1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002-2012 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, see
18 <http://www.gnu.org/licenses/>. */
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 */
41 #if defined HAVE_STDBOOL_H || defined _LIBC
43 #endif /* HAVE_STDBOOL_H || _LIBC */
44 #if defined HAVE_STDINT_H || defined _LIBC
46 #endif /* HAVE_STDINT_H || _LIBC */
48 # include <bits/libc-lock.h>
50 # define __libc_lock_define(CLASS,NAME)
51 # define __libc_lock_init(NAME) do { } while (0)
52 # define __libc_lock_lock(NAME) do { } while (0)
53 # define __libc_lock_unlock(NAME) do { } while (0)
56 /* In case that the system doesn't have isblank(). */
57 #if !defined _LIBC && !defined HAVE_ISBLANK && !defined isblank
58 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
62 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
63 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
64 # include <locale/localeinfo.h>
65 # include <locale/elem-hash.h>
66 # include <locale/coll-lookup.h>
70 /* This is for other GNU distributions with internationalized messages. */
71 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
75 # define gettext(msgid) \
76 __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
79 # define gettext(msgid) (msgid)
83 /* This define is so xgettext can find the internationalizable
85 # define gettext_noop(String) String
88 /* For loser systems without the definition. */
90 # define SIZE_MAX ((size_t) -1)
93 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
94 # define RE_ENABLE_I18N
98 # define BE(expr, val) __builtin_expect (expr, val)
100 # define BE(expr, val) (expr)
104 /* Number of single byte character. */
107 #define COLL_ELEM_LEN_MAX 8
109 /* The character which represents newline. */
110 #define NEWLINE_CHAR '\n'
111 #define WIDE_NEWLINE_CHAR L'\n'
113 /* Rename to standard API for using out of glibc. */
115 # define __wctype wctype
116 # define __iswctype iswctype
117 # define __btowc btowc
118 # define __mbrtowc mbrtowc
119 # define __mempcpy mempcpy
120 # define __wcrtomb wcrtomb
121 # define __regfree regfree
122 # define attribute_hidden
123 #endif /* not _LIBC */
126 # define __attribute(arg) __attribute__ (arg)
128 # define __attribute(arg)
131 extern const char __re_error_msgid[] attribute_hidden;
132 extern const size_t __re_error_msgid_idx[] attribute_hidden;
134 /* An integer used to represent a set of bits. It must be unsigned,
135 and must be at least as wide as unsigned int. */
136 typedef unsigned long int bitset_word_t;
137 /* All bits set in a bitset_word_t. */
138 #define BITSET_WORD_MAX ULONG_MAX
139 /* Number of bits in a bitset_word_t. */
140 #define BITSET_WORD_BITS (sizeof (bitset_word_t) * CHAR_BIT)
141 /* Number of bitset_word_t in a bit_set. */
142 #define BITSET_WORDS (SBC_MAX / BITSET_WORD_BITS)
143 typedef bitset_word_t bitset_t[BITSET_WORDS];
144 typedef bitset_word_t *re_bitset_ptr_t;
145 typedef const bitset_word_t *re_const_bitset_ptr_t;
147 #define bitset_set(set,i) \
148 (set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS)
149 #define bitset_clear(set,i) \
150 (set[i / BITSET_WORD_BITS] &= ~((bitset_word_t) 1 << i % BITSET_WORD_BITS))
151 #define bitset_contain(set,i) \
152 (set[i / BITSET_WORD_BITS] & ((bitset_word_t) 1 << i % BITSET_WORD_BITS))
153 #define bitset_empty(set) memset (set, '\0', sizeof (bitset_t))
154 #define bitset_set_all(set) memset (set, '\xff', sizeof (bitset_t))
155 #define bitset_copy(dest,src) memcpy (dest, src, sizeof (bitset_t))
157 #define PREV_WORD_CONSTRAINT 0x0001
158 #define PREV_NOTWORD_CONSTRAINT 0x0002
159 #define NEXT_WORD_CONSTRAINT 0x0004
160 #define NEXT_NOTWORD_CONSTRAINT 0x0008
161 #define PREV_NEWLINE_CONSTRAINT 0x0010
162 #define NEXT_NEWLINE_CONSTRAINT 0x0020
163 #define PREV_BEGBUF_CONSTRAINT 0x0040
164 #define NEXT_ENDBUF_CONSTRAINT 0x0080
165 #define WORD_DELIM_CONSTRAINT 0x0100
166 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
170 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
171 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
172 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
173 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
174 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
175 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
176 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
177 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
178 WORD_DELIM = WORD_DELIM_CONSTRAINT,
179 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
193 /* Node type, These are used by token, node, tree. */
199 #ifdef RE_ENABLE_I18N
202 #endif /* RE_ENABLE_I18N */
204 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
205 when the debugger shows values of this enum type. */
206 #define EPSILON_BIT 8
207 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
208 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
209 OP_ALT = EPSILON_BIT | 2,
210 OP_DUP_ASTERISK = EPSILON_BIT | 3,
211 ANCHOR = EPSILON_BIT | 4,
213 /* Tree type, these are used only by tree. */
217 /* Token type, these are used only by token. */
229 OP_CLOSE_EQUIV_CLASS,
240 #ifdef RE_ENABLE_I18N
243 /* Multibyte characters. */
246 /* Collating symbols. */
251 /* Equivalence classes. */
253 int32_t *equiv_classes;
256 /* Range expressions. */
258 uint32_t *range_starts;
259 uint32_t *range_ends;
260 # else /* not _LIBC */
261 wchar_t *range_starts;
263 # endif /* not _LIBC */
265 /* Character classes. */
266 wctype_t *char_classes;
268 /* If this character set is the non-matching list. */
269 unsigned int non_match : 1;
271 /* # of multibyte characters. */
274 /* # of collating symbols. */
277 /* # of equivalence classes. */
280 /* # of range expressions. */
283 /* # of character classes. */
286 #endif /* RE_ENABLE_I18N */
292 unsigned char c; /* for CHARACTER */
293 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
294 #ifdef RE_ENABLE_I18N
295 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
296 #endif /* RE_ENABLE_I18N */
297 int idx; /* for BACK_REF */
298 re_context_type ctx_type; /* for ANCHOR */
301 re_token_type_t type : 8;
303 re_token_type_t type;
305 unsigned int constraint : 10; /* context constraint */
306 unsigned int duplicated : 1;
307 unsigned int opt_subexp : 1;
308 #ifdef RE_ENABLE_I18N
309 unsigned int accept_mb : 1;
310 /* These 2 bits can be moved into the union if needed (e.g. if running out
311 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
312 unsigned int mb_partial : 1;
314 unsigned int word_char : 1;
317 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
321 /* Indicate the raw buffer which is the original string passed as an
322 argument of regexec(), re_search(), etc.. */
323 const unsigned char *raw_mbs;
324 /* Store the multibyte string. In case of "case insensitive mode" like
325 REG_ICASE, upper cases of the string are stored, otherwise MBS points
326 the same address that RAW_MBS points. */
328 #ifdef RE_ENABLE_I18N
329 /* Store the wide character string which is corresponding to MBS. */
334 /* Index in RAW_MBS. Each character mbs[i] corresponds to
335 raw_mbs[raw_mbs_idx + i]. */
337 /* The length of the valid characters in the buffers. */
339 /* The corresponding number of bytes in raw_mbs array. */
341 /* The length of the buffers MBS and WCS. */
343 /* The index in MBS, which is updated by re_string_fetch_byte. */
345 /* length of RAW_MBS array. */
347 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
349 /* End of the buffer may be shorter than its length in the cases such
350 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
353 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
356 /* The context of mbs[0]. We store the context independently, since
357 the context of mbs[0] may be different from raw_mbs[0], which is
358 the beginning of the input string. */
359 unsigned int tip_context;
360 /* The translation passed as a part of an argument of re_compile_pattern. */
361 RE_TRANSLATE_TYPE trans;
362 /* Copy of re_dfa_t's word_char. */
363 re_const_bitset_ptr_t word_char;
364 /* 1 if REG_ICASE. */
366 unsigned char is_utf8;
367 unsigned char map_notascii;
368 unsigned char mbs_allocated;
369 unsigned char offsets_needed;
370 unsigned char newline_anchor;
371 unsigned char word_ops_used;
374 typedef struct re_string_t re_string_t;
378 typedef struct re_dfa_t re_dfa_t;
382 # define internal_function __attribute ((regparm (3), stdcall))
384 # define internal_function
389 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
392 # ifdef RE_ENABLE_I18N
393 static void build_wcs_buffer (re_string_t *pstr) internal_function;
394 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
396 # endif /* RE_ENABLE_I18N */
397 static void build_upper_buffer (re_string_t *pstr) internal_function;
398 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
399 static unsigned int re_string_context_at (const re_string_t *input, int idx,
401 internal_function __attribute ((pure));
403 #define re_string_peek_byte(pstr, offset) \
404 ((pstr)->mbs[(pstr)->cur_idx + offset])
405 #define re_string_fetch_byte(pstr) \
406 ((pstr)->mbs[(pstr)->cur_idx++])
407 #define re_string_first_byte(pstr, idx) \
408 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
409 #define re_string_is_single_byte_char(pstr, idx) \
410 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
411 || (pstr)->wcs[(idx) + 1] != WEOF))
412 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
413 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
414 #define re_string_get_buffer(pstr) ((pstr)->mbs)
415 #define re_string_length(pstr) ((pstr)->len)
416 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
417 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
418 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
426 /* The OS usually guarantees only one guard page at the bottom of the stack,
427 and a page size can be as small as 4096 bytes. So we cannot safely
428 allocate anything larger than 4096 bytes. Also care for the possibility
429 of a few compiler-allocated temporary stack slots. */
430 # define __libc_use_alloca(n) ((n) < 4032)
432 /* alloca is implemented with malloc, so just use malloc. */
433 # define __libc_use_alloca(n) 0
437 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
438 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
439 #define re_free(p) free (p)
443 struct bin_tree_t *parent;
444 struct bin_tree_t *left;
445 struct bin_tree_t *right;
446 struct bin_tree_t *first;
447 struct bin_tree_t *next;
451 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
452 Otherwise `type' indicate the type of this node. */
455 typedef struct bin_tree_t bin_tree_t;
457 #define BIN_TREE_STORAGE_SIZE \
458 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
460 struct bin_tree_storage_t
462 struct bin_tree_storage_t *next;
463 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
465 typedef struct bin_tree_storage_t bin_tree_storage_t;
467 #define CONTEXT_WORD 1
468 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
469 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
470 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
472 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
473 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
474 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
475 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
476 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
478 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
479 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
480 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
481 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
483 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
484 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
485 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
486 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
487 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
489 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
490 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
491 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
492 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
493 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
499 re_node_set non_eps_nodes;
500 re_node_set inveclosure;
501 re_node_set *entrance_nodes;
502 struct re_dfastate_t **trtable, **word_trtable;
503 unsigned int context : 4;
504 unsigned int halt : 1;
505 /* If this state can accept `multi byte'.
506 Note that we refer to multibyte characters, and multi character
507 collating elements as `multi byte'. */
508 unsigned int accept_mb : 1;
509 /* If this state has backreference node(s). */
510 unsigned int has_backref : 1;
511 unsigned int has_constraint : 1;
513 typedef struct re_dfastate_t re_dfastate_t;
515 struct re_state_table_entry
519 re_dfastate_t **array;
522 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
528 re_dfastate_t **array;
531 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
536 int str_idx; /* The position NODE match at. */
538 } re_sub_match_last_t;
540 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
541 And information about the node, whose type is OP_CLOSE_SUBEXP,
542 corresponding to NODE is stored in LASTS. */
549 int alasts; /* Allocation size of LASTS. */
550 int nlasts; /* The number of LASTS. */
551 re_sub_match_last_t **lasts;
552 } re_sub_match_top_t;
554 struct re_backref_cache_entry
562 unsigned short int eps_reachable_subexps_map;
567 /* The string object corresponding to the input string. */
569 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
570 const re_dfa_t *const dfa;
574 /* EFLAGS of the argument of regexec. */
576 /* Where the matching ends. */
579 /* The state log used by the matcher. */
580 re_dfastate_t **state_log;
582 /* Back reference cache. */
585 struct re_backref_cache_entry *bkref_ents;
589 re_sub_match_top_t **sub_tops;
590 } re_match_context_t;
594 re_dfastate_t **sifted_states;
595 re_dfastate_t **limited_states;
601 struct re_fail_stack_ent_t
606 re_node_set eps_via_nodes;
609 struct re_fail_stack_t
613 struct re_fail_stack_ent_t *stack;
624 re_node_set *eclosures;
625 re_node_set *inveclosures;
626 struct re_state_table_entry *state_table;
627 re_dfastate_t *init_state;
628 re_dfastate_t *init_state_word;
629 re_dfastate_t *init_state_nl;
630 re_dfastate_t *init_state_begbuf;
631 bin_tree_t *str_tree;
632 bin_tree_storage_t *str_tree_storage;
633 re_bitset_ptr_t sb_char;
634 int str_tree_storage_idx;
636 /* number of subexpressions `re_nsub' is in regex_t. */
637 unsigned int state_hash_mask;
639 int nbackref; /* The number of backreference in this dfa. */
641 /* Bitmap expressing which backreference is used. */
642 bitset_word_t used_bkref_map;
643 bitset_word_t completed_bkref_map;
645 unsigned int has_plural_match : 1;
646 /* If this dfa has "multibyte node", which is a backreference or
647 a node which can accept multibyte character or multi character
648 collating element. */
649 unsigned int has_mb_node : 1;
650 unsigned int is_utf8 : 1;
651 unsigned int map_notascii : 1;
652 unsigned int word_ops_used : 1;
660 __libc_lock_define (, lock)
663 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
664 #define re_node_set_remove(set,id) \
665 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
666 #define re_node_set_empty(p) ((p)->nelem = 0)
667 #define re_node_set_free(set) re_free ((set)->elems)
681 bracket_elem_type type;
691 /* Inline functions for bitset operation. */
693 bitset_not (bitset_t set)
696 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
697 set[bitset_i] = ~set[bitset_i];
701 bitset_merge (bitset_t dest, const bitset_t src)
704 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
705 dest[bitset_i] |= src[bitset_i];
709 bitset_mask (bitset_t dest, const bitset_t src)
712 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
713 dest[bitset_i] &= src[bitset_i];
716 #ifdef RE_ENABLE_I18N
717 /* Inline functions for re_string. */
719 internal_function __attribute ((pure))
720 re_string_char_size_at (const re_string_t *pstr, int idx)
723 if (pstr->mb_cur_max == 1)
725 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
726 if (pstr->wcs[idx + byte_idx] != WEOF)
732 internal_function __attribute ((pure))
733 re_string_wchar_at (const re_string_t *pstr, int idx)
735 if (pstr->mb_cur_max == 1)
736 return (wint_t) pstr->mbs[idx];
737 return (wint_t) pstr->wcs[idx];
742 internal_function __attribute ((pure))
743 re_string_elem_size_at (const re_string_t *pstr, int idx)
746 const unsigned char *p, *extra;
747 const int32_t *table, *indirect;
748 # include <locale/weight.h>
749 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
753 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
754 extra = (const unsigned char *)
755 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
756 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
757 _NL_COLLATE_INDIRECTMB);
759 findidx (&p, pstr->len - idx);
760 return p - pstr->mbs - idx;
767 #endif /* RE_ENABLE_I18N */
769 #endif /* _REGEX_INTERNAL_H */