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 I32 seen_evals; /* number of eval groups in the pattern - for security checks */
86 HV *paren_names; /* Optional hash of paren names */
88 /* Refcount of this regexp */
89 I32 refcnt; /* Refcount of this regexp */
92 /* used for high speed searches */
93 typedef struct re_scream_pos_data_s
95 char **scream_olds; /* match pos */
96 I32 *scream_pos; /* Internal iterator of scream. */
99 /* regexp_engine structure. This is the dispatch table for regexes.
100 * Any regex engine implementation must be able to build one of these.
102 typedef struct regexp_engine {
103 regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm);
104 I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend,
105 char* strbeg, I32 minend, SV* screamer,
106 void* data, U32 flags);
107 char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos,
108 char *strend, U32 flags,
109 struct re_scream_pos_data_s *data);
110 SV* (*checkstr) (pTHX_ regexp *prog);
111 void (*free) (pTHX_ struct regexp* r);
113 void* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param);
117 /* Flags stored in regexp->extflags
118 * These are used by code external to the regexp engine
120 * Note that flags starting with RXf_PMf_ have exact equivalents
121 * stored in op_pmflags and which are defined in op.h, they are defined
122 * numerically here only for clarity.
125 /* Anchor and GPOS related stuff */
126 #define RXf_ANCH_BOL 0x00000001
127 #define RXf_ANCH_MBOL 0x00000002
128 #define RXf_ANCH_SBOL 0x00000004
129 #define RXf_ANCH_GPOS 0x00000008
130 #define RXf_GPOS_SEEN 0x00000010
131 #define RXf_GPOS_FLOAT 0x00000020
133 #define RXf_ANCH (RXf_ANCH_BOL|RXf_ANCH_MBOL|RXf_ANCH_GPOS|RXf_ANCH_SBOL)
134 #define RXf_GPOS_CHECK (RXf_GPOS_SEEN|RXf_ANCH_GPOS)
135 #define RXf_ANCH_SINGLE (RXf_ANCH_SBOL|RXf_ANCH_GPOS)
137 /* Flags indicating special patterns */
138 #define RXf_START_ONLY 0x00000200 /* Pattern is /^/ */
139 #define RXf_WHITE 0x00000400 /* Pattern is /\s+/ */
141 /* 0xF800 of extflags is used by (RXf_)PMf_COMPILETIME */
142 #define RXf_PMf_LOCALE 0x00000800 /* use locale */
143 #define RXf_PMf_MULTILINE 0x00001000 /* /m */
144 #define RXf_PMf_SINGLELINE 0x00002000 /* /s */
145 #define RXf_PMf_FOLD 0x00004000 /* /i */
146 #define RXf_PMf_EXTENDED 0x00008000 /* /x */
147 #define RXf_PMf_KEEPCOPY 0x00010000 /* /k */
148 /* these flags are transfered from the PMOP->op_pmflags member during compilation */
149 #define RXf_PMf_STD_PMMOD (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_FOLD|RXf_PMf_EXTENDED)
150 #define RXf_PMf_COMPILETIME (RXf_PMf_MULTILINE|RXf_PMf_SINGLELINE|RXf_PMf_LOCALE|RXf_PMf_FOLD|RXf_PMf_EXTENDED|RXf_PMf_KEEPCOPY)
152 #define CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl) \
153 case IGNORE_PAT_MOD: *(pmfl) |= RXf_PMf_FOLD; break; \
154 case MULTILINE_PAT_MOD: *(pmfl) |= RXf_PMf_MULTILINE; break; \
155 case SINGLE_PAT_MOD: *(pmfl) |= RXf_PMf_SINGLELINE; break; \
156 case XTENDED_PAT_MOD: *(pmfl) |= RXf_PMf_EXTENDED; break
158 /* What we have seen */
159 #define RXf_LOOKBEHIND_SEEN 0x00020000
160 #define RXf_EVAL_SEEN 0x00040000
161 #define RXf_CANY_SEEN 0x00080000
164 #define RXf_NOSCAN 0x00100000
165 #define RXf_CHECK_ALL 0x00200000
168 #define RXf_UTF8 0x00400000
169 #define RXf_MATCH_UTF8 0x00800000
172 #define RXf_USE_INTUIT_NOML 0x01000000
173 #define RXf_USE_INTUIT_ML 0x02000000
174 #define RXf_INTUIT_TAIL 0x04000000
176 #define RXf_USE_INTUIT (RXf_USE_INTUIT_NOML|RXf_USE_INTUIT_ML)
178 /* Copy and tainted info */
179 #define RXf_COPY_DONE 0x10000000
180 #define RXf_TAINTED_SEEN 0x20000000
184 #define RX_HAS_CUTGROUP(prog) ((prog)->intflags & PREGf_CUTGROUP_SEEN)
185 #define RX_MATCH_TAINTED(prog) ((prog)->extflags & RXf_TAINTED_SEEN)
186 #define RX_MATCH_TAINTED_on(prog) ((prog)->extflags |= RXf_TAINTED_SEEN)
187 #define RX_MATCH_TAINTED_off(prog) ((prog)->extflags &= ~RXf_TAINTED_SEEN)
188 #define RX_MATCH_TAINTED_set(prog, t) ((t) \
189 ? RX_MATCH_TAINTED_on(prog) \
190 : RX_MATCH_TAINTED_off(prog))
192 #define RX_MATCH_COPIED(prog) ((prog)->extflags & RXf_COPY_DONE)
193 #define RX_MATCH_COPIED_on(prog) ((prog)->extflags |= RXf_COPY_DONE)
194 #define RX_MATCH_COPIED_off(prog) ((prog)->extflags &= ~RXf_COPY_DONE)
195 #define RX_MATCH_COPIED_set(prog,t) ((t) \
196 ? RX_MATCH_COPIED_on(prog) \
197 : RX_MATCH_COPIED_off(prog))
199 #endif /* PLUGGABLE_RE_EXTENSION */
201 /* Stuff that needs to be included in the plugable extension goes below here */
203 #ifdef PERL_OLD_COPY_ON_WRITE
204 #define RX_MATCH_COPY_FREE(rx) \
205 STMT_START {if (rx->saved_copy) { \
206 SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \
208 if (RX_MATCH_COPIED(rx)) { \
209 Safefree(rx->subbeg); \
210 RX_MATCH_COPIED_off(rx); \
213 #define RX_MATCH_COPY_FREE(rx) \
214 STMT_START {if (RX_MATCH_COPIED(rx)) { \
215 Safefree(rx->subbeg); \
216 RX_MATCH_COPIED_off(rx); \
220 #define RX_MATCH_UTF8(prog) ((prog)->extflags & RXf_MATCH_UTF8)
221 #define RX_MATCH_UTF8_on(prog) ((prog)->extflags |= RXf_MATCH_UTF8)
222 #define RX_MATCH_UTF8_off(prog) ((prog)->extflags &= ~RXf_MATCH_UTF8)
223 #define RX_MATCH_UTF8_set(prog, t) ((t) \
224 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
225 : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
227 #define REXEC_COPY_STR 0x01 /* Need to copy the string. */
228 #define REXEC_CHECKED 0x02 /* check_substr already checked. */
229 #define REXEC_SCREAM 0x04 /* use scream table. */
230 #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
231 #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
233 #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
234 #define ReREFCNT_dec(re) CALLREGFREE(re)
236 #define FBMcf_TAIL_DOLLAR 1
237 #define FBMcf_TAIL_DOLLARM 2
238 #define FBMcf_TAIL_Z 4
239 #define FBMcf_TAIL_z 8
240 #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
242 #define FBMrf_MULTILINE 1
244 /* an accepting state/position*/
245 struct _reg_trie_accepted {
249 typedef struct _reg_trie_accepted reg_trie_accepted;
251 /* some basic information about the current match that is created by
252 * Perl_regexec_flags and then passed to regtry(), regmatch() etc */
264 /* structures for holding and saving the state maintained by regmatch() */
266 #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50
268 typedef I32 CHECKPOINT;
270 typedef struct regmatch_state {
271 int resume_state; /* where to jump to on return */
272 char *locinput; /* where to backtrack in string on failure */
276 /* this is a fake union member that matches the first element
277 * of each member that needs to store positive backtrack
280 struct regmatch_state *prev_yes_state;
284 /* this first element must match u.yes */
285 struct regmatch_state *prev_yes_state;
286 reg_trie_accepted *accept_buff;
287 U32 accepted; /* how many accepting states we have seen */
288 U16 *jump; /* positive offsets from me */
289 regnode *B; /* node following the trie */
290 regnode *me; /* Which node am I - needed for jump tries*/
294 /* this first element must match u.yes */
295 struct regmatch_state *prev_yes_state;
296 struct regmatch_state *prev_eval;
297 struct regmatch_state *prev_curlyx;
299 U32 toggle_reg_flags; /* what bits in PL_reg_flags to
300 flip when transitioning between
301 inner and outer rexen */
302 CHECKPOINT cp; /* remember current savestack indexes */
304 regnode *B; /* the node following us */
305 U32 close_paren; /* which close bracket is our end */
309 /* this first element must match u.yes */
310 struct regmatch_state *prev_yes_state;
311 struct regmatch_state *prev_curlyx; /* previous cur_curlyx */
312 CHECKPOINT cp; /* remember current savestack index */
314 int parenfloor;/* how far back to strip paren data */
315 int min; /* the minimal number of A's to match */
316 int max; /* the maximal number of A's to match */
317 regnode *A, *B; /* the nodes corresponding to /A*B/ */
319 /* these two are modified by WHILEM */
320 int count; /* how many instances of A we've matched */
321 char *lastloc;/* where previous A matched (0-len detect) */
325 /* this first element must match u.yes */
326 struct regmatch_state *prev_yes_state;
327 struct regmatch_state *save_curlyx;
328 CHECKPOINT cp; /* remember current savestack indexes */
330 char *save_lastloc; /* previous curlyx.lastloc */
336 /* this first element must match u.yes */
337 struct regmatch_state *prev_yes_state;
339 regnode *next_branch; /* next branch node */
344 /* this first element must match u.yes */
345 struct regmatch_state *prev_yes_state;
346 I32 c1, c2; /* case fold search */
348 I32 alen; /* length of first-matched A string */
351 regnode *A, *B; /* the nodes corresponding to /A*B/ */
352 regnode *me; /* the curlym node */
358 I32 c1, c2; /* case fold search */
359 char *maxpos; /* highest possible point in string to match */
360 char *oldloc; /* the previous locinput */
362 int min, max; /* {m,n} */
363 regnode *A, *B; /* the nodes corresponding to /A*B/ */
364 } curly; /* and CURLYN/PLUS/STAR */
367 /* this first element must match u.yes */
368 struct regmatch_state *prev_yes_state;
370 I32 logical; /* saved copy of 'logical' var */
371 regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */
372 } ifmatch; /* and SUSPEND/UNLESSM */
375 /* this first element must match u.yes */
376 struct regmatch_state *prev_yes_state;
377 struct regmatch_state *prev_mark;
388 /* how many regmatch_state structs to allocate as a single slab.
389 * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev
390 * pointers, plus 1 for any mythical malloc overhead. */
392 #define PERL_REGMATCH_SLAB_SLOTS \
393 ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state))
395 typedef struct regmatch_slab {
396 regmatch_state states[PERL_REGMATCH_SLAB_SLOTS];
397 struct regmatch_slab *prev, *next;
400 #define PL_reg_flags PL_reg_state.re_state_reg_flags
401 #define PL_bostr PL_reg_state.re_state_bostr
402 #define PL_reginput PL_reg_state.re_state_reginput
403 #define PL_regeol PL_reg_state.re_state_regeol
404 #define PL_regstartp PL_reg_state.re_state_regstartp
405 #define PL_regendp PL_reg_state.re_state_regendp
406 #define PL_reglastparen PL_reg_state.re_state_reglastparen
407 #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen
408 #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp
409 #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl
410 #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set
411 #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8
412 #define PL_reg_magic PL_reg_state.re_state_reg_magic
413 #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos
414 #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm
415 #define PL_reg_curpm PL_reg_state.re_state_reg_curpm
416 #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved
417 #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen
418 #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter
419 #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter
420 #define PL_reg_poscache PL_reg_state.re_state_reg_poscache
421 #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size
422 #define PL_regsize PL_reg_state.re_state_regsize
423 #define PL_reg_starttry PL_reg_state.re_state_reg_starttry
424 #define PL_nrs PL_reg_state.re_state_nrs
426 struct re_save_state {
427 U32 re_state_reg_flags; /* from regexec.c */
428 char *re_state_bostr;
429 char *re_state_reginput; /* String-input pointer. */
430 char *re_state_regeol; /* End of input, for $ check. */
431 I32 *re_state_regstartp; /* Pointer to startp array. */
432 I32 *re_state_regendp; /* Ditto for endp. */
433 U32 *re_state_reglastparen; /* Similarly for lastparen. */
434 U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */
435 char **re_state_reg_start_tmp; /* from regexec.c */
436 U32 re_state_reg_start_tmpl; /* from regexec.c */
437 I32 re_state_reg_eval_set; /* from regexec.c */
438 bool re_state_reg_match_utf8; /* from regexec.c */
439 MAGIC *re_state_reg_magic; /* from regexec.c */
440 I32 re_state_reg_oldpos; /* from regexec.c */
441 PMOP *re_state_reg_oldcurpm; /* from regexec.c */
442 PMOP *re_state_reg_curpm; /* from regexec.c */
443 char *re_state_reg_oldsaved; /* old saved substr during match */
444 STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */
445 I32 re_state_reg_maxiter; /* max wait until caching pos */
446 I32 re_state_reg_leftiter; /* wait until caching pos */
447 char *re_state_reg_poscache; /* cache of pos of WHILEM */
448 STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */
449 U32 re_state_regsize; /* from regexec.c */
450 char *re_state_reg_starttry; /* from regexec.c */
451 #ifdef PERL_OLD_COPY_ON_WRITE
452 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 */
456 #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \
457 (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack)))
461 * c-indentation-style: bsd
463 * indent-tabs-mode: t
466 * ex: set ts=8 sts=4 sw=4 noet: