1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003 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
35 #if defined HAVE_LOCALE_H || defined _LIBC
38 #if defined HAVE_WCHAR_H || defined _LIBC
40 #endif /* HAVE_WCHAR_H || _LIBC */
41 #if defined HAVE_WCTYPE_H || defined _LIBC
43 #endif /* HAVE_WCTYPE_H || _LIBC */
45 /* In case that the system doesn't have isblank(). */
46 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
47 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
51 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
52 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
53 # include <locale/localeinfo.h>
54 # include <locale/elem-hash.h>
55 # include <locale/coll-lookup.h>
59 /* This is for other GNU distributions with internationalized messages. */
60 #if HAVE_LIBINTL_H || defined _LIBC
64 # define gettext(msgid) \
65 INTUSE(__dcgettext) (INTUSE(_libc_intl_domainname), msgid, LC_MESSAGES)
68 # define gettext(msgid) (msgid)
72 /* This define is so xgettext can find the internationalizable
74 # define gettext_noop(String) String
77 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
78 # define RE_ENABLE_I18N
82 # define BE(expr, val) __builtin_expect (expr, val)
84 # define BE(expr, val) (expr)
88 /* Number of bits in a byte. */
90 /* Number of single byte character. */
93 #define COLL_ELEM_LEN_MAX 8
95 /* The character which represents newline. */
96 #define NEWLINE_CHAR '\n'
97 #define WIDE_NEWLINE_CHAR L'\n'
99 /* Rename to standard API for using out of glibc. */
101 # define __wctype wctype
102 # define __iswctype iswctype
103 # define __btowc btowc
104 # define __mempcpy mempcpy
105 # define __wcrtomb wcrtomb
106 # define attribute_hidden
107 #endif /* not _LIBC */
110 # define __attribute __attribute__
115 extern const char __re_error_msgid
[] attribute_hidden
;
116 extern const size_t __re_error_msgid_idx
[] attribute_hidden
;
118 /* Number of bits in an unsinged int. */
119 #define UINT_BITS (sizeof (unsigned int) * BYTE_BITS)
120 /* Number of unsigned int in an bit_set. */
121 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
122 typedef unsigned int bitset
[BITSET_UINTS
];
123 typedef unsigned int *re_bitset_ptr_t
;
125 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1 << i % UINT_BITS)
126 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1 << i % UINT_BITS))
127 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1 << i % UINT_BITS))
128 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
129 #define bitset_set_all(set) \
130 memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
131 #define bitset_copy(dest,src) \
132 memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
133 static inline void bitset_not (bitset set
);
134 static inline void bitset_merge (bitset dest
, const bitset src
);
135 static inline void bitset_not_merge (bitset dest
, const bitset src
);
137 #define PREV_WORD_CONSTRAINT 0x0001
138 #define PREV_NOTWORD_CONSTRAINT 0x0002
139 #define NEXT_WORD_CONSTRAINT 0x0004
140 #define NEXT_NOTWORD_CONSTRAINT 0x0008
141 #define PREV_NEWLINE_CONSTRAINT 0x0010
142 #define NEXT_NEWLINE_CONSTRAINT 0x0020
143 #define PREV_BEGBUF_CONSTRAINT 0x0040
144 #define NEXT_ENDBUF_CONSTRAINT 0x0080
145 #define DUMMY_CONSTRAINT 0x0100
149 INSIDE_WORD
= PREV_WORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
150 WORD_FIRST
= PREV_NOTWORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
151 WORD_LAST
= PREV_WORD_CONSTRAINT
| NEXT_NOTWORD_CONSTRAINT
,
152 LINE_FIRST
= PREV_NEWLINE_CONSTRAINT
,
153 LINE_LAST
= NEXT_NEWLINE_CONSTRAINT
,
154 BUF_FIRST
= PREV_BEGBUF_CONSTRAINT
,
155 BUF_LAST
= NEXT_ENDBUF_CONSTRAINT
,
156 WORD_DELIM
= DUMMY_CONSTRAINT
170 /* Token type, these are used only by token. */
180 OP_CLOSE_EQUIV_CLASS
,
189 /* Tree type, these are used only by tree. */
194 #ifdef RE_ENABLE_I18N
196 #endif /* RE_ENABLE_I18N */
198 /* Node type, These are used by token, node, tree. */
215 #ifdef RE_ENABLE_I18N
218 /* Multibyte characters. */
221 /* Collating symbols. */
226 /* Equivalence classes. */
228 int32_t *equiv_classes
;
231 /* Range expressions. */
233 uint32_t *range_starts
;
234 uint32_t *range_ends
;
235 # else /* not _LIBC */
236 wchar_t *range_starts
;
238 # endif /* not _LIBC */
240 /* Character classes. */
241 wctype_t *char_classes
;
243 /* If this character set is the non-matching list. */
244 unsigned int non_match
: 1;
246 /* # of multibyte characters. */
249 /* # of collating symbols. */
252 /* # of equivalence classes. */
255 /* # of range expressions. */
258 /* # of character classes. */
261 #endif /* RE_ENABLE_I18N */
267 unsigned char c
; /* for CHARACTER */
268 re_bitset_ptr_t sbcset
; /* for SIMPLE_BRACKET */
269 #ifdef RE_ENABLE_I18N
270 re_charset_t
*mbcset
; /* for COMPLEX_BRACKET */
271 #endif /* RE_ENABLE_I18N */
272 int idx
; /* for BACK_REF */
273 re_context_type ctx_type
; /* for ANCHOR */
276 re_token_type_t type
: 8;
278 re_token_type_t type
;
280 unsigned int constraint
: 10; /* context constraint */
281 unsigned int duplicated
: 1;
282 #ifdef RE_ENABLE_I18N
283 unsigned int mb_partial
: 1;
287 #define IS_EPSILON_NODE(type) \
288 ((type) == OP_ALT || (type) == OP_DUP_ASTERISK || (type) == OP_DUP_PLUS \
289 || (type) == OP_DUP_QUESTION || (type) == ANCHOR \
290 || (type) == OP_OPEN_SUBEXP || (type) == OP_CLOSE_SUBEXP)
292 #define ACCEPT_MB_NODE(type) \
293 ((type) == COMPLEX_BRACKET || (type) == OP_PERIOD)
297 /* Indicate the raw buffer which is the original string passed as an
298 argument of regexec(), re_search(), etc.. */
299 const unsigned char *raw_mbs
;
300 /* Store the multibyte string. In case of "case insensitive mode" like
301 REG_ICASE, upper cases of the string are stored, otherwise MBS points
302 the same address that RAW_MBS points. */
304 /* Store the case sensitive multibyte string. In case of
305 "case insensitive mode", the original string are stored,
306 otherwise MBS_CASE points the same address that MBS points. */
307 unsigned char *mbs_case
;
308 #ifdef RE_ENABLE_I18N
309 /* Store the wide character string which is corresponding to MBS. */
313 /* Index in RAW_MBS. Each character mbs[i] corresponds to
314 raw_mbs[raw_mbs_idx + i]. */
316 /* The length of the valid characters in the buffers. */
318 /* The length of the buffers MBS, MBS_CASE, and WCS. */
320 /* The index in MBS, which is updated by re_string_fetch_byte. */
322 /* This is length_of_RAW_MBS - RAW_MBS_IDX. */
324 /* End of the buffer may be shorter than its length in the cases such
325 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
329 /* The context of mbs[0]. We store the context independently, since
330 the context of mbs[0] may be different from raw_mbs[0], which is
331 the beginning of the input string. */
332 unsigned int tip_context
;
333 /* The translation passed as a part of an argument of re_compile_pattern. */
334 RE_TRANSLATE_TYPE trans
;
335 /* 1 if REG_ICASE. */
336 unsigned int icase
: 1;
337 unsigned int is_utf8
: 1;
338 unsigned int map_notascii
: 1;
341 typedef struct re_string_t re_string_t
;
342 /* In case of REG_ICASE, we allocate the buffer dynamically for mbs. */
343 #define MBS_ALLOCATED(pstr) (pstr->icase)
344 /* In case that we need translation, we allocate the buffer dynamically
345 for mbs_case. Note that mbs == mbs_case if not REG_ICASE. */
346 #define MBS_CASE_ALLOCATED(pstr) (pstr->trans != NULL)
350 typedef struct re_dfa_t re_dfa_t
;
351 #ifndef RE_NO_INTERNAL_PROTOTYPES
352 static reg_errcode_t
re_string_allocate (re_string_t
*pstr
, const char *str
,
353 int len
, int init_len
,
354 RE_TRANSLATE_TYPE trans
, int icase
,
355 const re_dfa_t
*dfa
);
356 static reg_errcode_t
re_string_construct (re_string_t
*pstr
, const char *str
,
357 int len
, RE_TRANSLATE_TYPE trans
,
358 int icase
, const re_dfa_t
*dfa
);
359 static reg_errcode_t
re_string_reconstruct (re_string_t
*pstr
, int idx
,
360 int eflags
, int newline
);
361 static reg_errcode_t
re_string_realloc_buffers (re_string_t
*pstr
,
363 # ifdef RE_ENABLE_I18N
364 static void build_wcs_buffer (re_string_t
*pstr
);
365 static void build_wcs_upper_buffer (re_string_t
*pstr
);
366 # endif /* RE_ENABLE_I18N */
367 static void build_upper_buffer (re_string_t
*pstr
);
368 static void re_string_translate_buffer (re_string_t
*pstr
);
369 static void re_string_destruct (re_string_t
*pstr
);
370 # ifdef RE_ENABLE_I18N
371 static int re_string_elem_size_at (const re_string_t
*pstr
, int idx
);
372 static inline int re_string_char_size_at (const re_string_t
*pstr
, int idx
);
373 static inline wint_t re_string_wchar_at (const re_string_t
*pstr
, int idx
);
374 # endif /* RE_ENABLE_I18N */
375 static unsigned int re_string_context_at (const re_string_t
*input
, int idx
,
376 int eflags
, int newline_anchor
);
378 #define re_string_peek_byte(pstr, offset) \
379 ((pstr)->mbs[(pstr)->cur_idx + offset])
380 #define re_string_peek_byte_case(pstr, offset) \
381 ((pstr)->mbs_case[(pstr)->cur_idx + offset])
382 #define re_string_fetch_byte(pstr) \
383 ((pstr)->mbs[(pstr)->cur_idx++])
384 #define re_string_fetch_byte_case(pstr) \
385 ((pstr)->mbs_case[(pstr)->cur_idx++])
386 #define re_string_first_byte(pstr, idx) \
387 ((idx) == (pstr)->len || (pstr)->wcs[idx] != WEOF)
388 #define re_string_is_single_byte_char(pstr, idx) \
389 ((pstr)->wcs[idx] != WEOF && ((pstr)->len == (idx) \
390 || (pstr)->wcs[(idx) + 1] != WEOF))
391 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
392 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
393 #define re_string_get_buffer(pstr) ((pstr)->mbs)
394 #define re_string_length(pstr) ((pstr)->len)
395 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
396 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
397 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
399 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
400 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
401 #define re_free(p) free (p)
405 struct bin_tree_t
*parent
;
406 struct bin_tree_t
*left
;
407 struct bin_tree_t
*right
;
409 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
410 Otherwise `type' indicate the type of this node. */
411 re_token_type_t type
;
416 re_node_set eclosure
;
418 typedef struct bin_tree_t bin_tree_t
;
421 #define CONTEXT_WORD 1
422 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
423 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
424 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
426 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
427 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
428 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
429 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
430 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
432 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
433 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
434 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
435 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
437 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
438 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
439 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
440 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
441 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
443 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
444 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
445 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
446 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
447 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
453 re_node_set
*entrance_nodes
;
454 struct re_dfastate_t
**trtable
;
455 struct re_dfastate_t
**trtable_search
;
456 /* If this state is a special state.
457 A state is a special state if the state is the halt state, or
459 unsigned int context
: 2;
460 unsigned int halt
: 1;
461 /* If this state can accept `multi byte'.
462 Note that we refer to multibyte characters, and multi character
463 collating elements as `multi byte'. */
464 unsigned int accept_mb
: 1;
465 /* If this state has backreference node(s). */
466 unsigned int has_backref
: 1;
467 unsigned int has_constraint
: 1;
469 typedef struct re_dfastate_t re_dfastate_t
;
473 /* start <= node < end */
478 struct re_state_table_entry
482 re_dfastate_t
**array
;
485 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
491 re_dfastate_t
**array
;
494 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
499 int str_idx
; /* The position NODE match at. */
501 } re_sub_match_last_t
;
503 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
504 And information about the node, whose type is OP_CLOSE_SUBEXP,
505 corresponding to NODE is stored in LASTS. */
511 int next_last_offset
;
513 int alasts
; /* Allocation size of LASTS. */
514 int nlasts
; /* The number of LASTS. */
515 re_sub_match_last_t
**lasts
;
516 } re_sub_match_top_t
;
518 struct re_backref_cache_entry
529 /* EFLAGS of the argument of regexec. */
531 /* Where the matching ends. */
534 /* The string object corresponding to the input string. */
536 /* The state log used by the matcher. */
537 re_dfastate_t
**state_log
;
539 /* Back reference cache. */
542 struct re_backref_cache_entry
*bkref_ents
;
546 re_sub_match_top_t
**sub_tops
;
547 } re_match_context_t
;
554 re_dfastate_t
**sifted_states
;
555 re_dfastate_t
**limited_states
;
564 struct re_fail_stack_ent_t
569 re_node_set eps_via_nodes
;
572 struct re_fail_stack_t
576 struct re_fail_stack_ent_t
*stack
;
581 re_bitset_ptr_t word_char
;
582 re_subexp_t
*subexps
;
586 bin_tree_t
*str_tree
;
590 re_node_set
*eclosures
;
591 re_node_set
*inveclosures
;
592 struct re_state_table_entry
*state_table
;
593 re_dfastate_t
*init_state
;
594 re_dfastate_t
*init_state_word
;
595 re_dfastate_t
*init_state_nl
;
596 re_dfastate_t
*init_state_begbuf
;
598 /* number of subexpressions `re_nsub' is in regex_t. */
600 unsigned int state_hash_mask
;
603 int nbackref
; /* The number of backreference in this dfa. */
604 /* Bitmap expressing which backreference is used. */
605 unsigned int used_bkref_map
;
609 unsigned int has_plural_match
: 1;
610 /* If this dfa has "multibyte node", which is a backreference or
611 a node which can accept multibyte character or multi character
612 collating element. */
613 unsigned int has_mb_node
: 1;
614 unsigned int is_utf8
: 1;
615 unsigned int map_notascii
: 1;
619 #ifndef RE_NO_INTERNAL_PROTOTYPES
620 static reg_errcode_t
re_node_set_alloc (re_node_set
*set
, int size
);
621 static reg_errcode_t
re_node_set_init_1 (re_node_set
*set
, int elem
);
622 static reg_errcode_t
re_node_set_init_2 (re_node_set
*set
, int elem1
,
624 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
625 static reg_errcode_t
re_node_set_init_copy (re_node_set
*dest
,
626 const re_node_set
*src
);
627 static reg_errcode_t
re_node_set_add_intersect (re_node_set
*dest
,
628 const re_node_set
*src1
,
629 const re_node_set
*src2
);
630 static reg_errcode_t
re_node_set_init_union (re_node_set
*dest
,
631 const re_node_set
*src1
,
632 const re_node_set
*src2
);
633 static reg_errcode_t
re_node_set_merge (re_node_set
*dest
,
634 const re_node_set
*src
);
635 static int re_node_set_insert (re_node_set
*set
, int elem
);
636 static int re_node_set_compare (const re_node_set
*set1
,
637 const re_node_set
*set2
);
638 static int re_node_set_contains (const re_node_set
*set
, int elem
);
639 static void re_node_set_remove_at (re_node_set
*set
, int idx
);
640 #define re_node_set_remove(set,id) \
641 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
642 #define re_node_set_empty(p) ((p)->nelem = 0)
643 #define re_node_set_free(set) re_free ((set)->elems)
644 static int re_dfa_add_node (re_dfa_t
*dfa
, re_token_t token
, int mode
);
645 static re_dfastate_t
*re_acquire_state (reg_errcode_t
*err
, re_dfa_t
*dfa
,
646 const re_node_set
*nodes
);
647 static re_dfastate_t
*re_acquire_state_context (reg_errcode_t
*err
,
649 const re_node_set
*nodes
,
650 unsigned int context
);
651 static void free_state (re_dfastate_t
*state
);
666 bracket_elem_type type
;
676 /* Inline functions for bitset operation. */
682 for (bitset_i
= 0; bitset_i
< BITSET_UINTS
; ++bitset_i
)
683 set
[bitset_i
] = ~set
[bitset_i
];
687 bitset_merge (dest
, src
)
692 for (bitset_i
= 0; bitset_i
< BITSET_UINTS
; ++bitset_i
)
693 dest
[bitset_i
] |= src
[bitset_i
];
697 bitset_not_merge (dest
, src
)
702 for (i
= 0; i
< BITSET_UINTS
; ++i
)
706 #if defined RE_ENABLE_I18N && !defined RE_NO_INTERNAL_PROTOTYPES
707 /* Inline functions for re_string. */
709 re_string_char_size_at (pstr
, idx
)
710 const re_string_t
*pstr
;
714 if (pstr
->mb_cur_max
== 1)
716 for (byte_idx
= 1; idx
+ byte_idx
< pstr
->valid_len
; ++byte_idx
)
717 if (pstr
->wcs
[idx
+ byte_idx
] != WEOF
)
723 re_string_wchar_at (pstr
, idx
)
724 const re_string_t
*pstr
;
727 if (pstr
->mb_cur_max
== 1)
728 return (wint_t) pstr
->mbs
[idx
];
729 return (wint_t) pstr
->wcs
[idx
];
733 re_string_elem_size_at (pstr
, idx
)
734 const re_string_t
*pstr
;
738 const unsigned char *p
, *extra
;
739 const int32_t *table
, *indirect
;
741 # include <locale/weight.h>
742 uint_fast32_t nrules
= _NL_CURRENT_WORD (LC_COLLATE
, _NL_COLLATE_NRULES
);
746 table
= (const int32_t *) _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_TABLEMB
);
747 extra
= (const unsigned char *)
748 _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_EXTRAMB
);
749 indirect
= (const int32_t *) _NL_CURRENT (LC_COLLATE
,
750 _NL_COLLATE_INDIRECTMB
);
753 return p
- pstr
->mbs
- idx
;
759 #endif /* RE_ENABLE_I18N */
761 #endif /* _REGEX_INTERNAL_H */