1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
11 The GNU C Library is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 Lesser General Public License for more details.
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, write to the Free
18 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
21 #ifndef _REGEX_INTERNAL_H
22 #define _REGEX_INTERNAL_H 1
24 /* Number of bits in a byte. */
26 /* Number of single byte character. */
29 #define COLL_ELEM_LEN_MAX 8
31 /* The character which represents newline. */
32 #define NEWLINE_CHAR '\n'
33 #define WIDE_NEWLINE_CHAR L'\n'
35 /* Rename to standard API for using out of glibc. */
37 # define __wctype wctype
38 # define __iswctype iswctype
39 # define __btowc btowc
40 # define __mempcpy memcpy
41 # define attribute_hidden
42 #endif /* not _LIBC */
44 extern const char __re_error_msgid
[] attribute_hidden
;
45 extern const size_t __re_error_msgid_idx
[] attribute_hidden
;
47 /* Number of bits in an unsinged int. */
48 #define UINT_BITS (sizeof (unsigned int) * BYTE_BITS)
49 /* Number of unsigned int in an bit_set. */
50 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
51 typedef unsigned int bitset
[BITSET_UINTS
];
52 typedef unsigned int *re_bitset_ptr_t
;
54 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1 << i % UINT_BITS)
55 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1 << i % UINT_BITS))
56 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1 << i % UINT_BITS))
57 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
58 #define bitset_set_all(set) \
59 memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
60 #define bitset_copy(dest,src) \
61 memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
62 static inline void bitset_not (bitset set
);
63 static inline void bitset_merge (bitset dest
, const bitset src
);
64 static inline void bitset_not_merge (bitset dest
, const bitset src
);
66 #define PREV_WORD_CONSTRAINT 0x0001
67 #define PREV_NOTWORD_CONSTRAINT 0x0002
68 #define NEXT_WORD_CONSTRAINT 0x0004
69 #define NEXT_NOTWORD_CONSTRAINT 0x0008
70 #define PREV_NEWLINE_CONSTRAINT 0x0010
71 #define NEXT_NEWLINE_CONSTRAINT 0x0020
72 #define PREV_BEGBUF_CONSTRAINT 0x0040
73 #define NEXT_ENDBUF_CONSTRAINT 0x0080
74 #define DUMMY_CONSTRAINT 0x0100
78 INSIDE_WORD
= PREV_WORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
79 WORD_FIRST
= PREV_NOTWORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
80 WORD_LAST
= PREV_WORD_CONSTRAINT
| NEXT_NOTWORD_CONSTRAINT
,
81 LINE_FIRST
= PREV_NEWLINE_CONSTRAINT
,
82 LINE_LAST
= NEXT_NEWLINE_CONSTRAINT
,
83 BUF_FIRST
= PREV_BEGBUF_CONSTRAINT
,
84 BUF_LAST
= NEXT_ENDBUF_CONSTRAINT
,
85 WORD_DELIM
= DUMMY_CONSTRAINT
99 /* Token type, these are used only by token. */
109 OP_CLOSE_EQUIV_CLASS
,
116 /* Tree type, these are used only by tree. */
121 #ifdef RE_ENABLE_I18N
123 #endif /* RE_ENABLE_I18N */
125 /* Node type, These are used by token, node, tree. */
142 #ifdef RE_ENABLE_I18N
145 /* Multibyte characters. */
148 /* Collating symbols. */
153 /* Equivalence classes. */
155 int32_t *equiv_classes
;
158 /* Range expressions. */
160 uint32_t *range_starts
;
161 uint32_t *range_ends
;
162 # else /* not _LIBC */
163 wchar_t *range_starts
;
165 # endif /* not _LIBC */
167 /* Character classes. */
168 wctype_t *char_classes
;
170 /* If this character set is the non-matching list. */
171 unsigned int non_match
: 1;
173 /* # of multibyte characters. */
176 /* # of collating symbols. */
179 /* # of equivalence classes. */
182 /* # of range expressions. */
185 /* # of character classes. */
188 #endif /* RE_ENABLE_I18N */
194 unsigned char c
; /* for CHARACTER */
195 re_bitset_ptr_t sbcset
; /* for SIMPLE_BRACKET */
196 #ifdef RE_ENABLE_I18N
197 re_charset_t
*mbcset
; /* for COMPLEX_BRACKET */
198 #endif /* RE_ENABLE_I18N */
199 int idx
; /* for BACK_REF */
200 re_context_type ctx_type
; /* for ANCHOR */
203 re_token_type_t type
: 8;
205 re_token_type_t type
;
207 unsigned int constraint
: 10; /* context constraint */
208 unsigned int duplicated
: 1;
209 #ifdef RE_ENABLE_I18N
210 unsigned int mb_partial
: 1;
214 #define IS_EPSILON_NODE(type) \
215 ((type) == OP_ALT || (type) == OP_DUP_ASTERISK || (type) == OP_DUP_PLUS \
216 || (type) == OP_DUP_QUESTION || (type) == ANCHOR \
217 || (type) == OP_OPEN_SUBEXP || (type) == OP_CLOSE_SUBEXP)
219 #define ACCEPT_MB_NODE(type) \
220 ((type) == COMPLEX_BRACKET || (type) == OP_PERIOD)
224 /* Indicate the raw buffer which is the original string passed as an
225 argument of regexec(), re_search(), etc.. */
226 const unsigned char *raw_mbs
;
227 /* Store the multibyte string. In case of "case insensitive mode" like
228 REG_ICASE, upper cases of the string are stored, otherwise MBS points
229 the same address that RAW_MBS points. */
231 /* Store the case sensitive multibyte string. In case of
232 "case insensitive mode", the original string are stored,
233 otherwise MBS_CASE points the same address that MBS points. */
234 unsigned char *mbs_case
;
235 #ifdef RE_ENABLE_I18N
236 /* Store the wide character string which is corresponding to MBS. */
240 /* Index in RAW_MBS. Each character mbs[i] corresponds to
241 raw_mbs[raw_mbs_idx + i]. */
243 /* The length of the valid characters in the buffers. */
245 /* The length of the buffers MBS, MBS_CASE, and WCS. */
247 /* The index in MBS, which is updated by re_string_fetch_byte. */
249 /* This is length_of_RAW_MBS - RAW_MBS_IDX. */
251 /* End of the buffer may be shorter than its length in the cases such
252 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
256 /* The context of mbs[0]. We store the context independently, since
257 the context of mbs[0] may be different from raw_mbs[0], which is
258 the beginning of the input string. */
259 unsigned int tip_context
;
260 /* The translation passed as a part of an argument of re_compile_pattern. */
261 RE_TRANSLATE_TYPE trans
;
262 /* 1 if REG_ICASE. */
263 unsigned int icase
: 1;
265 typedef struct re_string_t re_string_t
;
266 /* In case of REG_ICASE, we allocate the buffer dynamically for mbs. */
267 #define MBS_ALLOCATED(pstr) (pstr->icase)
268 /* In case that we need translation, we allocate the buffer dynamically
269 for mbs_case. Note that mbs == mbs_case if not REG_ICASE. */
270 #define MBS_CASE_ALLOCATED(pstr) (pstr->trans != NULL)
273 static reg_errcode_t
re_string_allocate (re_string_t
*pstr
, const char *str
,
274 int len
, int init_len
,
275 RE_TRANSLATE_TYPE trans
, int icase
);
276 static reg_errcode_t
re_string_construct (re_string_t
*pstr
, const char *str
,
277 int len
, RE_TRANSLATE_TYPE trans
,
279 static reg_errcode_t
re_string_reconstruct (re_string_t
*pstr
, int idx
,
280 int eflags
, int newline
);
281 static reg_errcode_t
re_string_realloc_buffers (re_string_t
*pstr
,
283 #ifdef RE_ENABLE_I18N
284 static void build_wcs_buffer (re_string_t
*pstr
);
285 static void build_wcs_upper_buffer (re_string_t
*pstr
);
286 #endif /* RE_ENABLE_I18N */
287 static void build_upper_buffer (re_string_t
*pstr
);
288 static void re_string_translate_buffer (re_string_t
*pstr
);
289 static void re_string_destruct (re_string_t
*pstr
);
290 #ifdef RE_ENABLE_I18N
291 static int re_string_elem_size_at (const re_string_t
*pstr
, int idx
);
292 static inline int re_string_char_size_at (const re_string_t
*pstr
, int idx
);
293 static inline wint_t re_string_wchar_at (const re_string_t
*pstr
, int idx
);
294 #endif /* RE_ENABLE_I18N */
295 static unsigned int re_string_context_at (const re_string_t
*input
, int idx
,
296 int eflags
, int newline_anchor
);
297 #define re_string_peek_byte(pstr, offset) \
298 ((pstr)->mbs[(pstr)->cur_idx + offset])
299 #define re_string_peek_byte_case(pstr, offset) \
300 ((pstr)->mbs_case[(pstr)->cur_idx + offset])
301 #define re_string_fetch_byte(pstr) \
302 ((pstr)->mbs[(pstr)->cur_idx++])
303 #define re_string_fetch_byte_case(pstr) \
304 ((pstr)->mbs_case[(pstr)->cur_idx++])
305 #define re_string_first_byte(pstr, idx) \
306 ((idx) == (pstr)->len || (pstr)->wcs[idx] != WEOF)
307 #define re_string_is_single_byte_char(pstr, idx) \
308 ((pstr)->wcs[idx] != WEOF && ((pstr)->len == (idx) \
309 || (pstr)->wcs[(idx) + 1] != WEOF))
310 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
311 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
312 #define re_string_get_buffer(pstr) ((pstr)->mbs)
313 #define re_string_length(pstr) ((pstr)->len)
314 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
315 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
316 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
318 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
319 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
320 #define re_free(p) free (p)
324 struct bin_tree_t
*parent
;
325 struct bin_tree_t
*left
;
326 struct bin_tree_t
*right
;
328 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
329 Otherwise `type' indicate the type of this node. */
330 re_token_type_t type
;
335 re_node_set eclosure
;
337 typedef struct bin_tree_t bin_tree_t
;
340 #define CONTEXT_WORD 1
341 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
342 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
343 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
345 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
346 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
347 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
348 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
349 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
351 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
352 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
353 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
354 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
356 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
357 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
358 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
359 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
360 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
362 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
363 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
364 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
365 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
366 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
372 re_node_set
*entrance_nodes
;
373 struct re_dfastate_t
**trtable
;
374 struct re_dfastate_t
**trtable_search
;
375 /* If this state is a special state.
376 A state is a special state if the state is the halt state, or
378 unsigned int context
: 2;
379 unsigned int halt
: 1;
380 /* If this state can accept `multi byte'.
381 Note that we refer to multibyte characters, and multi character
382 collating elements as `multi byte'. */
383 unsigned int accept_mb
: 1;
384 /* If this state has backreference node(s). */
385 unsigned int has_backref
: 1;
386 unsigned int has_constraint
: 1;
388 typedef struct re_dfastate_t re_dfastate_t
;
392 /* start <= node < end */
397 struct re_state_table_entry
401 re_dfastate_t
**array
;
404 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
410 re_dfastate_t
**array
;
413 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
418 int str_idx
; /* The position NODE match at. */
420 } re_sub_match_last_t
;
422 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
423 And information about the node, whose type is OP_CLOSE_SUBEXP,
424 corresponding to NODE is stored in LASTS. */
430 int next_last_offset
;
432 int alasts
; /* Allocation size of LASTS. */
433 int nlasts
; /* The number of LASTS. */
434 re_sub_match_last_t
**lasts
;
435 } re_sub_match_top_t
;
437 struct re_backref_cache_entry
448 /* EFLAGS of the argument of regexec. */
450 /* Where the matching ends. */
453 /* The string object corresponding to the input string. */
455 /* The state log used by the matcher. */
456 re_dfastate_t
**state_log
;
458 /* Back reference cache. */
461 struct re_backref_cache_entry
*bkref_ents
;
465 re_sub_match_top_t
**sub_tops
;
466 } re_match_context_t
;
473 re_dfastate_t
**sifted_states
;
474 re_dfastate_t
**limited_states
;
483 struct re_fail_stack_ent_t
488 re_node_set eps_via_nodes
;
491 struct re_fail_stack_t
495 struct re_fail_stack_ent_t
*stack
;
500 re_bitset_ptr_t word_char
;
502 /* number of subexpressions `re_nsub' is in regex_t. */
504 re_subexp_t
*subexps
;
509 bin_tree_t
*str_tree
;
512 re_node_set
*eclosures
;
513 re_node_set
*inveclosures
;
514 struct re_state_table_entry
*state_table
;
515 unsigned int state_hash_mask
;
516 re_dfastate_t
*init_state
;
517 re_dfastate_t
*init_state_word
;
518 re_dfastate_t
*init_state_nl
;
519 re_dfastate_t
*init_state_begbuf
;
522 int nbackref
; /* The number of backreference in this dfa. */
523 /* Bitmap expressing which backreference is used. */
524 unsigned int used_bkref_map
;
528 unsigned int has_plural_match
: 1;
529 /* If this dfa has "multibyte node", which is a backreference or
530 a node which can accept multibyte character or multi character
531 collating element. */
532 unsigned int has_mb_node
: 1;
534 typedef struct re_dfa_t re_dfa_t
;
536 static reg_errcode_t
re_node_set_alloc (re_node_set
*set
, int size
);
537 static reg_errcode_t
re_node_set_init_1 (re_node_set
*set
, int elem
);
538 static reg_errcode_t
re_node_set_init_2 (re_node_set
*set
, int elem1
,
540 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
541 static reg_errcode_t
re_node_set_init_copy (re_node_set
*dest
,
542 const re_node_set
*src
);
543 static reg_errcode_t
re_node_set_add_intersect (re_node_set
*dest
,
544 const re_node_set
*src1
,
545 const re_node_set
*src2
);
546 static reg_errcode_t
re_node_set_init_union (re_node_set
*dest
,
547 const re_node_set
*src1
,
548 const re_node_set
*src2
);
549 static reg_errcode_t
re_node_set_merge (re_node_set
*dest
,
550 const re_node_set
*src
);
551 static int re_node_set_insert (re_node_set
*set
, int elem
);
552 static int re_node_set_compare (const re_node_set
*set1
,
553 const re_node_set
*set2
);
554 static int re_node_set_contains (const re_node_set
*set
, int elem
);
555 static void re_node_set_remove_at (re_node_set
*set
, int idx
);
556 #define re_node_set_remove(set,id) \
557 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
558 #define re_node_set_empty(p) ((p)->nelem = 0)
559 #define re_node_set_free(set) re_free ((set)->elems)
560 static int re_dfa_add_node (re_dfa_t
*dfa
, re_token_t token
, int mode
);
561 static re_dfastate_t
*re_acquire_state (reg_errcode_t
*err
, re_dfa_t
*dfa
,
562 const re_node_set
*nodes
);
563 static re_dfastate_t
*re_acquire_state_context (reg_errcode_t
*err
,
565 const re_node_set
*nodes
,
566 unsigned int context
);
567 static void free_state (re_dfastate_t
*state
);
581 bracket_elem_type type
;
591 /* Inline functions for bitset operation. */
597 for (bitset_i
= 0; bitset_i
< BITSET_UINTS
; ++bitset_i
)
598 set
[bitset_i
] = ~set
[bitset_i
];
602 bitset_merge (dest
, src
)
607 for (bitset_i
= 0; bitset_i
< BITSET_UINTS
; ++bitset_i
)
608 dest
[bitset_i
] |= src
[bitset_i
];
612 bitset_not_merge (dest
, src
)
617 for (i
= 0; i
< BITSET_UINTS
; ++i
)
621 #ifdef RE_ENABLE_I18N
622 /* Inline functions for re_string. */
624 re_string_char_size_at (pstr
, idx
)
625 const re_string_t
*pstr
;
631 for (byte_idx
= 1; idx
+ byte_idx
< pstr
->len
; ++byte_idx
)
632 if (pstr
->wcs
[idx
+ byte_idx
] != WEOF
)
638 re_string_wchar_at (pstr
, idx
)
639 const re_string_t
*pstr
;
643 return (wint_t) pstr
->mbs
[idx
];
644 return (wint_t) pstr
->wcs
[idx
];
648 re_string_elem_size_at (pstr
, idx
)
649 const re_string_t
*pstr
;
653 const unsigned char *p
, *extra
;
654 const int32_t *table
, *indirect
;
656 # include <locale/weight.h>
657 uint_fast32_t nrules
= _NL_CURRENT_WORD (LC_COLLATE
, _NL_COLLATE_NRULES
);
661 table
= (const int32_t *) _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_TABLEMB
);
662 extra
= (const unsigned char *)
663 _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_EXTRAMB
);
664 indirect
= (const int32_t *) _NL_CURRENT (LC_COLLATE
,
665 _NL_COLLATE_INDIRECTMB
);
668 return p
- pstr
->mbs
- idx
;
674 #endif /* RE_ENABLE_I18N */
676 #endif /* _REGEX_INTERNAL_H */