH_TAINTED_off(prog))
201 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
202 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
203 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
204 #define RX_MATCH_COPIED_set(prog,t) ((t) \
205 ? RX_MATCH_COPIED_on(prog) \
206 : RX_MATCH_COPIED_off(prog))
208 #endif /* PLUGGABLE_RE_EXTENSION */
210 /* Stuff that needs to be included in the plugable extension goes below here */
212 #ifdef PERL_OLD_COPY_ON_WRITE
213 #define RX_MATCH_COPY_FREE(rx) \
214 STMT_START {if (rx->saved_copy) { \
215 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
217 if (RX_MATCH_COPIED(rx)) { \
218 Safefree(rx->subbeg); \
219 RX_MATCH_COPIED_off(rx); \
222 #define RX_MATCH_COPY_FREE(rx) \
223 STMT_START {if (RX_MATCH_COPIED(rx)) { \
224 Safefree(rx->subbeg); \
225 RX_MATCH_COPIED_off(rx); \
229 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
230 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
231 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
232 #define RX_MATCH_UTF8_set(prog, t) ((t) \
233 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
234 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
236 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
237 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
238 #define REXEC_SCREAM 0x04 /* use scream table. */
239 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
240 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
242 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
243 #define ReREFCNT_dec(re) CALLREGFREE(re)
245 #define FBMcf_TAIL_DOLLAR 1
246 #define FBMcf_TAIL_DOLLARM 2
247 #define FBMcf_TAIL_Z 4
248 #define FBMcf_TAIL_z 8
249 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
251 #define FBMrf_MULTILINE 1
253 /* an accepting state/position*/
254 struct _reg_trie_accepted {
258 typedef struct _reg_trie_accepted reg_trie_accepted;
260 /* some basic information about the current match that is created by
261 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
273 /* structures for holding and saving the state maintained by regmatch() */
275 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
277 typedef I32 CHECKPOINT;
279 typedef struct regmatch_state {
280 int resume_state; /* where to jump to on return */
281 char *locinput; /* where to backtrack in string on failure */
285 /* this is a fake union member that matches the first element
286 * of each member that needs to store positive backtrack
289 struct regmatch_state *prev_yes_state;
293 /* this first element must match u.yes */
294 struct regmatch_state *prev_yes_state;
295 reg_trie_accepted *accept_buff;
296 U32 accepted; /* how many accepting states we have seen */
297 U16 *jump; /* positive offsets from me */
298 regnode *B; /* node following the trie */
299 regnode *me; /* Which node am I - needed for jump tries*/
303 /* this first element must match u.yes */
304 struct regmatch_state *prev_yes_state;
305 struct regmatch_state *prev_eval;
306 struct regmatch_state *prev_curlyx;
308 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
309 flip when transitioning between
310 inner and outer rexen */
311 CHECKPOINT cp; /* remember current savestack indexes */
313 regnode *B; /* the node following us */
314 U32 close_paren; /* which close bracket is our end */
318 /* this first element must match u.yes */
319 struct regmatch_state *prev_yes_state;
320 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
321 CHECKPOINT cp; /* remember current savestack index */
323 int parenfloor;/* how far back to strip paren data */
324 int min; /* the minimal number of A's to match */
325 int max; /* the maximal number of A's to match */
326 regnode *A, *B; /* the nodes corresponding to /A*B/ */
328 /* these two are modified by WHILEM */
329 int count; /* how many instances of A we've matched */
330 char *lastloc;/* where previous A matched (0-len detect) */
334 /* this first element must match u.yes */
335 struct regmatch_state *prev_yes_state;
336 struct regmatch_state *save_curlyx;
337 CHECKPOINT cp; /* remember current savestack indexes */
339 char *save_lastloc; /* previous curlyx.lastloc */
345 /* this first element must match u.yes */
346 struct regmatch_state *prev_yes_state;
348 regnode *next_branch; /* next branch node */
353 /* this first element must match u.yes */
354 struct regmatch_state *prev_yes_state;
355 I32 c1, c2; /* case fold search */
357 I32 alen; /* length of first-matched A string */
360 regnode *A, *B; /* the nodes corresponding to /A*B/ */
361 regnode *me; /* the curlym node */
367 I32 c1, c2; /* case fold search */
368 char *maxpos; /* highest possible point in string to match */
369 char *oldloc; /* the previous locinput */
371 int min, max; /* {m,n} */
372 regnode *A, *B; /* the nodes corresponding to /A*B/ */
373 } curly; /* and CURLYN/PLUS/STAR */
376 /* this first element must match u.yes */
377 struct regmatch_state *prev_yes_state;
379 I32 logical; /* saved copy of 'logical' var */
380 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
381 } ifmatch; /* and SUSPEND/UNLESSM */
384 /* this first element must match u.yes */
385 struct regmatch_state *prev_yes_state;
386 struct regmatch_state *prev_mark;
393 /* how many regmatch_state structs to allocate as a single slab.
394 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
395 * pointers, plus 1 for any mythical malloc overhead. */
397 #define PERL_REGMATCH_SLAB_SLOTS \
398 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
400 typedef struct regmatch_slab {
401 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
402 struct regmatch_slab *prev, *next;
405 #define PL_reg_flags PL_reg_state.re_state_reg_flags
406 #define PL_bostr PL_reg_state.re_state_bostr
407 #define PL_reginput PL_reg_state.re_state_reginput
408 #define PL_regeol PL_reg_state.re_state_regeol
409 #define PL_regstartp PL_reg_state.re_state_regstartp
410 #define PL_regendp PL_reg_state.re_state_regendp
411 #define PL_reglastparen PL_reg_state.re_state_reglastparen
412 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
413 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
414 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
415 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
416 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
417 #define PL_reg_magic PL_reg_state.re_state_reg_magic
418 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
419 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
420 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
421 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
422 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
423 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
424 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
425 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
426 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
427 #define PL_regsize PL_reg_state.re_state_regsize
428 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
429 #define PL_nrs PL_reg_state.re_state_nrs
431 struct re_save_state {
432 U32 re_state_reg_flags; /* from regexec.c */
433 char *re_state_bostr;
434 char *re_state_reginput; /* String-input pointer. */
435 char *re_state_regeol; /* End of input, for $ check. */
436 I32 *re_state_regstartp; /* Pointer to startp array. */
437 I32 *re_state_regendp; /* Ditto for endp. */
438 U32 *re_state_reglastparen; /* Similarly for lastparen. */
439 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
440 char **re_state_reg_start_tmp; /* from regexec.c */
441 U32 re_state_reg_start_tmpl; /* from regexec.c */
442 I32 re_state_reg_eval_set; /* from regexec.c */
443 bool re_state_reg_match_utf8; /* from regexec.c */
444 MAGIC *re_state_reg_magic; /* from regexec.c */
445 I32 re_state_reg_oldpos; /* from regexec.c */
446 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
447 PMOP *re_state_reg_curpm; /* from regexec.c */
448 char *re_state_reg_oldsaved; /* old saved substr during match */
449 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
450 I32 re_state_reg_maxiter; /* max wait until caching pos */
451 I32 re_state_reg_leftiter; /* wait until caching pos */
452 char *re_state_reg_poscache; /* cache of pos of WHILEM */
453 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
454 U32 re_state_regsize; /* from regexec.c */
455 char *re_state_reg_starttry; /* from regexec.c */
456 #ifdef PERL_OLD_COPY_ON_WRITE
457 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 */
461 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
462 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
465 * c-indentation-style: bsd
467 * indent-tabs-mode: t
470 * ex: set ts=8 sts=4 sw=4 noet: