3 * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
4 * 2005, 2006 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 of
19 an external regex engine based on the core one - like re 'debug'*/
27 typedef struct regnode regnode;
29 struct reg_substr_data;
35 struct reg_substr_datum {
38 SV *substr; /* non-utf8 variant */
39 SV *utf8_substr; /* utf8 variant */
42 struct reg_substr_data {
43 struct reg_substr_datum data[3]; /* Actual array */
46 #ifdef PERL_OLD_COPY_ON_WRITE
47 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
51 /* this is ordered such that the most commonly used
52 fields are at the start of the struct */
53 typedef struct regexp {
54 /* what engine created this regexp? */
55 const struct regexp_engine* engine;
57 /* Information about the match that the perl core uses to manage things */
58 U32 extflags; /* Flags used both externally and internally */
59 I32 minlen; /* mininum possible length of string to match */
60 I32 minlenret; /* mininum possible length of $& */
61 U32 gofs; /* chars left of pos that we search from */
62 struct reg_substr_data *substrs; /* substring data about strings that must appear
63 in the final match, used for optimisations */
64 U32 nparens; /* number of capture buffers */
66 /* private engine specific data */
67 U32 intflags; /* Engine Specific Internal flags */
68 void *pprivate; /* Data private to the regex engine which
69 created this object. */
71 /* Data about the last/current match. These are modified during matching*/
72 U32 lastparen; /* last open paren matched */
73 U32 lastcloseparen; /* last close paren matched */
74 I32 *startp; /* Array of offsets from start of string (@-) */
75 I32 *endp; /* Array of offsets from start of string (@+) */
76 char *subbeg; /* saved or original string
77 so \digit works forever. */
78 I32 sublen; /* Length of string pointed by subbeg */
79 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
82 /* Information about the match that isn't often used */
83 char *precomp; /* pre-compilation regular expression */
84 I32 prelen; /* length of precomp */
85 char *wrapped; /* wrapped version of the pattern */
86 I32 wraplen; /* length of wrapped */
87 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
88 HV *paren_names; /* Optional hash of paren names */
90 /* Refcount of this regexp */
91 I32 refcnt; /* Refcount of this regexp */
94 /* used for high speed searches */
95 typedef struct re_scream_pos_data_s
97 char **scream_olds; /* match pos */
98 I32 *scream_pos; /* Internal iterator of scream. */
101 /* regexp_engine structure. This is the dispatch table for regexes.
102 * Any regex engine implementation must be able to build one of these.
104 typedef struct regexp_engine {
105 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
106 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
107 char* strbeg, I32 minend, SV* screamer,
108 void* data, U32 flags);
109 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
110 char *strend, U32 flags,
111 struct re_scream_pos_data_s *data);
112 SV* (*checkstr) (pTHX_ regexp *prog);
113 void (*free) (pTHX_ struct regexp* r);
114 SV* (*numbered_buff_get) (pTHX_ const REGEXP * const rx, I32 paren, SV* usesv);
115 SV* (*named_buff_get)(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags);
117 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
121 /* Flags stored in regexp->extflags
122 * These are used by code external to the regexp engine
124 * Note that flags starting with RXf_PMf_ have exact equivalents
125 * stored in op_pmflags and which are defined in op.h, they are defined
126 * numerically here only for clarity.
129 /* Anchor and GPOS related stuff */
130 #define RXf_ANCH_BOL 0x00000001
131 #define RXf_ANCH_MBOL 0x00000002
132 #define RXf_ANCH_SBOL 0x00000004
133 #define RXf_ANCH_GPOS 0x00000008
134 #define RXf_GPOS_SEEN 0x00000010
135 #define RXf_GPOS_FLOAT 0x00000020
137 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
138 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
139 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
141 /* Flags indicating special patterns */
142 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
143 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
145 /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */
146 #define RXf_PMf_LOCALE 0x00000800 /* use locale */
147 #define RXf_PMf_MULTILINE 0x00001000 /* /m */
148 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */
149 #define RXf_PMf_FOLD 0x00004000 /* /i */
150 #define RXf_PMf_EXTENDED 0x00008000 /* /x */
151 #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */
152 /* these flags are transfered from the PMOP->op_pmflags member during compilation */
153 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
154 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
156 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
157 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
158 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
159 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
160 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
162 /* chars and strings used as regex pattern modifiers
163 * Singlular is a 'c'har, plural is a "string"
165 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
166 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
167 * for its own uses. So 'k' is out as well.
169 #define EXEC_PAT_MOD 'e'
170 #define KEEPCOPY_PAT_MOD 'p'
171 #define ONCE_PAT_MOD 'o'
172 #define GLOBAL_PAT_MOD 'g'
173 #define CONTINUE_PAT_MOD 'c'
174 #define MULTILINE_PAT_MOD 'm'
175 #define SINGLE_PAT_MOD 's'
176 #define IGNORE_PAT_MOD 'i'
177 #define XTENDED_PAT_MOD 'x'
179 #define ONCE_PAT_MODS "o"
180 #define KEEPCOPY_PAT_MODS "p"
181 #define EXEC_PAT_MODS "e"
182 #define LOOP_PAT_MODS "gc"
184 #define STD_PAT_MODS "msix"
186 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
188 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
189 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
190 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
191 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
194 /* What we have seen */
195 #define RXf_LOOKBEHIND_SEEN 0x00020000
196 #define RXf_EVAL_SEEN 0x00040000
197 #define RXf_CANY_SEEN 0x00080000
200 #define RXf_NOSCAN 0x00100000
201 #define RXf_CHECK_ALL 0x00200000
204 #define RXf_UTF8 0x00400000
205 #define RXf_MATCH_UTF8 0x00800000
208 #define RXf_USE_INTUIT_NOML 0x01000000
209 #define RXf_USE_INTUIT_ML 0x02000000
210 #define RXf_INTUIT_TAIL 0x04000000
212 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
214 /* Copy and tainted info */
215 #define RXf_COPY_DONE 0x10000000
216 #define RXf_TAINTED_SEEN 0x20000000
220 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
221 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
222 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
223 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
224 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
225 ? RX_MATCH_TAINTED_on(prog) \
226 : RX_MATCH_TAINTED_off(prog))
228 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
229 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
230 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
231 #define RX_MATCH_COPIED_set(prog,t) ((t) \
232 ? RX_MATCH_COPIED_on(prog) \
233 : RX_MATCH_COPIED_off(prog))
235 #endif /* PLUGGABLE_RE_EXTENSION */
237 /* Stuff that needs to be included in the plugable extension goes below here */
239 #ifdef PERL_OLD_COPY_ON_WRITE
240 #define RX_MATCH_COPY_FREE(rx) \
241 STMT_START {if (rx->saved_copy) { \
242 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
244 if (RX_MATCH_COPIED(rx)) { \
245 Safefree(rx->subbeg); \
246 RX_MATCH_COPIED_off(rx); \
249 #define RX_MATCH_COPY_FREE(rx) \
250 STMT_START {if (RX_MATCH_COPIED(rx)) { \
251 Safefree(rx->subbeg); \
252 RX_MATCH_COPIED_off(rx); \
256 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
257 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
258 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
259 #define RX_MATCH_UTF8_set(prog, t) ((t) \
260 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
261 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
263 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
264 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
265 #define REXEC_SCREAM 0x04 /* use scream table. */
266 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
267 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
269 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
270 #define ReREFCNT_dec(re) CALLREGFREE(re)
272 #define FBMcf_TAIL_DOLLAR 1
273 #define FBMcf_TAIL_DOLLARM 2
274 #define FBMcf_TAIL_Z 4
275 #define FBMcf_TAIL_z 8
276 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
278 #define FBMrf_MULTILINE 1
280 /* an accepting state/position*/
281 struct _reg_trie_accepted {
285 typedef struct _reg_trie_accepted reg_trie_accepted;
287 /* some basic information about the current match that is created by
288 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
300 /* structures for holding and saving the state maintained by regmatch() */
302 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
303 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
306 typedef I32 CHECKPOINT;
308 typedef struct regmatch_state {
309 int resume_state; /* where to jump to on return */
310 char *locinput; /* where to backtrack in string on failure */
314 /* this is a fake union member that matches the first element
315 * of each member that needs to store positive backtrack
318 struct regmatch_state *prev_yes_state;
321 /* branchlike members */
322 /* this is a fake union member that matches the first elements
323 * of each member that needs to behave like a branch */
325 /* this first element must match u.yes */
326 struct regmatch_state *prev_yes_state;
333 /* the first elements must match u.branchlike */
334 struct regmatch_state *prev_yes_state;
338 regnode *next_branch; /* next branch node */
342 /* the first elements must match u.branchlike */
343 struct regmatch_state *prev_yes_state;
347 reg_trie_accepted *accept_buff; /* accepting states we have seen */
348 U32 accepted; /* how many accepting states we have seen */
349 U16 *jump; /* positive offsets from me */
350 regnode *B; /* node following the trie */
351 regnode *me; /* Which node am I - needed for jump tries*/
354 /* special types - these members are used to store state for special
355 regops like eval, if/then, lookaround and the markpoint state */
357 /* this first element must match u.yes */
358 struct regmatch_state *prev_yes_state;
359 struct regmatch_state *prev_eval;
360 struct regmatch_state *prev_curlyx;
362 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
363 flip when transitioning between
364 inner and outer rexen */
365 CHECKPOINT cp; /* remember current savestack indexes */
367 regnode *B; /* the node following us */
368 U32 close_paren; /* which close bracket is our end */
372 /* this first element must match u.yes */
373 struct regmatch_state *prev_yes_state;
375 I32 logical; /* saved copy of 'logical' var */
376 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
377 } ifmatch; /* and SUSPEND/UNLESSM */
380 /* this first element must match u.yes */
381 struct regmatch_state *prev_yes_state;
382 struct regmatch_state *prev_mark;
391 /* quantifiers - these members are used for storing state for
392 for the regops used to implement quantifiers */
394 /* this first element must match u.yes */
395 struct regmatch_state *prev_yes_state;
396 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
397 CHECKPOINT cp; /* remember current savestack index */
399 int parenfloor;/* how far back to strip paren data */
400 int min; /* the minimal number of A's to match */
401 int max; /* the maximal number of A's to match */
402 regnode *A, *B; /* the nodes corresponding to /A*B/ */
404 /* these two are modified by WHILEM */
405 int count; /* how many instances of A we've matched */
406 char *lastloc;/* where previous A matched (0-len detect) */
410 /* this first element must match u.yes */
411 struct regmatch_state *prev_yes_state;
412 struct regmatch_state *save_curlyx;
413 CHECKPOINT cp; /* remember current savestack indexes */
415 char *save_lastloc; /* previous curlyx.lastloc */
421 /* this first element must match u.yes */
422 struct regmatch_state *prev_yes_state;
423 I32 c1, c2; /* case fold search */
425 I32 alen; /* length of first-matched A string */
428 regnode *A, *B; /* the nodes corresponding to /A*B/ */
429 regnode *me; /* the curlym node */
435 I32 c1, c2; /* case fold search */
436 char *maxpos; /* highest possible point in string to match */
437 char *oldloc; /* the previous locinput */
439 int min, max; /* {m,n} */
440 regnode *A, *B; /* the nodes corresponding to /A*B/ */
441 } curly; /* and CURLYN/PLUS/STAR */
446 /* how many regmatch_state structs to allocate as a single slab.
447 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
448 * pointers, plus 1 for any mythical malloc overhead. */
450 #define PERL_REGMATCH_SLAB_SLOTS \
451 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
453 typedef struct regmatch_slab {
454 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
455 struct regmatch_slab *prev, *next;
458 #define PL_reg_flags PL_reg_state.re_state_reg_flags
459 #define PL_bostr PL_reg_state.re_state_bostr
460 #define PL_reginput PL_reg_state.re_state_reginput
461 #define PL_regeol PL_reg_state.re_state_regeol
462 #define PL_regstartp PL_reg_state.re_state_regstartp
463 #define PL_regendp PL_reg_state.re_state_regendp
464 #define PL_reglastparen PL_reg_state.re_state_reglastparen
465 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
466 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
467 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
468 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
469 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
470 #define PL_reg_magic PL_reg_state.re_state_reg_magic
471 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
472 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
473 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
474 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
475 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
476 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
477 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
478 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
479 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
480 #define PL_regsize PL_reg_state.re_state_regsize
481 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
482 #define PL_nrs PL_reg_state.re_state_nrs
484 struct re_save_state {
485 U32 re_state_reg_flags; /* from regexec.c */
486 char *re_state_bostr;
487 char *re_state_reginput; /* String-input pointer. */
488 char *re_state_regeol; /* End of input, for $ check. */
489 I32 *re_state_regstartp; /* Pointer to startp array. */
490 I32 *re_state_regendp; /* Ditto for endp. */
491 U32 *re_state_reglastparen; /* Similarly for lastparen. */
492 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
493 char **re_state_reg_start_tmp; /* from regexec.c */
494 U32 re_state_reg_start_tmpl; /* from regexec.c */
495 I32 re_state_reg_eval_set; /* from regexec.c */
496 bool re_state_reg_match_utf8; /* from regexec.c */
497 MAGIC *re_state_reg_magic; /* from regexec.c */
498 I32 re_state_reg_oldpos; /* from regexec.c */
499 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
500 PMOP *re_state_reg_curpm; /* from regexec.c */
501 char *re_state_reg_oldsaved; /* old saved substr during match */
502 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
503 I32 re_state_reg_maxiter; /* max wait until caching pos */
504 I32 re_state_reg_leftiter; /* wait until caching pos */
505 char *re_state_reg_poscache; /* cache of pos of WHILEM */
506 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
507 U32 re_state_regsize; /* from regexec.c */
508 char *re_state_reg_starttry; /* from regexec.c */
509 #ifdef PERL_OLD_COPY_ON_WRITE
510 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 */
514 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
515 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
519 * c-indentation-style: bsd
521 * indent-tabs-mode: t
524 * ex: set ts=8 sts=4 sw=4 noet: