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 /* swap buffer for paren structs */
54 typedef struct regexp_paren_ofs {
59 /* this is ordered such that the most commonly used
60 fields are at the start of the struct */
61 typedef struct regexp {
62 /* what engine created this regexp? */
63 const struct regexp_engine* engine;
64 struct regexp* mother_re; /* what re is this a lightweight copy of? */
66 /* Information about the match that the perl core uses to manage things */
67 U32 extflags; /* Flags used both externally and internally */
68 I32 minlen; /* mininum possible length of string to match */
69 I32 minlenret; /* mininum possible length of $& */
70 U32 gofs; /* chars left of pos that we search from */
71 struct reg_substr_data *substrs; /* substring data about strings that must appear
72 in the final match, used for optimisations */
73 U32 nparens; /* number of capture buffers */
75 /* private engine specific data */
76 U32 intflags; /* Engine Specific Internal flags */
77 void *pprivate; /* Data private to the regex engine which
78 created this object. */
80 /* Data about the last/current match. These are modified during matching*/
81 U32 lastparen; /* last open paren matched */
82 U32 lastcloseparen; /* last close paren matched */
83 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
84 I32 *startp; /* Array of offsets from start of string (@-) */
85 I32 *endp; /* Array of offsets from start of string (@+) */
87 char *subbeg; /* saved or original string
88 so \digit works forever. */
89 I32 sublen; /* Length of string pointed by subbeg */
90 SV_SAVED_COPY /* If non-NULL, SV which is COW from original */
93 /* Information about the match that isn't often used */
94 char *precomp; /* pre-compilation regular expression */
95 I32 prelen; /* length of precomp */
96 char *wrapped; /* wrapped version of the pattern */
97 I32 wraplen; /* length of wrapped */
98 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
99 HV *paren_names; /* Optional hash of paren names */
101 /* Refcount of this regexp */
102 I32 refcnt; /* Refcount of this regexp */
105 /* used for high speed searches */
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 /* regexp_engine structure. This is the dispatch table for regexes.
113 * Any regex engine implementation must be able to build one of these.
115 typedef struct regexp_engine {
116 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
117 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
118 char* strbeg, I32 minend, SV* screamer,
119 void* data, U32 flags);
120 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
121 char *strend, U32 flags,
122 struct re_scream_pos_data_s *data);
123 SV* (*checkstr) (pTHX_ regexp *prog);
124 void (*free) (pTHX_ struct regexp* r);
125 SV* (*numbered_buff_get) (pTHX_ const REGEXP * const rx, I32 paren, SV* usesv);
126 SV* (*named_buff_get)(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags);
128 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
132 /* Flags stored in regexp->extflags
133 * These are used by code external to the regexp engine
135 * Note that flags starting with RXf_PMf_ have exact equivalents
136 * stored in op_pmflags and which are defined in op.h, they are defined
137 * numerically here only for clarity.
140 /* Anchor and GPOS related stuff */
141 #define RXf_ANCH_BOL 0x00000001
142 #define RXf_ANCH_MBOL 0x00000002
143 #define RXf_ANCH_SBOL 0x00000004
144 #define RXf_ANCH_GPOS 0x00000008
145 #define RXf_GPOS_SEEN 0x00000010
146 #define RXf_GPOS_FLOAT 0x00000020
148 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
149 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
150 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
152 /* Flags indicating special patterns */
153 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
154 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
156 /* 0x1F800 of extflags is used by (RXf_)PMf_COMPILETIME */
157 #define RXf_PMf_LOCALE 0x00000800 /* use locale */
158 #define RXf_PMf_MULTILINE 0x00001000 /* /m */
159 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */
160 #define RXf_PMf_FOLD 0x00004000 /* /i */
161 #define RXf_PMf_EXTENDED 0x00008000 /* /x */
162 #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */
163 /* these flags are transfered from the PMOP->op_pmflags member during compilation */
164 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
165 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
167 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
168 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
169 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
170 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
171 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
173 /* chars and strings used as regex pattern modifiers
174 * Singlular is a 'c'har, plural is a "string"
176 * NOTE, KEEPCOPY was originally 'k', but was changed to 'p' for preserve
177 * for compatibility reasons with Regexp::Common which highjacked (?k:...)
178 * for its own uses. So 'k' is out as well.
180 #define EXEC_PAT_MOD 'e'
181 #define KEEPCOPY_PAT_MOD 'p'
182 #define ONCE_PAT_MOD 'o'
183 #define GLOBAL_PAT_MOD 'g'
184 #define CONTINUE_PAT_MOD 'c'
185 #define MULTILINE_PAT_MOD 'm'
186 #define SINGLE_PAT_MOD 's'
187 #define IGNORE_PAT_MOD 'i'
188 #define XTENDED_PAT_MOD 'x'
190 #define ONCE_PAT_MODS "o"
191 #define KEEPCOPY_PAT_MODS "p"
192 #define EXEC_PAT_MODS "e"
193 #define LOOP_PAT_MODS "gc"
195 #define STD_PAT_MODS "msix"
197 #define INT_PAT_MODS STD_PAT_MODS KEEPCOPY_PAT_MODS
199 #define EXT_PAT_MODS ONCE_PAT_MODS KEEPCOPY_PAT_MODS
200 #define QR_PAT_MODS STD_PAT_MODS EXT_PAT_MODS
201 #define M_PAT_MODS QR_PAT_MODS LOOP_PAT_MODS
202 #define S_PAT_MODS M_PAT_MODS EXEC_PAT_MODS
205 /* What we have seen */
206 #define RXf_LOOKBEHIND_SEEN 0x00020000
207 #define RXf_EVAL_SEEN 0x00040000
208 #define RXf_CANY_SEEN 0x00080000
211 #define RXf_NOSCAN 0x00100000
212 #define RXf_CHECK_ALL 0x00200000
215 #define RXf_UTF8 0x00400000
216 #define RXf_MATCH_UTF8 0x00800000
219 #define RXf_USE_INTUIT_NOML 0x01000000
220 #define RXf_USE_INTUIT_ML 0x02000000
221 #define RXf_INTUIT_TAIL 0x04000000
223 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
225 /* Copy and tainted info */
226 #define RXf_COPY_DONE 0x10000000
227 #define RXf_TAINTED_SEEN 0x20000000
230 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
231 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
232 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
233 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
234 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
235 ? RX_MATCH_TAINTED_on(prog) \
236 : RX_MATCH_TAINTED_off(prog))
238 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
239 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
240 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
241 #define RX_MATCH_COPIED_set(prog,t) ((t) \
242 ? RX_MATCH_COPIED_on(prog) \
243 : RX_MATCH_COPIED_off(prog))
245 #endif /* PLUGGABLE_RE_EXTENSION */
247 /* Stuff that needs to be included in the plugable extension goes below here */
249 #ifdef PERL_OLD_COPY_ON_WRITE
250 #define RX_MATCH_COPY_FREE(rx) \
251 STMT_START {if (rx->saved_copy) { \
252 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
254 if (RX_MATCH_COPIED(rx)) { \
255 Safefree(rx->subbeg); \
256 RX_MATCH_COPIED_off(rx); \
259 #define RX_MATCH_COPY_FREE(rx) \
260 STMT_START {if (RX_MATCH_COPIED(rx)) { \
261 Safefree(rx->subbeg); \
262 RX_MATCH_COPIED_off(rx); \
266 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
267 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
268 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
269 #define RX_MATCH_UTF8_set(prog, t) ((t) \
270 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
271 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
273 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
274 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
275 #define REXEC_SCREAM 0x04 /* use scream table. */
276 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
277 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
279 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
280 #define ReREFCNT_dec(re) CALLREGFREE(re)
282 #define FBMcf_TAIL_DOLLAR 1
283 #define FBMcf_TAIL_DOLLARM 2
284 #define FBMcf_TAIL_Z 4
285 #define FBMcf_TAIL_z 8
286 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
288 #define FBMrf_MULTILINE 1
290 /* an accepting state/position*/
291 struct _reg_trie_accepted {
295 typedef struct _reg_trie_accepted reg_trie_accepted;
297 /* some basic information about the current match that is created by
298 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
310 /* structures for holding and saving the state maintained by regmatch() */
312 #ifndef MAX_RECURSE_EVAL_NOCHANGE_DEPTH
313 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 1000
316 typedef I32 CHECKPOINT;
318 typedef struct regmatch_state {
319 int resume_state; /* where to jump to on return */
320 char *locinput; /* where to backtrack in string on failure */
324 /* this is a fake union member that matches the first element
325 * of each member that needs to store positive backtrack
328 struct regmatch_state *prev_yes_state;
331 /* branchlike members */
332 /* this is a fake union member that matches the first elements
333 * of each member that needs to behave like a branch */
335 /* this first element must match u.yes */
336 struct regmatch_state *prev_yes_state;
343 /* the first elements must match u.branchlike */
344 struct regmatch_state *prev_yes_state;
348 regnode *next_branch; /* next branch node */
352 /* the first elements must match u.branchlike */
353 struct regmatch_state *prev_yes_state;
357 reg_trie_accepted *accept_buff; /* accepting states we have seen */
358 U32 accepted; /* how many accepting states we have seen */
359 U16 *jump; /* positive offsets from me */
360 regnode *B; /* node following the trie */
361 regnode *me; /* Which node am I - needed for jump tries*/
364 /* special types - these members are used to store state for special
365 regops like eval, if/then, lookaround and the markpoint state */
367 /* this first element must match u.yes */
368 struct regmatch_state *prev_yes_state;
369 struct regmatch_state *prev_eval;
370 struct regmatch_state *prev_curlyx;
372 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
373 flip when transitioning between
374 inner and outer rexen */
375 CHECKPOINT cp; /* remember current savestack indexes */
377 regnode *B; /* the node following us */
378 U32 close_paren; /* which close bracket is our end */
382 /* this first element must match u.yes */
383 struct regmatch_state *prev_yes_state;
385 I32 logical; /* saved copy of 'logical' var */
386 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
387 } ifmatch; /* and SUSPEND/UNLESSM */
390 /* this first element must match u.yes */
391 struct regmatch_state *prev_yes_state;
392 struct regmatch_state *prev_mark;
401 /* quantifiers - these members are used for storing state for
402 for the regops used to implement quantifiers */
404 /* this first element must match u.yes */
405 struct regmatch_state *prev_yes_state;
406 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
407 CHECKPOINT cp; /* remember current savestack index */
409 int parenfloor;/* how far back to strip paren data */
410 int min; /* the minimal number of A's to match */
411 int max; /* the maximal number of A's to match */
412 regnode *A, *B; /* the nodes corresponding to /A*B/ */
414 /* these two are modified by WHILEM */
415 int count; /* how many instances of A we've matched */
416 char *lastloc;/* where previous A matched (0-len detect) */
420 /* this first element must match u.yes */
421 struct regmatch_state *prev_yes_state;
422 struct regmatch_state *save_curlyx;
423 CHECKPOINT cp; /* remember current savestack indexes */
425 char *save_lastloc; /* previous curlyx.lastloc */
431 /* this first element must match u.yes */
432 struct regmatch_state *prev_yes_state;
433 I32 c1, c2; /* case fold search */
435 I32 alen; /* length of first-matched A string */
438 regnode *A, *B; /* the nodes corresponding to /A*B/ */
439 regnode *me; /* the curlym node */
445 I32 c1, c2; /* case fold search */
446 char *maxpos; /* highest possible point in string to match */
447 char *oldloc; /* the previous locinput */
449 int min, max; /* {m,n} */
450 regnode *A, *B; /* the nodes corresponding to /A*B/ */
451 } curly; /* and CURLYN/PLUS/STAR */
456 /* how many regmatch_state structs to allocate as a single slab.
457 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
458 * pointers, plus 1 for any mythical malloc overhead. */
460 #define PERL_REGMATCH_SLAB_SLOTS \
461 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
463 typedef struct regmatch_slab {
464 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
465 struct regmatch_slab *prev, *next;
468 #define PL_reg_flags PL_reg_state.re_state_reg_flags
469 #define PL_bostr PL_reg_state.re_state_bostr
470 #define PL_reginput PL_reg_state.re_state_reginput
471 #define PL_regeol PL_reg_state.re_state_regeol
472 #define PL_regstartp PL_reg_state.re_state_regstartp
473 #define PL_regendp PL_reg_state.re_state_regendp
474 #define PL_reglastparen PL_reg_state.re_state_reglastparen
475 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
476 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
477 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
478 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
479 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
480 #define PL_reg_magic PL_reg_state.re_state_reg_magic
481 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
482 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
483 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
484 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
485 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
486 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
487 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
488 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
489 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
490 #define PL_regsize PL_reg_state.re_state_regsize
491 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
492 #define PL_nrs PL_reg_state.re_state_nrs
494 struct re_save_state {
495 U32 re_state_reg_flags; /* from regexec.c */
496 char *re_state_bostr;
497 char *re_state_reginput; /* String-input pointer. */
498 char *re_state_regeol; /* End of input, for $ check. */
499 I32 *re_state_regstartp; /* Pointer to startp array. */
500 I32 *re_state_regendp; /* Ditto for endp. */
501 U32 *re_state_reglastparen; /* Similarly for lastparen. */
502 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
503 char **re_state_reg_start_tmp; /* from regexec.c */
504 U32 re_state_reg_start_tmpl; /* from regexec.c */
505 I32 re_state_reg_eval_set; /* from regexec.c */
506 bool re_state_reg_match_utf8; /* from regexec.c */
507 MAGIC *re_state_reg_magic; /* from regexec.c */
508 I32 re_state_reg_oldpos; /* from regexec.c */
509 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
510 PMOP *re_state_reg_curpm; /* from regexec.c */
511 char *re_state_reg_oldsaved; /* old saved substr during match */
512 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
513 I32 re_state_reg_maxiter; /* max wait until caching pos */
514 I32 re_state_reg_leftiter; /* wait until caching pos */
515 char *re_state_reg_poscache; /* cache of pos of WHILEM */
516 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
517 U32 re_state_regsize; /* from regexec.c */
518 char *re_state_reg_starttry; /* from regexec.c */
519 #ifdef PERL_OLD_COPY_ON_WRITE
520 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 */
524 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
525 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
529 * c-indentation-style: bsd
531 * indent-tabs-mode: t
534 * ex: set ts=8 sts=4 sw=4 noet: