annotate lib/regex_internal.h @ 8153:0f48fdcc7a64

Fix regex code so it doesn't rely on strcasecmp. * lib/regex_internal.h: Include <langinfo.h> only if _LIBC is defined. Otherwise, include gnulib's langinfo.h. * lib/regcomp.c (init_dfa): Don't use strcasecmp, as it can have undesirable behavior in non-C locales. Instead, rely on locale_charset. * m4/regex.m4 (gl_PREREQ_REGEX): Don't require AM_LANGINFO_CODESET. * modules/regex (FILES): Remove m4/codeset.m4. (Depends-on): Add localcharset. Remove strcase.
author Paul Eggert <eggert@cs.ucla.edu>
date Thu, 15 Feb 2007 00:16:55 +0000
parents eaa00773406b
children 28ffab893f7a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
1 /* Extended regular expression matching and search library.
7888
b6376840b47b * modules/fnmatch (Depends-on): Depend on wchar.
Eric Blake <ebb9@byu.net>
parents: 7733
diff changeset
2 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
3 This file is part of the GNU C Library.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
4 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
5
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
6 This program is free software; you can redistribute it and/or modify
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
7 it under the terms of the GNU General Public License as published by
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
8 the Free Software Foundation; either version 2, or (at your option)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
9 any later version.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
10
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
11 This program is distributed in the hope that it will be useful,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
14 GNU General Public License for more details.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
15
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
16 You should have received a copy of the GNU General Public License along
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
17 with this program; if not, write to the Free Software Foundation,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
18 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
19
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
20 #ifndef _REGEX_INTERNAL_H
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
21 #define _REGEX_INTERNAL_H 1
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
22
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
23 #include <assert.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
24 #include <ctype.h>
6195
25eaa608fc4e Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6184
diff changeset
25 #include <stdbool.h>
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
26 #include <stdio.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
27 #include <stdlib.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
28 #include <string.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
29
8153
0f48fdcc7a64 Fix regex code so it doesn't rely on strcasecmp.
Paul Eggert <eggert@cs.ucla.edu>
parents: 8073
diff changeset
30 #ifdef _LIBC
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
31 # include <langinfo.h>
8153
0f48fdcc7a64 Fix regex code so it doesn't rely on strcasecmp.
Paul Eggert <eggert@cs.ucla.edu>
parents: 8073
diff changeset
32 #else
0f48fdcc7a64 Fix regex code so it doesn't rely on strcasecmp.
Paul Eggert <eggert@cs.ucla.edu>
parents: 8073
diff changeset
33 # include "localcharset.h"
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
34 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
35 #if defined HAVE_LOCALE_H || defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
36 # include <locale.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
37 #endif
7888
b6376840b47b * modules/fnmatch (Depends-on): Depend on wchar.
Eric Blake <ebb9@byu.net>
parents: 7733
diff changeset
38
b6376840b47b * modules/fnmatch (Depends-on): Depend on wchar.
Eric Blake <ebb9@byu.net>
parents: 7733
diff changeset
39 #include <wchar.h>
7733
9d60f39f35c1 * MODULES.html.sh: New module wctype.
Paul Eggert <eggert@cs.ucla.edu>
parents: 7709
diff changeset
40 #include <wctype.h>
7162
19c2e5121b2f Add and change modules to make it easier for coreutils to use
Paul Eggert <eggert@cs.ucla.edu>
parents: 7066
diff changeset
41 #include <stdint.h>
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
42 #if defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
43 # include <bits/libc-lock.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
44 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
45 # define __libc_lock_init(NAME) do { } while (0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
46 # define __libc_lock_lock(NAME) do { } while (0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
47 # define __libc_lock_unlock(NAME) do { } while (0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
48 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
49
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
50 /* In case that the system doesn't have isblank(). */
6932
6aeb4d6c28d9 * lib/.cppi-disable: Add wcwidth.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6726
diff changeset
51 #if !defined _LIBC && !HAVE_DECL_ISBLANK && !defined isblank
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
52 # define isblank(ch) ((ch) == ' ' || (ch) == '\t')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
53 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
54
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
55 #ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
56 # ifndef _RE_DEFINE_LOCALE_FUNCTIONS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
57 # define _RE_DEFINE_LOCALE_FUNCTIONS 1
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
58 # include <locale/localeinfo.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
59 # include <locale/elem-hash.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
60 # include <locale/coll-lookup.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
61 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
62 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
63
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
64 /* This is for other GNU distributions with internationalized messages. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
65 #if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
66 # include <libintl.h>
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
67 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
68 # undef gettext
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
69 # define gettext(msgid) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
70 INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
71 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
72 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
73 # define gettext(msgid) (msgid)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
74 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
75
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
76 #ifndef gettext_noop
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
77 /* This define is so xgettext can find the internationalizable
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
78 strings. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
79 # define gettext_noop(String) String
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
80 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
81
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
82 /* For loser systems without the definition. */
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
83 #ifndef SIZE_MAX
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
84 # define SIZE_MAX ((size_t) -1)
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
85 #endif
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
86
7888
b6376840b47b * modules/fnmatch (Depends-on): Depend on wchar.
Eric Blake <ebb9@byu.net>
parents: 7733
diff changeset
87 #if (defined MB_CUR_MAX && HAVE_LOCALE_H && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCRTOMB && HAVE_MBRTOWC && HAVE_WCSCOLL) || _LIBC
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
88 # define RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
89 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
90
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
91 #if __GNUC__ >= 3
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
92 # define BE(expr, val) __builtin_expect (expr, val)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
93 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
94 # define BE(expr, val) (expr)
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
95 # ifdef _LIBC
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
96 # define inline
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
97 # endif
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
98 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
99
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
100 /* Number of ASCII characters. */
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
101 #define ASCII_CHARS 0x80
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
102
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
103 /* Number of single byte characters. */
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
104 #define SBC_MAX (UCHAR_MAX + 1)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
105
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
106 #define COLL_ELEM_LEN_MAX 8
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
107
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
108 /* The character which represents newline. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
109 #define NEWLINE_CHAR '\n'
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
110 #define WIDE_NEWLINE_CHAR L'\n'
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
111
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
112 /* Rename to standard API for using out of glibc. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
113 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
114 # define __wctype wctype
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
115 # define __iswctype iswctype
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
116 # define __btowc btowc
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
117 # define __wcrtomb wcrtomb
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
118 # define __regfree regfree
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
119 # define attribute_hidden
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
120 #endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
121
6296
ac6a92dc851d * regex_internal.h (__GNUC_PREREQ, always_inline, inline, pure):
Paul Eggert <eggert@cs.ucla.edu>
parents: 6294
diff changeset
122 #if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
123 # define __attribute(arg) __attribute__ (arg)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
124 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
125 # define __attribute(arg)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
126 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
127
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
128 typedef __re_idx_t Idx;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
129
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
130 /* Special return value for failure to match. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
131 #define REG_MISSING ((Idx) -1)
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
132
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
133 /* Special return value for internal error. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
134 #define REG_ERROR ((Idx) -2)
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
135
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
136 /* Test whether N is a valid index, and is not one of the above. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
137 #ifdef _REGEX_LARGE_OFFSETS
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
138 # define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR)
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
139 #else
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
140 # define REG_VALID_INDEX(n) (0 <= (n))
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
141 #endif
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
142
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
143 /* Test whether N is a valid nonzero index. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
144 #ifdef _REGEX_LARGE_OFFSETS
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
145 # define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1))
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
146 #else
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
147 # define REG_VALID_NONZERO_INDEX(n) (0 < (n))
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
148 #endif
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
149
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
150 /* A hash value, suitable for computing hash tables. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
151 typedef __re_size_t re_hashval_t;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
152
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
153 /* An integer used to represent a set of bits. It must be unsigned,
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
154 and must be at least as wide as unsigned int. */
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
155 typedef unsigned long int bitset_word_t;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
156 /* All bits set in a bitset_word_t. */
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
157 #define BITSET_WORD_MAX ULONG_MAX
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
158
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
159 /* Number of bits in a bitset_word_t. For portability to hosts with
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
160 padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)';
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
161 instead, deduce it directly from BITSET_WORD_MAX. Avoid
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
162 greater-than-32-bit integers and unconditional shifts by more than
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
163 31 bits, as they're not portable. */
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
164 #if BITSET_WORD_MAX == 0xffffffff
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
165 # define BITSET_WORD_BITS 32
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
166 #elif BITSET_WORD_MAX >> 31 >> 5 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
167 # define BITSET_WORD_BITS 36
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
168 #elif BITSET_WORD_MAX >> 31 >> 16 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
169 # define BITSET_WORD_BITS 48
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
170 #elif BITSET_WORD_MAX >> 31 >> 28 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
171 # define BITSET_WORD_BITS 60
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
172 #elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
173 # define BITSET_WORD_BITS 64
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
174 #elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
175 # define BITSET_WORD_BITS 72
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
176 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
177 # define BITSET_WORD_BITS 128
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
178 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
179 # define BITSET_WORD_BITS 256
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
180 #elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
181 # define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
182 # if BITSET_WORD_BITS <= SBC_MAX
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
183 # error "Invalid SBC_MAX"
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
184 # endif
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
185 #elif BITSET_WORD_MAX == (0xffffffff + 2) * 0xffffffff
6539
5a1c91276525 * regex_internal.h (BITSET_WORD_BITS):
Paul Eggert <eggert@cs.ucla.edu>
parents: 6296
diff changeset
186 /* Work around a bug in 64-bit PGC (before version 6.1-2), where the
5a1c91276525 * regex_internal.h (BITSET_WORD_BITS):
Paul Eggert <eggert@cs.ucla.edu>
parents: 6296
diff changeset
187 preprocessor mishandles large unsigned values as if they were signed. */
5a1c91276525 * regex_internal.h (BITSET_WORD_BITS):
Paul Eggert <eggert@cs.ucla.edu>
parents: 6296
diff changeset
188 # define BITSET_WORD_BITS 64
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
189 #else
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
190 # error "Add case for new bitset_word_t size"
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
191 #endif
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
192
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
193 /* Number of bitset_word_t values in a bitset_t. */
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
194 #define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS)
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
195
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
196 typedef bitset_word_t bitset_t[BITSET_WORDS];
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
197 typedef bitset_word_t *re_bitset_ptr_t;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
198 typedef const bitset_word_t *re_const_bitset_ptr_t;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
199
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
200 #define PREV_WORD_CONSTRAINT 0x0001
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
201 #define PREV_NOTWORD_CONSTRAINT 0x0002
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
202 #define NEXT_WORD_CONSTRAINT 0x0004
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
203 #define NEXT_NOTWORD_CONSTRAINT 0x0008
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
204 #define PREV_NEWLINE_CONSTRAINT 0x0010
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
205 #define NEXT_NEWLINE_CONSTRAINT 0x0020
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
206 #define PREV_BEGBUF_CONSTRAINT 0x0040
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
207 #define NEXT_ENDBUF_CONSTRAINT 0x0080
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
208 #define WORD_DELIM_CONSTRAINT 0x0100
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
209 #define NOT_WORD_DELIM_CONSTRAINT 0x0200
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
210
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
211 typedef enum
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
212 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
213 INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
214 WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
215 WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
216 INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
217 LINE_FIRST = PREV_NEWLINE_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
218 LINE_LAST = NEXT_NEWLINE_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
219 BUF_FIRST = PREV_BEGBUF_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
220 BUF_LAST = NEXT_ENDBUF_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
221 WORD_DELIM = WORD_DELIM_CONSTRAINT,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
222 NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
223 } re_context_type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
224
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
225 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
226 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
227 Idx alloc;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
228 Idx nelem;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
229 Idx *elems;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
230 } re_node_set;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
231
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
232 typedef enum
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
233 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
234 NON_TYPE = 0,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
235
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
236 /* Node type, These are used by token, node, tree. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
237 CHARACTER = 1,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
238 END_OF_RE = 2,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
239 SIMPLE_BRACKET = 3,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
240 OP_BACK_REF = 4,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
241 OP_PERIOD = 5,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
242 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
243 COMPLEX_BRACKET = 6,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
244 OP_UTF8_PERIOD = 7,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
245 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
246
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
247 /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
248 when the debugger shows values of this enum type. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
249 #define EPSILON_BIT 8
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
250 OP_OPEN_SUBEXP = EPSILON_BIT | 0,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
251 OP_CLOSE_SUBEXP = EPSILON_BIT | 1,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
252 OP_ALT = EPSILON_BIT | 2,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
253 OP_DUP_ASTERISK = EPSILON_BIT | 3,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
254 ANCHOR = EPSILON_BIT | 4,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
255
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
256 /* Tree type, these are used only by tree. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
257 CONCAT = 16,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
258 SUBEXP = 17,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
259
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
260 /* Token type, these are used only by token. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
261 OP_DUP_PLUS = 18,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
262 OP_DUP_QUESTION,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
263 OP_OPEN_BRACKET,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
264 OP_CLOSE_BRACKET,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
265 OP_CHARSET_RANGE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
266 OP_OPEN_DUP_NUM,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
267 OP_CLOSE_DUP_NUM,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
268 OP_NON_MATCH_LIST,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
269 OP_OPEN_COLL_ELEM,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
270 OP_CLOSE_COLL_ELEM,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
271 OP_OPEN_EQUIV_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
272 OP_CLOSE_EQUIV_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
273 OP_OPEN_CHAR_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
274 OP_CLOSE_CHAR_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
275 OP_WORD,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
276 OP_NOTWORD,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
277 OP_SPACE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
278 OP_NOTSPACE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
279 BACK_SLASH
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
280
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
281 } re_token_type_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
282
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
283 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
284 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
285 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
286 /* Multibyte characters. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
287 wchar_t *mbchars;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
288
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
289 /* Collating symbols. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
290 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
291 int32_t *coll_syms;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
292 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
293
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
294 /* Equivalence classes. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
295 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
296 int32_t *equiv_classes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
297 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
298
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
299 /* Range expressions. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
300 # ifdef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
301 uint32_t *range_starts;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
302 uint32_t *range_ends;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
303 # else /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
304 wchar_t *range_starts;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
305 wchar_t *range_ends;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
306 # endif /* not _LIBC */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
307
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
308 /* Character classes. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
309 wctype_t *char_classes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
310
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
311 /* If this character set is the non-matching list. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
312 unsigned int non_match : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
313
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
314 /* # of multibyte characters. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
315 Idx nmbchars;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
316
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
317 /* # of collating symbols. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
318 Idx ncoll_syms;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
319
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
320 /* # of equivalence classes. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
321 Idx nequiv_classes;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
322
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
323 /* # of range expressions. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
324 Idx nranges;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
325
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
326 /* # of character classes. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
327 Idx nchar_classes;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
328 } re_charset_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
329 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
330
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
331 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
332 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
333 union
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
334 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
335 unsigned char c; /* for CHARACTER */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
336 re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
337 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
338 re_charset_t *mbcset; /* for COMPLEX_BRACKET */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
339 #endif /* RE_ENABLE_I18N */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
340 Idx idx; /* for BACK_REF */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
341 re_context_type ctx_type; /* for ANCHOR */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
342 } opr;
7689
b4806b1ec9e4 Fix some incompatibilities with gcc -ansi -pedantic.
Paul Eggert <eggert@cs.ucla.edu>
parents: 7162
diff changeset
343 #if __GNUC__ >= 2 && !__STRICT_ANSI__
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
344 re_token_type_t type : 8;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
345 #else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
346 re_token_type_t type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
347 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
348 unsigned int constraint : 10; /* context constraint */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
349 unsigned int duplicated : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
350 unsigned int opt_subexp : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
351 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
352 unsigned int accept_mb : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
353 /* These 2 bits can be moved into the union if needed (e.g. if running out
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
354 of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
355 unsigned int mb_partial : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
356 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
357 unsigned int word_char : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
358 } re_token_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
359
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
360 #define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
361
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
362 struct re_string_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
363 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
364 /* Indicate the raw buffer which is the original string passed as an
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
365 argument of regexec(), re_search(), etc.. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
366 const unsigned char *raw_mbs;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
367 /* Store the multibyte string. In case of "case insensitive mode" like
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
368 REG_ICASE, upper cases of the string are stored, otherwise MBS points
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
369 the same address that RAW_MBS points. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
370 unsigned char *mbs;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
371 #ifdef RE_ENABLE_I18N
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
372 /* Store the wide character string which is corresponding to MBS. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
373 wint_t *wcs;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
374 Idx *offsets;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
375 mbstate_t cur_state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
376 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
377 /* Index in RAW_MBS. Each character mbs[i] corresponds to
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
378 raw_mbs[raw_mbs_idx + i]. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
379 Idx raw_mbs_idx;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
380 /* The length of the valid characters in the buffers. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
381 Idx valid_len;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
382 /* The corresponding number of bytes in raw_mbs array. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
383 Idx valid_raw_len;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
384 /* The length of the buffers MBS and WCS. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
385 Idx bufs_len;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
386 /* The index in MBS, which is updated by re_string_fetch_byte. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
387 Idx cur_idx;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
388 /* length of RAW_MBS array. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
389 Idx raw_len;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
390 /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
391 Idx len;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
392 /* End of the buffer may be shorter than its length in the cases such
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
393 as re_match_2, re_search_2. Then, we use STOP for end of the buffer
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
394 instead of LEN. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
395 Idx raw_stop;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
396 /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
397 Idx stop;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
398
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
399 /* The context of mbs[0]. We store the context independently, since
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
400 the context of mbs[0] may be different from raw_mbs[0], which is
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
401 the beginning of the input string. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
402 unsigned int tip_context;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
403 /* The translation passed as a part of an argument of re_compile_pattern. */
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
404 RE_TRANSLATE_TYPE trans;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
405 /* Copy of re_dfa_t's word_char. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
406 re_const_bitset_ptr_t word_char;
6195
25eaa608fc4e Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6184
diff changeset
407 /* true if REG_ICASE. */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
408 unsigned char icase;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
409 unsigned char is_utf8;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
410 unsigned char map_notascii;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
411 unsigned char mbs_allocated;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
412 unsigned char offsets_needed;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
413 unsigned char newline_anchor;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
414 unsigned char word_ops_used;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
415 int mb_cur_max;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
416 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
417 typedef struct re_string_t re_string_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
418
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
419
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
420 struct re_dfa_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
421 typedef struct re_dfa_t re_dfa_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
422
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
423 #ifndef _LIBC
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
424 # ifdef __i386__
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
425 # define internal_function __attribute ((regparm (3), stdcall))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
426 # else
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
427 # define internal_function
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
428 # endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
429 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
430
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
431 static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr,
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
432 Idx new_buf_len)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
433 internal_function;
6076
e2dd51f6e259 * config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents: 5968
diff changeset
434 #ifdef RE_ENABLE_I18N
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
435 static void build_wcs_buffer (re_string_t *pstr) internal_function;
6119
c3bf2ea44695 Make regex safe for g++. This fixes one real bug (an "err"
Paul Eggert <eggert@cs.ucla.edu>
parents: 6101
diff changeset
436 static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr)
c3bf2ea44695 Make regex safe for g++. This fixes one real bug (an "err"
Paul Eggert <eggert@cs.ucla.edu>
parents: 6101
diff changeset
437 internal_function;
6076
e2dd51f6e259 * config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents: 5968
diff changeset
438 #endif /* RE_ENABLE_I18N */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
439 static void build_upper_buffer (re_string_t *pstr) internal_function;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
440 static void re_string_translate_buffer (re_string_t *pstr) internal_function;
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
441 static unsigned int re_string_context_at (const re_string_t *input, Idx idx,
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
442 int eflags)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
443 internal_function __attribute ((pure));
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
444 #define re_string_peek_byte(pstr, offset) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
445 ((pstr)->mbs[(pstr)->cur_idx + offset])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
446 #define re_string_fetch_byte(pstr) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
447 ((pstr)->mbs[(pstr)->cur_idx++])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
448 #define re_string_first_byte(pstr, idx) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
449 ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
450 #define re_string_is_single_byte_char(pstr, idx) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
451 ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
452 || (pstr)->wcs[(idx) + 1] != WEOF))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
453 #define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
454 #define re_string_cur_idx(pstr) ((pstr)->cur_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
455 #define re_string_get_buffer(pstr) ((pstr)->mbs)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
456 #define re_string_length(pstr) ((pstr)->len)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
457 #define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx])
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
458 #define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
459 #define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
460
6125
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
461 #include <alloca.h>
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
462
6126
3690c96b6da9 Fix typo: __LIBC -> _LIBC.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6125
diff changeset
463 #ifndef _LIBC
6125
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
464 # if HAVE_ALLOCA
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
465 /* The OS usually guarantees only one guard page at the bottom of the stack,
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
466 and a page size can be as small as 4096 bytes. So we cannot safely
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
467 allocate anything larger than 4096 bytes. Also care for the possibility
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
468 of a few compiler-allocated temporary stack slots. */
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
469 # define __libc_use_alloca(n) ((n) < 4032)
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
470 # else
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
471 /* alloca is implemented with malloc, so just use malloc. */
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
472 # define __libc_use_alloca(n) 0
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
473 # endif
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
474 #endif
4af02335816a * config/srclist.txt: Add glibc bug 1245.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6119
diff changeset
475
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
476 #ifndef MAX
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
477 # define MAX(a,b) ((a) < (b) ? (b) : (a))
6206
ca2f5d46eeb6 Check for arithmetic overflow when calculating sizes, to prevent
Paul Eggert <eggert@cs.ucla.edu>
parents: 6195
diff changeset
478 #endif
ca2f5d46eeb6 Check for arithmetic overflow when calculating sizes, to prevent
Paul Eggert <eggert@cs.ucla.edu>
parents: 6195
diff changeset
479
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
480 #define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t)))
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
481 #define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t)))
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
482 #define re_free(p) free (p)
6206
ca2f5d46eeb6 Check for arithmetic overflow when calculating sizes, to prevent
Paul Eggert <eggert@cs.ucla.edu>
parents: 6195
diff changeset
483
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
484 struct bin_tree_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
485 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
486 struct bin_tree_t *parent;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
487 struct bin_tree_t *left;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
488 struct bin_tree_t *right;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
489 struct bin_tree_t *first;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
490 struct bin_tree_t *next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
491
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
492 re_token_t token;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
493
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
494 /* `node_idx' is the index in dfa->nodes, if `type' == 0.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
495 Otherwise `type' indicate the type of this node. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
496 Idx node_idx;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
497 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
498 typedef struct bin_tree_t bin_tree_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
499
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
500 #define BIN_TREE_STORAGE_SIZE \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
501 ((1024 - sizeof (void *)) / sizeof (bin_tree_t))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
502
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
503 struct bin_tree_storage_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
504 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
505 struct bin_tree_storage_t *next;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
506 bin_tree_t data[BIN_TREE_STORAGE_SIZE];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
507 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
508 typedef struct bin_tree_storage_t bin_tree_storage_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
509
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
510 #define CONTEXT_WORD 1
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
511 #define CONTEXT_NEWLINE (CONTEXT_WORD << 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
512 #define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
513 #define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
514
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
515 #define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
516 #define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
517 #define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
518 #define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
519 #define IS_ORDINARY_CONTEXT(c) ((c) == 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
520
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
521 #define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
522 #define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
523 #define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_')
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
524 #define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
525
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
526 #define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
527 ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
528 || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
529 || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
530 || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
531
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
532 #define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
533 ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
534 || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
535 || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
536 || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context)))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
537
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
538 struct re_dfastate_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
539 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
540 re_hashval_t hash;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
541 re_node_set nodes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
542 re_node_set non_eps_nodes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
543 re_node_set inveclosure;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
544 re_node_set *entrance_nodes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
545 struct re_dfastate_t **trtable, **word_trtable;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
546 unsigned int context : 4;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
547 unsigned int halt : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
548 /* If this state can accept `multi byte'.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
549 Note that we refer to multibyte characters, and multi character
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
550 collating elements as `multi byte'. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
551 unsigned int accept_mb : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
552 /* If this state has backreference node(s). */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
553 unsigned int has_backref : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
554 unsigned int has_constraint : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
555 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
556 typedef struct re_dfastate_t re_dfastate_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
557
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
558 struct re_state_table_entry
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
559 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
560 Idx num;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
561 Idx alloc;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
562 re_dfastate_t **array;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
563 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
564
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
565 /* Array type used in re_sub_match_last_t and re_sub_match_top_t. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
566
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
567 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
568 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
569 Idx next_idx;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
570 Idx alloc;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
571 re_dfastate_t **array;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
572 } state_array_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
573
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
574 /* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
575
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
576 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
577 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
578 Idx node;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
579 Idx str_idx; /* The position NODE match at. */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
580 state_array_t path;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
581 } re_sub_match_last_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
582
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
583 /* Store information about the node NODE whose type is OP_OPEN_SUBEXP.
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
584 And information about the node, whose type is OP_CLOSE_SUBEXP,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
585 corresponding to NODE is stored in LASTS. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
586
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
587 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
588 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
589 Idx str_idx;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
590 Idx node;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
591 state_array_t *path;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
592 Idx alasts; /* Allocation size of LASTS. */
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
593 Idx nlasts; /* The number of LASTS. */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
594 re_sub_match_last_t **lasts;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
595 } re_sub_match_top_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
596
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
597 struct re_backref_cache_entry
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
598 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
599 Idx node;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
600 Idx str_idx;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
601 Idx subexp_from;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
602 Idx subexp_to;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
603 char more;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
604 char unused;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
605 unsigned short int eps_reachable_subexps_map;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
606 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
607
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
608 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
609 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
610 /* The string object corresponding to the input string. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
611 re_string_t input;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
612 #if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
613 const re_dfa_t *const dfa;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
614 #else
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
615 const re_dfa_t *dfa;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
616 #endif
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
617 /* EFLAGS of the argument of regexec. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
618 int eflags;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
619 /* Where the matching ends. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
620 Idx match_last;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
621 Idx last_node;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
622 /* The state log used by the matcher. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
623 re_dfastate_t **state_log;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
624 Idx state_log_top;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
625 /* Back reference cache. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
626 Idx nbkref_ents;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
627 Idx abkref_ents;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
628 struct re_backref_cache_entry *bkref_ents;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
629 int max_mb_elem_len;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
630 Idx nsub_tops;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
631 Idx asub_tops;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
632 re_sub_match_top_t **sub_tops;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
633 } re_match_context_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
634
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
635 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
636 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
637 re_dfastate_t **sifted_states;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
638 re_dfastate_t **limited_states;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
639 Idx last_node;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
640 Idx last_str_idx;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
641 re_node_set limits;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
642 } re_sift_context_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
643
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
644 struct re_fail_stack_ent_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
645 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
646 Idx idx;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
647 Idx node;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
648 regmatch_t *regs;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
649 re_node_set eps_via_nodes;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
650 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
651
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
652 struct re_fail_stack_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
653 {
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
654 Idx num;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
655 Idx alloc;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
656 struct re_fail_stack_ent_t *stack;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
657 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
658
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
659 struct re_dfa_t
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
660 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
661 re_token_t *nodes;
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
662 size_t nodes_alloc;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
663 size_t nodes_len;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
664 Idx *nexts;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
665 Idx *org_indices;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
666 re_node_set *edests;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
667 re_node_set *eclosures;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
668 re_node_set *inveclosures;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
669 struct re_state_table_entry *state_table;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
670 re_dfastate_t *init_state;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
671 re_dfastate_t *init_state_word;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
672 re_dfastate_t *init_state_nl;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
673 re_dfastate_t *init_state_begbuf;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
674 bin_tree_t *str_tree;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
675 bin_tree_storage_t *str_tree_storage;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
676 re_bitset_ptr_t sb_char;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
677 int str_tree_storage_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
678
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
679 /* number of subexpressions `re_nsub' is in regex_t. */
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
680 re_hashval_t state_hash_mask;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
681 Idx init_node;
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
682 Idx nbackref; /* The number of backreference in this dfa. */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
683
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
684 /* Bitmap expressing which backreference is used. */
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
685 bitset_word_t used_bkref_map;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
686 bitset_word_t completed_bkref_map;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
687
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
688 unsigned int has_plural_match : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
689 /* If this dfa has "multibyte node", which is a backreference or
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
690 a node which can accept multibyte character or multi character
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
691 collating element. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
692 unsigned int has_mb_node : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
693 unsigned int is_utf8 : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
694 unsigned int map_notascii : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
695 unsigned int word_ops_used : 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
696 int mb_cur_max;
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
697 bitset_t word_char;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
698 reg_syntax_t syntax;
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
699 Idx *subexp_map;
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
700 #ifdef DEBUG
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
701 char* re_str;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
702 #endif
7066
3288bb324a28 * regex_internal.h (struct re_dfa_t) [!_LIBC]: Avoid invalid C89
Eric Blake <ebb9@byu.net>
parents: 6932
diff changeset
703 #ifdef _LIBC
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
704 __libc_lock_define (, lock)
7066
3288bb324a28 * regex_internal.h (struct re_dfa_t) [!_LIBC]: Avoid invalid C89
Eric Blake <ebb9@byu.net>
parents: 6932
diff changeset
705 #endif
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
706 };
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
707
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
708 #define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
709 #define re_node_set_remove(set,id) \
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
710 (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1))
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
711 #define re_node_set_empty(p) ((p)->nelem = 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
712 #define re_node_set_free(set) re_free ((set)->elems)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
713
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
714
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
715 typedef enum
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
716 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
717 SB_CHAR,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
718 MB_CHAR,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
719 EQUIV_CLASS,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
720 COLL_SYM,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
721 CHAR_CLASS
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
722 } bracket_elem_type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
723
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
724 typedef struct
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
725 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
726 bracket_elem_type type;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
727 union
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
728 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
729 unsigned char ch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
730 unsigned char *name;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
731 wchar_t wch;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
732 } opr;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
733 } bracket_elem_t;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
734
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
735
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
736 /* Inline functions for bitset_t operation. */
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
737
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
738 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
739 bitset_set (bitset_t set, Idx i)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
740 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
741 set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS;
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
742 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
743
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
744 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
745 bitset_clear (bitset_t set, Idx i)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
746 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
747 set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS);
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
748 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
749
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
750 static inline bool
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
751 bitset_contain (const bitset_t set, Idx i)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
752 {
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
753 return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1;
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
754 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
755
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
756 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
757 bitset_empty (bitset_t set)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
758 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
759 memset (set, '\0', sizeof (bitset_t));
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
760 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
761
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
762 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
763 bitset_set_all (bitset_t set)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
764 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
765 memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS));
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
766 if (SBC_MAX % BITSET_WORD_BITS != 0)
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
767 set[BITSET_WORDS - 1] =
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
768 ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1;
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
769 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
770
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
771 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
772 bitset_copy (bitset_t dest, const bitset_t src)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
773 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
774 memcpy (dest, src, sizeof (bitset_t));
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
775 }
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
776
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
777 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
778 bitset_not (bitset_t set)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
779 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
780 int bitset_i;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
781 for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i)
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
782 set[bitset_i] = ~set[bitset_i];
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
783 if (SBC_MAX % BITSET_WORD_BITS != 0)
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
784 set[BITSET_WORDS - 1] =
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
785 ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1)
6214
afb93b90dcb8 Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents: 6206
diff changeset
786 & ~set[BITSET_WORDS - 1]);
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
787 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
788
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
789 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
790 bitset_merge (bitset_t dest, const bitset_t src)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
791 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
792 int bitset_i;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
793 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
794 dest[bitset_i] |= src[bitset_i];
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
795 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
796
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
797 static inline void
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
798 bitset_mask (bitset_t dest, const bitset_t src)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
799 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
800 int bitset_i;
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
801 for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i)
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
802 dest[bitset_i] &= src[bitset_i];
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
803 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
804
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
805 #ifdef RE_ENABLE_I18N
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
806 /* Inline functions for re_string. */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
807 static inline int
6129
2fce0cb93f59 * config/srclist.text: Add glibc bug 1248.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6126
diff changeset
808 internal_function __attribute ((pure))
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
809 re_string_char_size_at (const re_string_t *pstr, Idx idx)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
810 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
811 int byte_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
812 if (pstr->mb_cur_max == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
813 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
814 for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
815 if (pstr->wcs[idx + byte_idx] != WEOF)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
816 break;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
817 return byte_idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
818 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
819
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
820 static inline wint_t
6129
2fce0cb93f59 * config/srclist.text: Add glibc bug 1248.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6126
diff changeset
821 internal_function __attribute ((pure))
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
822 re_string_wchar_at (const re_string_t *pstr, Idx idx)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
823 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
824 if (pstr->mb_cur_max == 1)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
825 return (wint_t) pstr->mbs[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
826 return (wint_t) pstr->wcs[idx];
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
827 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
828
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
829 static int
6129
2fce0cb93f59 * config/srclist.text: Add glibc bug 1248.
Paul Eggert <eggert@cs.ucla.edu>
parents: 6126
diff changeset
830 internal_function __attribute ((pure))
6184
f1728546eca4 On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents: 6171
diff changeset
831 re_string_elem_size_at (const re_string_t *pstr, Idx idx)
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
832 {
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
833 # ifdef _LIBC
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
834 const unsigned char *p, *extra;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
835 const int32_t *table, *indirect;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
836 int32_t tmp;
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
837 # include <locale/weight.h>
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
838 uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
839
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
840 if (nrules != 0)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
841 {
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
842 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
843 extra = (const unsigned char *)
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
844 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
845 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE,
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
846 _NL_COLLATE_INDIRECTMB);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
847 p = pstr->mbs + idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
848 tmp = findidx (&p);
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
849 return p - pstr->mbs - idx;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
850 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
851 else
6726
af9abbcedfbd Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents: 6539
diff changeset
852 # endif /* _LIBC */
5968
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
853 return 1;
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
854 }
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
855 #endif /* RE_ENABLE_I18N */
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
856
541fed6ae301 * modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff changeset
857 #endif /* _REGEX_INTERNAL_H */