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;
36 struct reg_substr_datum {
39 SV *substr; /* non-utf8 variant */
40 SV *utf8_substr; /* utf8 variant */
43 struct reg_substr_data {
44 struct reg_substr_datum data[3]; /* Actual array */
47 #ifdef PERL_OLD_COPY_ON_WRITE
48 #define SV_SAVED_COPY SV *saved_copy; /* If non-NULL, SV which is COW from original */
53 typedef struct regexp_paren_pair {
58 /* this is ordered such that the most commonly used
59 fields are at the start of the struct */
60 typedef struct regexp {
61 /* what engine created this regexp? */
62 const struct regexp_engine* engine;
63 struct regexp* mother_re; /* what re is this a lightweight copy of? */
65 /* Information about the match that the perl core uses to manage things */
66 U32 extflags; /* Flags used both externally and internally */
67 I32 minlen; /* mininum possible length of string to match */
68 I32 minlenret; /* mininum possible length of $& */
69 U32 gofs; /* chars left of pos that we search from */
70 struct reg_substr_data *substrs; /* substring data about strings that must appear
71 in the final match, used for optimisations */
72 U32 nparens; /* number of capture buffers */
74 /* private engine specific data */
75 U32 intflags; /* Engine Specific Internal flags */
76 void *pprivate; /* Data private to the regex engine which
77 created this object. */
79 /* Data about the last/current match. These are modified during matching*/
80 U32 lastparen; /* last open paren matched */
81 U32 lastcloseparen; /* last close paren matched */
82 regexp_paren_pair *swap; /* Swap copy of *offs */
83 regexp_paren_pair *offs; /* Array of offsets for (@-) and (@+) */
85 char *subbeg; /* saved or original string
86 so \digit works forever. */
87 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
88 I32 sublen; /* Length of string pointed by subbeg */
91 /* Information about the match that isn't often used */
92 I32 prelen; /* length of precomp */
93 const char *precomp; /* pre-compilation regular expression */
94 /* wrapped can't be const char*, as it is returned by sv_2pv_flags */
95 char *wrapped; /* wrapped version of the pattern */
96 I32 wraplen; /* length of wrapped */
97 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
98 HV *paren_names; /* Optional hash of paren names */
100 /* Refcount of this regexp */
101 I32 refcnt; /* Refcount of this regexp */
104 /* used for high speed searches */
105 typedef struct re_scream_pos_data_s
107 char **scream_olds; /* match pos */
108 I32 *scream_pos; /* Internal iterator of scream. */
109 } re_scream_pos_data;
111 /* regexp_engine structure. This is the dispatch table for regexes.
112 * Any regex engine implementation must be able to build one of these.
114 typedef struct regexp_engine {
115 REGEXP* (*comp) (pTHX_ const SV * const pattern, const U32 flags);
116 I32 (*exec) (pTHX_ REGEXP * const rx, char* stringarg, char* strend,
117 char* strbeg, I32 minend, SV* screamer,
118 void* data, U32 flags);
119 char* (*intuit) (pTHX_ REGEXP * const rx, SV *sv, char *strpos,
120 char *strend, const U32 flags,
121 re_scream_pos_data *data);
122 SV* (*checkstr) (pTHX_ REGEXP * const rx);
123 void (*free) (pTHX_ REGEXP * const rx);
124 void (*numbered_buff_FETCH) (pTHX_ REGEXP * const rx, const I32 paren,
126 void (*numbered_buff_STORE) (pTHX_ REGEXP * const rx, const I32 paren,
127 SV const * const value);
128 I32 (*numbered_buff_LENGTH) (pTHX_ REGEXP * const rx, const SV * const sv,
130 SV* (*named_buff_FETCH) (pTHX_ REGEXP * const rx, SV * const key,
132 SV* (*qr_package)(pTHX_ REGEXP * const rx);
134 void* (*dupe) (pTHX_ REGEXP * const rx, CLONE_PARAMS *param);
138 /* Flags stored in regexp->extflags
139 * These are used by code external to the regexp engine
141 * Note that flags starting with RXf_PMf_ have exact equivalents
142 * stored in op_pmflags and which are defined in op.h, they are defined
143 * numerically here only for clarity.
146 /* Anchor and GPOS related stuff */
147 #define RXf_ANCH_BOL 0x00000001
148 #define RXf_ANCH_MBOL 0x00000002
149 #define RXf_ANCH_SBOL 0x00000004
150 #define RXf_ANCH_GPOS 0x00000008
151 #define RXf_GPOS_SEEN 0x00000010
152 #define RXf_GPOS_FLOAT 0x00000020
154 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
155 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
156 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
158 /* Flags indicating special patterns */
159 #define RXf_SKIPWHITE 0x00000100 /* Pattern is for a split / / */
160 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
161 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
163 /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */
164 #define RXf_PMf_LOCALE 0x00000800 /* use locale */
165 #define RXf_PMf_MULTILINE 0x00001000 /* /m */
166 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */
167 #define RXf_PMf_FOLD 0x00004000 /* /i */
168 #define RXf_PMf_EXTENDED 0x00008000 /* /x */
169 #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */
170 /* these flags are transfered from the PMOP->op_pmflags member during compilation */
171 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
172 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
174 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
175 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
176 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
177 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
178 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
180 /* chars and strings used as regex pattern modifiers
181 * Singlular is a 'c'har, plural is a "string"
183 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
184 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
185 * for its own uses. So 'k' is out as well.
187 #define EXEC_PAT_MOD 'e'
188 #define KEEPCOPY_PAT_MOD 'p'
189 #define ONCE_PAT_MOD 'o'
190 #define GLOBAL_PAT_MOD 'g'
191 #define CONTINUE_PAT_MOD 'c'
192 #define MULTILINE_PAT_MOD 'm'
193 #define SINGLE_PAT_MOD 's'
194 #define IGNORE_PAT_MOD 'i'
195 #define XTENDED_PAT_MOD 'x'
197 #define ONCE_PAT_MODS "o"
198 #define KEEPCOPY_PAT_MODS "p"
199 #define EXEC_PAT_MODS "e"
200 #define LOOP_PAT_MODS "gc"
202 #define STD_PAT_MODS "msix"
204 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
206 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
207 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
208 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
209 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
212 /* What we have seen */
213 #define RXf_LOOKBEHIND_SEEN 0x00020000
214 #define RXf_EVAL_SEEN 0x00040000
215 #define RXf_CANY_SEEN 0x00080000
218 #define RXf_NOSCAN 0x00100000
219 #define RXf_CHECK_ALL 0x00200000
222 #define RXf_UTF8 0x00400000
223 #define RXf_MATCH_UTF8 0x00800000
226 #define RXf_USE_INTUIT_NOML 0x01000000
227 #define RXf_USE_INTUIT_ML 0x02000000
228 #define RXf_INTUIT_TAIL 0x04000000
230 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
232 /* Copy and tainted info */
233 #define RXf_COPY_DONE 0x10000000
234 #define RXf_TAINTED_SEEN 0x20000000
235 #define RXf_TAINTED 0x80000000 /* this pattern is tainted */
238 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
239 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
240 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
241 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
242 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
243 ? RX_MATCH_TAINTED_on(prog) \
244 : RX_MATCH_TAINTED_off(prog))
246 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
247 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
248 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
249 #define RX_MATCH_COPIED_set(prog,t) ((t) \
250 ? RX_MATCH_COPIED_on(prog) \
251 : RX_MATCH_COPIED_off(prog))
253 #endif /* PLUGGABLE_RE_EXTENSION */
255 /* Stuff that needs to be included in the plugable extension goes below here */
257 #ifdef PERL_OLD_COPY_ON_WRITE
258 #define RX_MATCH_COPY_FREE(rx) \
259 STMT_START {if (rx->saved_copy) { \
260 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
262 if (RX_MATCH_COPIED(rx)) { \
263 Safefree(rx->subbeg); \
264 RX_MATCH_COPIED_off(rx); \
267 #define RX_MATCH_COPY_FREE(rx) \
268 STMT_START {if (RX_MATCH_COPIED(rx)) { \
269 Safefree(rx->subbeg); \
270 RX_MATCH_COPIED_off(rx); \
274 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
275 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
276 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
277 #define RX_MATCH_UTF8_set(prog, t) ((t) \
278 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
279 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
281 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
282 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
283 #define REXEC_SCREAM 0x04 /* use scream table. */
284 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
285 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
287 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
288 #define ReREFCNT_dec(re) CALLREGFREE(re)
290 #define FBMcf_TAIL_DOLLAR 1
291 #define FBMcf_TAIL_DOLLARM 2
292 #define FBMcf_TAIL_Z 4
293 #define FBMcf_TAIL_z 8
294 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
296 #define FBMrf_MULTILINE 1
298 /* an accepting state/position*/
299 struct _reg_trie_accepted {
303 typedef struct _reg_trie_accepted reg_trie_accepted;
305 /* some basic information about the current match that is created by
306 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
318 /* structures for holding and saving the state maintained by regmatch() */
320 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
321 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
324 typedef I32 CHECKPOINT;
326 typedef struct regmatch_state {
327 int resume_state; /* where to jump to on return */
328 char *locinput; /* where to backtrack in string on failure */
332 /* this is a fake union member that matches the first element
333 * of each member that needs to store positive backtrack
336 struct regmatch_state *prev_yes_state;
339 /* branchlike members */
340 /* this is a fake union member that matches the first elements
341 * of each member that needs to behave like a branch */
343 /* this first element must match u.yes */
344 struct regmatch_state *prev_yes_state;
351 /* the first elements must match u.branchlike */
352 struct regmatch_state *prev_yes_state;
356 regnode *next_branch; /* next branch node */
360 /* the first elements must match u.branchlike */
361 struct regmatch_state *prev_yes_state;
365 reg_trie_accepted *accept_buff; /* accepting states we have seen */
366 U32 accepted; /* how many accepting states we have seen */
367 U16 *jump; /* positive offsets from me */
368 regnode *B; /* node following the trie */
369 regnode *me; /* Which node am I - needed for jump tries*/
372 /* special types - these members are used to store state for special
373 regops like eval, if/then, lookaround and the markpoint state */
375 /* this first element must match u.yes */
376 struct regmatch_state *prev_yes_state;
377 struct regmatch_state *prev_eval;
378 struct regmatch_state *prev_curlyx;
380 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
381 flip when transitioning between
382 inner and outer rexen */
383 CHECKPOINT cp; /* remember current savestack indexes */
385 U32 close_paren; /* which close bracket is our end */
386 regnode *B; /* the node following us */
390 /* this first element must match u.yes */
391 struct regmatch_state *prev_yes_state;
393 I32 logical; /* saved copy of 'logical' var */
394 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
395 } ifmatch; /* and SUSPEND/UNLESSM */
398 /* this first element must match u.yes */
399 struct regmatch_state *prev_yes_state;
400 struct regmatch_state *prev_mark;
409 /* quantifiers - these members are used for storing state for
410 for the regops used to implement quantifiers */
412 /* this first element must match u.yes */
413 struct regmatch_state *prev_yes_state;
414 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
415 regnode *A, *B; /* the nodes corresponding to /A*B/ */
416 CHECKPOINT cp; /* remember current savestack index */
418 int parenfloor;/* how far back to strip paren data */
419 int min; /* the minimal number of A's to match */
420 int max; /* the maximal number of A's to match */
422 /* these two are modified by WHILEM */
423 int count; /* how many instances of A we've matched */
424 char *lastloc;/* where previous A matched (0-len detect) */
428 /* this first element must match u.yes */
429 struct regmatch_state *prev_yes_state;
430 struct regmatch_state *save_curlyx;
431 CHECKPOINT cp; /* remember current savestack indexes */
433 char *save_lastloc; /* previous curlyx.lastloc */
439 /* this first element must match u.yes */
440 struct regmatch_state *prev_yes_state;
441 I32 c1, c2; /* case fold search */
443 I32 alen; /* length of first-matched A string */
446 regnode *A, *B; /* the nodes corresponding to /A*B/ */
447 regnode *me; /* the curlym node */
453 I32 c1, c2; /* case fold search */
454 char *maxpos; /* highest possible point in string to match */
455 char *oldloc; /* the previous locinput */
457 int min, max; /* {m,n} */
458 regnode *A, *B; /* the nodes corresponding to /A*B/ */
459 } curly; /* and CURLYN/PLUS/STAR */
464 /* how many regmatch_state structs to allocate as a single slab.
465 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
466 * pointers, plus 1 for any mythical malloc overhead. */
468 #define PERL_REGMATCH_SLAB_SLOTS \
469 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
471 typedef struct regmatch_slab {
472 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
473 struct regmatch_slab *prev, *next;
476 #define PL_reg_flags PL_reg_state.re_state_reg_flags
477 #define PL_bostr PL_reg_state.re_state_bostr
478 #define PL_reginput PL_reg_state.re_state_reginput
479 #define PL_regeol PL_reg_state.re_state_regeol
480 #define PL_regoffs PL_reg_state.re_state_regoffs
481 #define PL_reglastparen PL_reg_state.re_state_reglastparen
482 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
483 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
484 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
485 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
486 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
487 #define PL_reg_magic PL_reg_state.re_state_reg_magic
488 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
489 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
490 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
491 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
492 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
493 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
494 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
495 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
496 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
497 #define PL_regsize PL_reg_state.re_state_regsize
498 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
499 #define PL_nrs PL_reg_state.re_state_nrs
501 struct re_save_state {
502 U32 re_state_reg_flags; /* from regexec.c */
503 U32 re_state_reg_start_tmpl; /* from regexec.c */
504 I32 re_state_reg_eval_set; /* from regexec.c */
505 bool re_state_reg_match_utf8; /* from regexec.c */
506 char *re_state_bostr;
507 char *re_state_reginput; /* String-input pointer. */
508 char *re_state_regeol; /* End of input, for $ check. */
509 regexp_paren_pair *re_state_regoffs; /* Pointer to start/end pairs */
510 U32 *re_state_reglastparen; /* Similarly for lastparen. */
511 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
512 char **re_state_reg_start_tmp; /* from regexec.c */
513 MAGIC *re_state_reg_magic; /* from regexec.c */
514 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
515 PMOP *re_state_reg_curpm; /* from regexec.c */
516 char *re_state_reg_oldsaved; /* old saved substr during match */
517 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
518 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
519 I32 re_state_reg_oldpos; /* from regexec.c */
520 I32 re_state_reg_maxiter; /* max wait until caching pos */
521 I32 re_state_reg_leftiter; /* wait until caching pos */
522 U32 re_state_regsize; /* from regexec.c */
523 char *re_state_reg_poscache; /* cache of pos of WHILEM */
524 char *re_state_reg_starttry; /* from regexec.c */
525 #ifdef PERL_OLD_COPY_ON_WRITE
526 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 */
530 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
531 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
535 * c-indentation-style: bsd
537 * indent-tabs-mode: t
540 * ex: set ts=8 sts=4 sw=4 noet: