1 /* Extended regular expression matching and search library.
2 Copyright (C) 2002-2025 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 <https://www.gnu.org/licenses/>. */
20 #ifndef _REGEX_INTERNAL_H
21 #define _REGEX_INTERNAL_H 1
32 #include <stdckdint.h>
36 # include <dynarray.h>
42 #if defined DEBUG && DEBUG != 0
44 # define DEBUG_ASSERT(x) assert (x)
46 # define DEBUG_ASSERT(x) assume (x)
50 # include <libc-lock.h>
51 # define lock_define(name) __libc_lock_define (, name)
52 # define lock_init(lock) (__libc_lock_init (lock), 0)
53 # define lock_fini(lock) ((void) 0)
54 # define lock_lock(lock) __libc_lock_lock (lock)
55 # define lock_unlock(lock) __libc_lock_unlock (lock)
56 #elif defined GNULIB_LOCK && !defined GNULIB_REGEX_SINGLE_THREAD
57 # include "glthread/lock.h"
58 # define lock_define(name) gl_lock_define (, name)
59 # define lock_init(lock) glthread_lock_init (&(lock))
60 # define lock_fini(lock) glthread_lock_destroy (&(lock))
61 # define lock_lock(lock) glthread_lock_lock (&(lock))
62 # define lock_unlock(lock) glthread_lock_unlock (&(lock))
63 #elif defined GNULIB_PTHREAD && !defined GNULIB_REGEX_SINGLE_THREAD
65 # define lock_define(name) pthread_mutex_t name;
66 # define lock_init(lock) pthread_mutex_init (&(lock), 0)
67 # define lock_fini(lock) pthread_mutex_destroy (&(lock))
68 # define lock_lock(lock) pthread_mutex_lock (&(lock))
69 # define lock_unlock(lock) pthread_mutex_unlock (&(lock))
71 # define lock_define(name)
72 # define lock_init(lock) 0
73 # define lock_fini(lock) ((void) 0)
74 /* The 'dfa' avoids an "unused variable 'dfa'" warning from GCC. */
75 # define lock_lock(lock) ((void) dfa)
76 # define lock_unlock(lock) ((void) 0)
79 /* In case that the system doesn't have isblank(). */
80 #if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK))
81 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
84 /* regex code assumes isascii has its usual numeric meaning,
85 even if the portable character set uses EBCDIC encoding,
86 and even if wint_t is wider than int. */
89 # define isascii(c) (((c) & ~0x7f) == 0)
93 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
94 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
95 # include <locale/localeinfo.h>
96 # include <locale/coll-lookup.h>
100 /* This is for other GNU distributions with internationalized messages. */
101 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
102 # include <libintl.h>
105 # define gettext(msgid) \
106 __dcgettext (_libc_intl_domainname, msgid, LC_MESSAGES)
108 # define gettext(msgid) dgettext ("gnulib", msgid)
112 # define gettext(msgid) (msgid)
116 /* This define is so xgettext can find the internationalizable
118 # define gettext_noop(String) String
121 /* Number of ASCII characters. */
122 #define ASCII_CHARS 0x80
124 /* Number of single byte characters. */
125 #define SBC_MAX (UCHAR_MAX + 1)
127 #define COLL_ELEM_LEN_MAX 8
129 /* The character which represents newline. */
130 #define NEWLINE_CHAR '\n'
131 #define WIDE_NEWLINE_CHAR L'\n'
133 /* Rename to standard API for using out of glibc. */
140 # define __wctype wctype
141 # define __iswalnum iswalnum
142 # define __iswctype iswctype
143 # define __towlower towlower
144 # define __towupper towupper
145 # define __btowc btowc
146 # define __mbrtowc mbrtowc
147 # define __wcrtomb wcrtomb
148 # define __regfree regfree
149 #endif /* not _LIBC */
151 /* Types related to integers. Unless protected by #ifdef _LIBC, the
152 regex code should avoid exact-width types like int32_t and uint64_t
153 as some non-GCC platforms lack them, an issue when this code is
157 # define ULONG_WIDTH REGEX_UINTEGER_WIDTH (ULONG_MAX)
158 /* The number of usable bits in an unsigned integer type with maximum
159 value MAX, as an int expression suitable in #if. Cover all known
160 practical hosts. This implementation exploits the fact that MAX is
161 1 less than a power of 2, and merely counts the number of 1 bits in
162 MAX; "COBn" means "count the number of 1 bits in the low-order n bits". */
163 # define REGEX_UINTEGER_WIDTH(max) REGEX_COB128 (max)
164 # define REGEX_COB128(n) (REGEX_COB64 ((n) >> 31 >> 31 >> 2) + REGEX_COB64 (n))
165 # define REGEX_COB64(n) (REGEX_COB32 ((n) >> 31 >> 1) + REGEX_COB32 (n))
166 # define REGEX_COB32(n) (REGEX_COB16 ((n) >> 16) + REGEX_COB16 (n))
167 # define REGEX_COB16(n) (REGEX_COB8 ((n) >> 8) + REGEX_COB8 (n))
168 # define REGEX_COB8(n) (REGEX_COB4 ((n) >> 4) + REGEX_COB4 (n))
169 # define REGEX_COB4(n) (!!((n) & 8) + !!((n) & 4) + !!((n) & 2) + ((n) & 1))
170 # if ULONG_MAX / 2 + 1 != 1ul << (ULONG_WIDTH - 1)
171 # error "ULONG_MAX out of range"
175 /* The type of indexes into strings. This is signed, not size_t,
176 since the API requires indexes to fit in regoff_t anyway, and using
177 signed integers makes the code a bit smaller and presumably faster.
178 The traditional GNU regex implementation uses int for indexes.
179 The POSIX-compatible implementation uses a possibly-wider type.
180 The name 'Idx' is three letters to minimize the hassle of
181 reindenting a lot of regex code that formerly used 'int'. */
182 typedef regoff_t Idx
;
183 #ifdef _REGEX_LARGE_OFFSETS
184 # define IDX_MAX SSIZE_MAX
186 # define IDX_MAX INT_MAX
189 /* A hash value, suitable for computing hash tables. */
190 typedef __re_size_t re_hashval_t
;
192 /* An integer used to represent a set of bits. It must be unsigned,
193 and must be at least as wide as unsigned int. */
194 typedef unsigned long int bitset_word_t
;
195 /* All bits set in a bitset_word_t. */
196 #define BITSET_WORD_MAX ULONG_MAX
197 /* Number of bits in a bitset_word_t. */
198 #define BITSET_WORD_BITS ULONG_WIDTH
200 /* Number of bitset_word_t values in a bitset_t. */
201 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
203 typedef bitset_word_t bitset_t
[BITSET_WORDS
];
204 typedef bitset_word_t
*re_bitset_ptr_t
;
205 typedef const bitset_word_t
*re_const_bitset_ptr_t
;
207 #define PREV_WORD_CONSTRAINT 0x0001
208 #define PREV_NOTWORD_CONSTRAINT 0x0002
209 #define NEXT_WORD_CONSTRAINT 0x0004
210 #define NEXT_NOTWORD_CONSTRAINT 0x0008
211 #define PREV_NEWLINE_CONSTRAINT 0x0010
212 #define NEXT_NEWLINE_CONSTRAINT 0x0020
213 #define PREV_BEGBUF_CONSTRAINT 0x0040
214 #define NEXT_ENDBUF_CONSTRAINT 0x0080
215 #define WORD_DELIM_CONSTRAINT 0x0100
216 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
220 INSIDE_WORD
= PREV_WORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
221 WORD_FIRST
= PREV_NOTWORD_CONSTRAINT
| NEXT_WORD_CONSTRAINT
,
222 WORD_LAST
= PREV_WORD_CONSTRAINT
| NEXT_NOTWORD_CONSTRAINT
,
223 INSIDE_NOTWORD
= PREV_NOTWORD_CONSTRAINT
| NEXT_NOTWORD_CONSTRAINT
,
224 LINE_FIRST
= PREV_NEWLINE_CONSTRAINT
,
225 LINE_LAST
= NEXT_NEWLINE_CONSTRAINT
,
226 BUF_FIRST
= PREV_BEGBUF_CONSTRAINT
,
227 BUF_LAST
= NEXT_ENDBUF_CONSTRAINT
,
228 WORD_DELIM
= WORD_DELIM_CONSTRAINT
,
229 NOT_WORD_DELIM
= NOT_WORD_DELIM_CONSTRAINT
243 /* Node type, These are used by token, node, tree. */
252 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
253 when the debugger shows values of this enum type. */
254 #define EPSILON_BIT 8
255 OP_OPEN_SUBEXP
= EPSILON_BIT
| 0,
256 OP_CLOSE_SUBEXP
= EPSILON_BIT
| 1,
257 OP_ALT
= EPSILON_BIT
| 2,
258 OP_DUP_ASTERISK
= EPSILON_BIT
| 3,
259 ANCHOR
= EPSILON_BIT
| 4,
261 /* Tree type, these are used only by tree. */
265 /* Token type, these are used only by token. */
277 OP_CLOSE_EQUIV_CLASS
,
290 /* Multibyte characters. */
294 /* Collating symbols. */
299 /* Equivalence classes. */
300 int32_t *equiv_classes
;
303 /* Range expressions. */
305 uint32_t *range_starts
;
306 uint32_t *range_ends
;
308 wchar_t *range_starts
;
312 /* Character classes. */
313 wctype_t *char_classes
;
315 /* If this character set is the non-matching list. */
316 unsigned int non_match
: 1;
318 /* # of multibyte characters. */
321 /* # of collating symbols. */
324 /* # of equivalence classes. */
327 /* # of range expressions. */
330 /* # of character classes. */
338 unsigned char c
; /* for CHARACTER */
339 re_bitset_ptr_t sbcset
; /* for SIMPLE_BRACKET */
340 re_charset_t
*mbcset
; /* for COMPLEX_BRACKET */
341 Idx idx
; /* for BACK_REF */
342 re_context_type ctx_type
; /* for ANCHOR */
344 #if (__GNUC__ >= 2 || defined __clang__) && !defined __STRICT_ANSI__
345 re_token_type_t type
: 8;
347 re_token_type_t type
;
349 unsigned int constraint
: 10; /* context constraint */
350 unsigned int duplicated
: 1;
351 unsigned int opt_subexp
: 1;
352 unsigned int accept_mb
: 1;
353 /* These 2 bits can be moved into the union if needed (e.g. if running out
354 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
355 unsigned int mb_partial
: 1;
356 unsigned int word_char
: 1;
359 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
363 /* Indicate the raw buffer which is the original string passed as an
364 argument of regexec(), re_search(), etc.. */
365 const unsigned char *raw_mbs
;
366 /* Store the multibyte string. In case of "case insensitive mode" like
367 REG_ICASE, upper cases of the string are stored, otherwise MBS points
368 the same address that RAW_MBS points. */
370 /* Store the wide character string which is corresponding to MBS. */
374 /* Index in RAW_MBS. Each character mbs[i] corresponds to
375 raw_mbs[raw_mbs_idx + i]. */
377 /* The length of the valid characters in the buffers. */
379 /* The corresponding number of bytes in raw_mbs array. */
381 /* The length of the buffers MBS and WCS. */
383 /* The index in MBS, which is updated by re_string_fetch_byte. */
385 /* length of RAW_MBS array. */
387 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
389 /* End of the buffer may be shorter than its length in the cases such
390 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
393 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
396 /* The context of mbs[0]. We store the context independently, since
397 the context of mbs[0] may be different from raw_mbs[0], which is
398 the beginning of the input string. */
399 unsigned int tip_context
;
400 /* The translation passed as a part of an argument of re_compile_pattern. */
401 RE_TRANSLATE_TYPE trans
;
402 /* Copy of re_dfa_t's word_char. */
403 re_const_bitset_ptr_t word_char
;
404 /* true if REG_ICASE. */
406 unsigned char is_utf8
;
407 unsigned char map_notascii
;
408 unsigned char mbs_allocated
;
409 unsigned char offsets_needed
;
410 unsigned char newline_anchor
;
411 unsigned char word_ops_used
;
414 typedef struct re_string_t re_string_t
;
418 typedef struct re_dfa_t re_dfa_t
;
421 # define IS_IN(libc) false
424 #define re_string_peek_byte(pstr, offset) \
425 ((pstr)->mbs[(pstr)->cur_idx + offset])
426 #define re_string_fetch_byte(pstr) \
427 ((pstr)->mbs[(pstr)->cur_idx++])
428 #define re_string_first_byte(pstr, idx) \
429 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
430 #define re_string_is_single_byte_char(pstr, idx) \
431 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
432 || (pstr)->wcs[(idx) + 1] != WEOF))
433 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
434 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
435 #define re_string_get_buffer(pstr) ((pstr)->mbs)
436 #define re_string_length(pstr) ((pstr)->len)
437 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
438 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
439 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
442 # define MAX(a,b) ((a) < (b) ? (b) : (a))
445 # define MIN(a,b) ((a) < (b) ? (a) : (b))
448 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
449 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
450 #define re_free(p) free (p)
454 struct bin_tree_t
*parent
;
455 struct bin_tree_t
*left
;
456 struct bin_tree_t
*right
;
457 struct bin_tree_t
*first
;
458 struct bin_tree_t
*next
;
462 /* 'node_idx' is the index in dfa->nodes, if 'type' == 0.
463 Otherwise 'type' indicate the type of this node. */
466 typedef struct bin_tree_t bin_tree_t
;
468 #define BIN_TREE_STORAGE_SIZE \
469 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
471 struct bin_tree_storage_t
473 struct bin_tree_storage_t
*next
;
474 bin_tree_t data
[BIN_TREE_STORAGE_SIZE
];
476 typedef struct bin_tree_storage_t bin_tree_storage_t
;
478 #define CONTEXT_WORD 1
479 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
480 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
481 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
483 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
484 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
485 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
486 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
487 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
489 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
490 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
491 #define IS_WIDE_WORD_CHAR(ch) (__iswalnum (ch) || (ch) == L'_')
492 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
494 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
495 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
496 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
497 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
498 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
500 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
501 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
502 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
503 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
504 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
510 re_node_set non_eps_nodes
;
511 re_node_set inveclosure
;
512 re_node_set
*entrance_nodes
;
513 struct re_dfastate_t
**trtable
, **word_trtable
;
514 unsigned int context
: 4;
515 unsigned int halt
: 1;
516 /* If this state can accept "multi byte".
517 Note that we refer to multibyte characters, and multi character
518 collating elements as "multi byte". */
519 unsigned int accept_mb
: 1;
520 /* If this state has backreference node(s). */
521 unsigned int has_backref
: 1;
522 unsigned int has_constraint
: 1;
524 typedef struct re_dfastate_t re_dfastate_t
;
526 struct re_state_table_entry
530 re_dfastate_t
**array
;
533 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
539 re_dfastate_t
**array
;
542 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
547 Idx str_idx
; /* The position NODE match at. */
549 } re_sub_match_last_t
;
551 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
552 And information about the node, whose type is OP_CLOSE_SUBEXP,
553 corresponding to NODE is stored in LASTS. */
560 Idx alasts
; /* Allocation size of LASTS. */
561 Idx nlasts
; /* The number of LASTS. */
562 re_sub_match_last_t
**lasts
;
563 } re_sub_match_top_t
;
565 struct re_backref_cache_entry
571 bitset_word_t eps_reachable_subexps_map
;
577 /* The string object corresponding to the input string. */
579 const re_dfa_t
*const dfa
;
580 /* EFLAGS of the argument of regexec. */
582 /* Where the matching ends. */
585 /* The state log used by the matcher. */
586 re_dfastate_t
**state_log
;
588 /* Back reference cache. */
591 struct re_backref_cache_entry
*bkref_ents
;
595 re_sub_match_top_t
**sub_tops
;
596 } re_match_context_t
;
600 re_dfastate_t
**sifted_states
;
601 re_dfastate_t
**limited_states
;
607 struct re_fail_stack_ent_t
612 re_node_set eps_via_nodes
;
615 struct re_fail_stack_t
619 struct re_fail_stack_ent_t
*stack
;
630 re_node_set
*eclosures
;
631 re_node_set
*inveclosures
;
632 struct re_state_table_entry
*state_table
;
633 re_dfastate_t
*init_state
;
634 re_dfastate_t
*init_state_word
;
635 re_dfastate_t
*init_state_nl
;
636 re_dfastate_t
*init_state_begbuf
;
637 bin_tree_t
*str_tree
;
638 bin_tree_storage_t
*str_tree_storage
;
639 re_bitset_ptr_t sb_char
;
640 int str_tree_storage_idx
;
642 /* number of subexpressions 're_nsub' is in regex_t. */
643 re_hashval_t state_hash_mask
;
645 Idx nbackref
; /* The number of backreference in this dfa. */
647 /* Bitmap expressing which backreference is used. */
648 bitset_word_t used_bkref_map
;
649 bitset_word_t completed_bkref_map
;
651 unsigned int has_plural_match
: 1;
652 /* If this dfa has "multibyte node", which is a backreference or
653 a node which can accept multibyte character or multi character
654 collating element. */
655 unsigned int has_mb_node
: 1;
656 unsigned int is_utf8
: 1;
657 unsigned int map_notascii
: 1;
658 unsigned int word_ops_used
: 1;
669 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
670 #define re_node_set_remove(set,id) \
671 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
672 #define re_node_set_empty(p) ((p)->nelem = 0)
673 #define re_node_set_free(set) re_free ((set)->elems)
687 bracket_elem_type type
;
697 /* Functions for bitset_t operation. */
700 bitset_set (bitset_t set
, Idx i
)
702 set
[i
/ BITSET_WORD_BITS
] |= (bitset_word_t
) 1 << i
% BITSET_WORD_BITS
;
706 bitset_clear (bitset_t set
, Idx i
)
708 set
[i
/ BITSET_WORD_BITS
] &= ~ ((bitset_word_t
) 1 << i
% BITSET_WORD_BITS
);
712 bitset_contain (const bitset_t set
, Idx i
)
714 return (set
[i
/ BITSET_WORD_BITS
] >> i
% BITSET_WORD_BITS
) & 1;
718 bitset_empty (bitset_t set
)
720 memset (set
, '\0', sizeof (bitset_t
));
724 bitset_set_all (bitset_t set
)
726 memset (set
, -1, sizeof (bitset_word_t
) * (SBC_MAX
/ BITSET_WORD_BITS
));
727 if (SBC_MAX
% BITSET_WORD_BITS
!= 0)
728 set
[BITSET_WORDS
- 1] =
729 ((bitset_word_t
) 1 << SBC_MAX
% BITSET_WORD_BITS
) - 1;
733 bitset_copy (bitset_t dest
, const bitset_t src
)
735 memcpy (dest
, src
, sizeof (bitset_t
));
739 bitset_not (bitset_t set
)
742 for (bitset_i
= 0; bitset_i
< SBC_MAX
/ BITSET_WORD_BITS
; ++bitset_i
)
743 set
[bitset_i
] = ~set
[bitset_i
];
744 if (SBC_MAX
% BITSET_WORD_BITS
!= 0)
745 set
[BITSET_WORDS
- 1] =
746 ((((bitset_word_t
) 1 << SBC_MAX
% BITSET_WORD_BITS
) - 1)
747 & ~set
[BITSET_WORDS
- 1]);
751 bitset_merge (bitset_t dest
, const bitset_t src
)
754 for (bitset_i
= 0; bitset_i
< BITSET_WORDS
; ++bitset_i
)
755 dest
[bitset_i
] |= src
[bitset_i
];
759 bitset_mask (bitset_t dest
, const bitset_t src
)
762 for (bitset_i
= 0; bitset_i
< BITSET_WORDS
; ++bitset_i
)
763 dest
[bitset_i
] &= src
[bitset_i
];
766 /* Functions for re_string. */
768 __attribute__ ((pure
, unused
))
769 re_string_char_size_at (const re_string_t
*pstr
, Idx idx
)
772 if (pstr
->mb_cur_max
== 1)
774 for (byte_idx
= 1; idx
+ byte_idx
< pstr
->valid_len
; ++byte_idx
)
775 if (pstr
->wcs
[idx
+ byte_idx
] != WEOF
)
781 __attribute__ ((pure
, unused
))
782 re_string_wchar_at (const re_string_t
*pstr
, Idx idx
)
784 if (pstr
->mb_cur_max
== 1)
785 return (wint_t) pstr
->mbs
[idx
];
786 return (wint_t) pstr
->wcs
[idx
];
790 # include <locale/weight.h>
794 __attribute__ ((pure
, unused
))
795 re_string_elem_size_at (const re_string_t
*pstr
, Idx idx
)
798 const unsigned char *p
, *extra
;
799 const int32_t *table
, *indirect
;
800 uint_fast32_t nrules
= _NL_CURRENT_WORD (LC_COLLATE
, _NL_COLLATE_NRULES
);
804 table
= (const int32_t *) _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_TABLEMB
);
805 extra
= (const unsigned char *)
806 _NL_CURRENT (LC_COLLATE
, _NL_COLLATE_EXTRAMB
);
807 indirect
= (const int32_t *) _NL_CURRENT (LC_COLLATE
,
808 _NL_COLLATE_INDIRECTMB
);
810 findidx (table
, indirect
, extra
, &p
, pstr
->len
- idx
);
811 return p
- pstr
->mbs
- idx
;
819 # if __glibc_has_attribute (__fallthrough__)
820 # define FALLTHROUGH __attribute__ ((__fallthrough__))
822 # define FALLTHROUGH ((void) 0)
825 # include "attribute.h"
828 #endif /* _REGEX_INTERNAL_H */