1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2, or (at your option)
11 This program 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
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License along
17 with this program; if not, write to the Free Software Foundation,
18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
29 #if defined HAVE_LANGINFO_H || defined HAVE_LANGINFO_CODESET || defined _LIBC
30 # include <langinfo.h>
32 #if defined HAVE_LOCALE_H || defined _LIBC
35 #if defined HAVE_WCHAR_H || defined _LIBC
37 #endif /* HAVE_WCHAR_H || _LIBC */
38 #if defined HAVE_WCTYPE_H || defined _LIBC
40 #endif /* HAVE_WCTYPE_H || _LIBC */
42 # include <bits/libc-lock.h>
44 # define __libc_lock_define(CLASS,NAME)
45 # define __libc_lock_init(NAME) do { } while (0)
46 # define __libc_lock_lock(NAME) do { } while (0)
47 # define __libc_lock_unlock(NAME) do { } while (0)
50 /* In case that the system doesn't have isblank(). */
51 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
52 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
56 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
57 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
58 # include <locale/localeinfo.h>
59 # include <locale/elem-hash.h>
60 # include <locale/coll-lookup.h>
64 /* This is for other GNU distributions with internationalized messages. */
65 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
69 # define gettext(msgid) \
70 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
73 # define gettext(msgid) (msgid)
77 /* This define is so xgettext can find the internationalizable
79 # define gettext_noop(String) String
82 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
83 # define RE_ENABLE_I18N
87 # define BE(expr, val) __builtin_expect (expr, val)
89 # define BE(expr, val) (expr)
93 /* Number of single byte character. */
96 #define COLL_ELEM_LEN_MAX 8
98 /* The character which represents newline. */
99 #define NEWLINE_CHAR '\n'
100 #define WIDE_NEWLINE_CHAR L'\n'
102 /* Rename to standard API for using out of glibc. */
104 # define __wctype wctype
105 # define __iswctype iswctype
106 # define __btowc btowc
108 # define __mempcpy mempcpy
110 # define __wcrtomb wcrtomb
111 # define __regfree regfree
112 # define attribute_hidden
113 #endif /* not _LIBC */
116 # define __attribute(arg) __attribute__ (arg)
118 # define __attribute(arg)
121 extern const char __re_error_msgid[] attribute_hidden;
122 extern const size_t __re_error_msgid_idx[] attribute_hidden;
124 typedef __re_idx_t Idx;
126 /* Special return value for failure to match. */
127 #define REG_MISSING ((Idx) -1)
129 /* Special return value for internal error. */
130 #define REG_ERROR ((Idx) -2)
132 /* Test whether N is a valid index, and is not one of the above. */
133 #ifdef _REGEX_LARGE_OFFSETS
134 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
136 # define REG_VALID_INDEX(n) (0 <= (n))
139 /* Test whether N is a valid nonzero index. */
140 #ifdef _REGEX_LARGE_OFFSETS
141 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
143 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
146 /* A hash value, suitable for computing hash tables. */
147 typedef __re_size_t re_hashval_t;
149 /* Number of bits in an unsinged int. */
150 #define UINT_BITS (sizeof (unsigned int) * CHAR_BIT)
151 /* Number of unsigned int in an bit_set. */
152 #define BITSET_UINTS ((SBC_MAX + UINT_BITS - 1) / UINT_BITS)
153 typedef unsigned int bitset[BITSET_UINTS];
154 typedef unsigned int *re_bitset_ptr_t;
155 typedef const unsigned int *re_const_bitset_ptr_t;
157 #define bitset_set(set,i) (set[i / UINT_BITS] |= 1u << i % UINT_BITS)
158 #define bitset_clear(set,i) (set[i / UINT_BITS] &= ~(1u << i % UINT_BITS))
159 #define bitset_contain(set,i) (set[i / UINT_BITS] & (1u << i % UINT_BITS))
160 #define bitset_empty(set) memset (set, 0, sizeof (unsigned int) * BITSET_UINTS)
161 #define bitset_set_all(set) \
162 memset (set, 255, sizeof (unsigned int) * BITSET_UINTS)
163 #define bitset_copy(dest,src) \
164 memcpy (dest, src, sizeof (unsigned int) * BITSET_UINTS)
166 #define PREV_WORD_CONSTRAINT 0x0001
167 #define PREV_NOTWORD_CONSTRAINT 0x0002
168 #define NEXT_WORD_CONSTRAINT 0x0004
169 #define NEXT_NOTWORD_CONSTRAINT 0x0008
170 #define PREV_NEWLINE_CONSTRAINT 0x0010
171 #define NEXT_NEWLINE_CONSTRAINT 0x0020
172 #define PREV_BEGBUF_CONSTRAINT 0x0040
173 #define NEXT_ENDBUF_CONSTRAINT 0x0080
174 #define WORD_DELIM_CONSTRAINT 0x0100
175 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
179 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
180 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
181 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
182 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
183 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
184 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
185 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
186 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
187 WORD_DELIM = WORD_DELIM_CONSTRAINT,
188 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
202 /* Node type, These are used by token, node, tree. */
208 #ifdef RE_ENABLE_I18N
211 #endif /* RE_ENABLE_I18N */
213 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
214 when the debugger shows values of this enum type. */
215 #define EPSILON_BIT 8
216 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
217 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
218 OP_ALT = EPSILON_BIT | 2,
219 OP_DUP_ASTERISK = EPSILON_BIT | 3,
220 ANCHOR = EPSILON_BIT | 4,
222 /* Tree type, these are used only by tree. */
226 /* Token type, these are used only by token. */
238 OP_CLOSE_EQUIV_CLASS,
249 #ifdef RE_ENABLE_I18N
252 /* Multibyte characters. */
255 /* Collating symbols. */
260 /* Equivalence classes. */
262 int32_t *equiv_classes;
265 /* Range expressions. */
267 uint32_t *range_starts;
268 uint32_t *range_ends;
269 # else /* not _LIBC */
270 wchar_t *range_starts;
272 # endif /* not _LIBC */
274 /* Character classes. */
275 wctype_t *char_classes;
277 /* If this character set is the non-matching list. */
278 unsigned int non_match : 1;
280 /* # of multibyte characters. */
283 /* # of collating symbols. */
286 /* # of equivalence classes. */
289 /* # of range expressions. */
292 /* # of character classes. */
295 #endif /* RE_ENABLE_I18N */
301 unsigned char c; /* for CHARACTER */
302 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
303 #ifdef RE_ENABLE_I18N
304 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
305 #endif /* RE_ENABLE_I18N */
306 Idx idx; /* for BACK_REF */
307 re_context_type ctx_type; /* for ANCHOR */
310 re_token_type_t type : 8;
312 re_token_type_t type;
314 unsigned int constraint : 10; /* context constraint */
315 unsigned int duplicated : 1;
316 unsigned int opt_subexp : 1;
317 #ifdef RE_ENABLE_I18N
318 unsigned int accept_mb : 1;
319 /* These 2 bits can be moved into the union if needed (e.g. if running out
320 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
321 unsigned int mb_partial : 1;
323 unsigned int word_char : 1;
326 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
330 /* Indicate the raw buffer which is the original string passed as an
331 argument of regexec(), re_search(), etc.. */
332 const unsigned char *raw_mbs;
333 /* Store the multibyte string. In case of "case insensitive mode" like
334 REG_ICASE, upper cases of the string are stored, otherwise MBS points
335 the same address that RAW_MBS points. */
337 #ifdef RE_ENABLE_I18N
338 /* Store the wide character string which is corresponding to MBS. */
343 /* Index in RAW_MBS. Each character mbs[i] corresponds to
344 raw_mbs[raw_mbs_idx + i]. */
346 /* The length of the valid characters in the buffers. */
348 /* The corresponding number of bytes in raw_mbs array. */
350 /* The length of the buffers MBS and WCS. */
352 /* The index in MBS, which is updated by re_string_fetch_byte. */
354 /* length of RAW_MBS array. */
356 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
358 /* End of the buffer may be shorter than its length in the cases such
359 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
362 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
365 /* The context of mbs[0]. We store the context independently, since
366 the context of mbs[0] may be different from raw_mbs[0], which is
367 the beginning of the input string. */
368 unsigned int tip_context;
369 /* The translation passed as a part of an argument of re_compile_pattern. */
370 unsigned REG_TRANSLATE_TYPE trans;
371 /* Copy of re_dfa_t's word_char. */
372 re_const_bitset_ptr_t word_char;
373 /* 1 if REG_ICASE. */
375 unsigned char is_utf8;
376 unsigned char map_notascii;
377 unsigned char mbs_allocated;
378 unsigned char offsets_needed;
379 unsigned char newline_anchor;
380 unsigned char word_ops_used;
383 typedef struct re_string_t re_string_t;
387 typedef struct re_dfa_t re_dfa_t;
391 # define internal_function __attribute ((regparm (3), stdcall))
393 # define internal_function
397 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
400 #ifdef RE_ENABLE_I18N
401 static void build_wcs_buffer (re_string_t *pstr) internal_function;
402 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
404 #endif /* RE_ENABLE_I18N */
405 static void build_upper_buffer (re_string_t *pstr) internal_function;
406 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
407 static unsigned int re_string_context_at (const re_string_t *input,
409 internal_function __attribute ((pure));
411 #define re_string_peek_byte(pstr, offset) \
412 ((pstr)->mbs[(pstr)->cur_idx + offset])
413 #define re_string_fetch_byte(pstr) \
414 ((pstr)->mbs[(pstr)->cur_idx++])
415 #define re_string_first_byte(pstr, idx) \
416 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
417 #define re_string_is_single_byte_char(pstr, idx) \
418 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
419 || (pstr)->wcs[(idx) + 1] != WEOF))
420 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
421 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
422 #define re_string_get_buffer(pstr) ((pstr)->mbs)
423 #define re_string_length(pstr) ((pstr)->len)
424 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
425 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
426 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
432 /* The OS usually guarantees only one guard page at the bottom of the stack,
433 and a page size can be as small as 4096 bytes. So we cannot safely
434 allocate anything larger than 4096 bytes. Also care for the possibility
435 of a few compiler-allocated temporary stack slots. */
436 # define __libc_use_alloca(n) ((n) < 4032)
438 /* alloca is implemented with malloc, so just use malloc. */
439 # define __libc_use_alloca(n) 0
443 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
444 #define re_calloc(t,n) ((t *) calloc (n, sizeof (t)))
445 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
446 #define re_free(p) free (p)
450 struct bin_tree_t *parent;
451 struct bin_tree_t *left;
452 struct bin_tree_t *right;
453 struct bin_tree_t *first;
454 struct bin_tree_t *next;
458 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
459 Otherwise `type' indicate the type of this node. */
462 typedef struct bin_tree_t bin_tree_t;
464 #define BIN_TREE_STORAGE_SIZE \
465 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
467 struct bin_tree_storage_t
469 struct bin_tree_storage_t *next;
470 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
472 typedef struct bin_tree_storage_t bin_tree_storage_t;
474 #define CONTEXT_WORD 1
475 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
476 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
477 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
479 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
480 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
481 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
482 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
483 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
485 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
486 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
487 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
488 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
490 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
491 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
492 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
493 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
494 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
496 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
497 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
498 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
499 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
500 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
506 re_node_set non_eps_nodes;
507 re_node_set inveclosure;
508 re_node_set *entrance_nodes;
509 struct re_dfastate_t **trtable, **word_trtable;
510 unsigned int context : 4;
511 unsigned int halt : 1;
512 /* If this state can accept `multi byte'.
513 Note that we refer to multibyte characters, and multi character
514 collating elements as `multi byte'. */
515 unsigned int accept_mb : 1;
516 /* If this state has backreference node(s). */
517 unsigned int has_backref : 1;
518 unsigned int has_constraint : 1;
520 typedef struct re_dfastate_t re_dfastate_t;
522 struct re_state_table_entry
526 re_dfastate_t **array;
529 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
535 re_dfastate_t **array;
538 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
543 Idx str_idx; /* The position NODE match at. */
545 } re_sub_match_last_t;
547 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
548 And information about the node, whose type is OP_CLOSE_SUBEXP,
549 corresponding to NODE is stored in LASTS. */
556 Idx alasts; /* Allocation size of LASTS. */
557 Idx nlasts; /* The number of LASTS. */
558 re_sub_match_last_t **lasts;
559 } re_sub_match_top_t;
561 struct re_backref_cache_entry
569 unsigned short int eps_reachable_subexps_map;
574 /* The string object corresponding to the input string. */
576 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
581 /* EFLAGS of the argument of regexec. */
583 /* Where the matching ends. */
586 /* The state log used by the matcher. */
587 re_dfastate_t **state_log;
589 /* Back reference cache. */
592 struct re_backref_cache_entry *bkref_ents;
596 re_sub_match_top_t **sub_tops;
597 } re_match_context_t;
601 re_dfastate_t **sifted_states;
602 re_dfastate_t **limited_states;
608 struct re_fail_stack_ent_t
613 re_node_set eps_via_nodes;
616 struct re_fail_stack_t
620 struct re_fail_stack_ent_t *stack;
631 re_node_set *eclosures;
632 re_node_set *inveclosures;
633 struct re_state_table_entry *state_table;
634 re_dfastate_t *init_state;
635 re_dfastate_t *init_state_word;
636 re_dfastate_t *init_state_nl;
637 re_dfastate_t *init_state_begbuf;
638 bin_tree_t *str_tree;
639 bin_tree_storage_t *str_tree_storage;
640 re_bitset_ptr_t sb_char;
641 int str_tree_storage_idx;
643 /* number of subexpressions `re_nsub' is in regex_t. */
644 re_hashval_t state_hash_mask;
646 Idx nbackref; /* The number of backreference in this dfa. */
648 /* Bitmap expressing which backreference is used. */
649 unsigned int used_bkref_map;
650 unsigned int completed_bkref_map;
652 unsigned int has_plural_match : 1;
653 /* If this dfa has "multibyte node", which is a backreference or
654 a node which can accept multibyte character or multi character
655 collating element. */
656 unsigned int has_mb_node : 1;
657 unsigned int is_utf8 : 1;
658 unsigned int map_notascii : 1;
659 unsigned int word_ops_used : 1;
667 __libc_lock_define (, lock)
670 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
671 #define re_node_set_remove(set,id) \
672 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
673 #define re_node_set_empty(p) ((p)->nelem = 0)
674 #define re_node_set_free(set) re_free ((set)->elems)
676 static void free_state (re_dfastate_t *state) internal_function;
690 bracket_elem_type type;
700 /* Inline functions for bitset operation. */
702 bitset_not (bitset set)
705 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
706 set[bitset_i] = ~set[bitset_i];
710 bitset_merge (bitset dest, const bitset src)
713 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
714 dest[bitset_i] |= src[bitset_i];
718 bitset_not_merge (bitset dest, const bitset src)
721 for (i = 0; i < BITSET_UINTS; ++i)
726 bitset_mask (bitset dest, const bitset src)
729 for (bitset_i = 0; bitset_i < BITSET_UINTS; ++bitset_i)
730 dest[bitset_i] &= src[bitset_i];
733 #if defined RE_ENABLE_I18N
734 /* Inline functions for re_string. */
736 internal_function __attribute ((pure))
737 re_string_char_size_at (const re_string_t *pstr, Idx idx)
740 if (pstr->mb_cur_max == 1)
742 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
743 if (pstr->wcs[idx + byte_idx] != WEOF)
749 internal_function __attribute ((pure))
750 re_string_wchar_at (const re_string_t *pstr, Idx idx)
752 if (pstr->mb_cur_max == 1)
753 return (wint_t) pstr->mbs[idx];
754 return (wint_t) pstr->wcs[idx];
758 internal_function __attribute ((pure))
759 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
762 const unsigned char *p, *extra;
763 const int32_t *table, *indirect;
765 # include <locale/weight.h>
766 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
770 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
771 extra = (const unsigned char *)
772 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
773 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
774 _NL_COLLATE_INDIRECTMB);
777 return p - pstr->mbs - idx;
783 #endif /* RE_ENABLE_I18N */
785 #endif /* _REGEX_INTERNAL_H */