annotate lib/unicodeio.c @ 3843:2f0e1c3ba75b

*** empty log message ***
author Jim Meyering <jim@meyering.net>
date Wed, 22 May 2002 08:36:32 +0000
parents f80e972daf0f
children 22d3032f0239
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
1 /* Unicode character output to streams with locale dependent encoding.
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
2
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
3 Copyright (C) 2000-2002 Free Software Foundation, Inc.
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
4
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
5 This program is free software; you can redistribute it and/or modify it
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
6 under the terms of the GNU Library General Public License as published
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
7 by the Free Software Foundation; either version 2, or (at your option)
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
8 any later version.
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
9
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
10 This program is distributed in the hope that it will be useful,
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
13 Library General Public License for more details.
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
14
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
15 You should have received a copy of the GNU Library General Public
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
16 License along with this program; if not, write to the Free Software
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
17 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
18 USA. */
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
19
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
20 /* Written by Bruno Haible <haible@clisp.cons.org>. */
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
21
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
22 /* Note: This file requires the locale_charset() function. See in
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
23 libiconv-1.7/libcharset/INTEGRATE for how to obtain it. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
24
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
25 #ifdef HAVE_CONFIG_H
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
26 # include <config.h>
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
27 #endif
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
28
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
29 #if HAVE_STDDEF_H
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
30 # include <stddef.h>
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
31 #endif
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
32
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
33 #include <stdio.h>
2352
84a00de8c81b (<string.h>): Include only #if HAVE_STRING_H.
Jim Meyering <jim@meyering.net>
parents: 2351
diff changeset
34 #if HAVE_STRING_H
84a00de8c81b (<string.h>): Include only #if HAVE_STRING_H.
Jim Meyering <jim@meyering.net>
parents: 2351
diff changeset
35 # include <string.h>
84a00de8c81b (<string.h>): Include only #if HAVE_STRING_H.
Jim Meyering <jim@meyering.net>
parents: 2351
diff changeset
36 #else
84a00de8c81b (<string.h>): Include only #if HAVE_STRING_H.
Jim Meyering <jim@meyering.net>
parents: 2351
diff changeset
37 # include <strings.h>
84a00de8c81b (<string.h>): Include only #if HAVE_STRING_H.
Jim Meyering <jim@meyering.net>
parents: 2351
diff changeset
38 #endif
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
39
2329
add6290501b6 (<errno.h>): Include it.
Jim Meyering <jim@meyering.net>
parents: 2305
diff changeset
40 #include <errno.h>
add6290501b6 (<errno.h>): Include it.
Jim Meyering <jim@meyering.net>
parents: 2305
diff changeset
41 #ifndef errno
add6290501b6 (<errno.h>): Include it.
Jim Meyering <jim@meyering.net>
parents: 2305
diff changeset
42 extern int errno;
add6290501b6 (<errno.h>): Include it.
Jim Meyering <jim@meyering.net>
parents: 2305
diff changeset
43 #endif
2969
6152b49dde12 Include wchar.h.
Jim Meyering <jim@meyering.net>
parents: 2650
diff changeset
44
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
45 #if HAVE_ICONV
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
46 # include <iconv.h>
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
47 #endif
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
48
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
49 #include <error.h>
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
50
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
51 #if ENABLE_NLS
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
52 # include <libintl.h>
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
53 #else
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
54 # define gettext(Text) Text
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
55 #endif
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
56 #define _(Text) gettext (Text)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
57 #define N_(Text) Text
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
58
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
59 /* Specification. */
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
60 #include "unicodeio.h"
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
61
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
62 /* When we pass a Unicode character to iconv(), we must pass it in a
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
63 suitable encoding. The standardized Unicode encodings are
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
64 UTF-8, UCS-2, UCS-4, UTF-16, UTF-16BE, UTF-16LE, UTF-7.
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
65 UCS-2 supports only characters up to \U0000FFFF.
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
66 UTF-16 and variants support only characters up to \U0010FFFF.
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
67 UTF-7 is way too complex and not supported by glibc-2.1.
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
68 UCS-4 specification leaves doubts about endianness and byte order
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
69 mark. glibc currently interprets it as big endian without byte order
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
70 mark, but this is not backed by an RFC.
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
71 So we use UTF-8. It supports characters up to \U7FFFFFFF and is
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
72 unambiguously defined. */
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
73
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
74 /* Stores the UTF-8 representation of the Unicode character wc in r[0..5].
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
75 Returns the number of bytes stored, or -1 if wc is out of range. */
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
76 static int
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
77 utf8_wctomb (unsigned char *r, unsigned int wc)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
78 {
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
79 int count;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
80
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
81 if (wc < 0x80)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
82 count = 1;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
83 else if (wc < 0x800)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
84 count = 2;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
85 else if (wc < 0x10000)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
86 count = 3;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
87 else if (wc < 0x200000)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
88 count = 4;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
89 else if (wc < 0x4000000)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
90 count = 5;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
91 else if (wc <= 0x7fffffff)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
92 count = 6;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
93 else
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
94 return -1;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
95
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
96 switch (count)
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
97 {
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
98 /* Note: code falls through cases! */
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
99 case 6: r[5] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x4000000;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
100 case 5: r[4] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x200000;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
101 case 4: r[3] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x10000;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
102 case 3: r[2] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0x800;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
103 case 2: r[1] = 0x80 | (wc & 0x3f); wc = wc >> 6; wc |= 0xc0;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
104 case 1: r[0] = wc;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
105 }
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
106
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
107 return count;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
108 }
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
109
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
110 /* Luckily, the encoding's name is platform independent. */
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
111 #define UTF8_NAME "UTF-8"
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
112
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
113 /* Converts the Unicode character CODE to its multibyte representation
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
114 in the current locale and calls the SUCCESS callback on the resulting
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
115 byte sequence. If an error occurs, invokes the FAILURE callback instead,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
116 passing it CODE and an English error string.
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
117 Returns whatever the callback returned.
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
118 Assumes that the locale doesn't change between two calls. */
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
119 long
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
120 unicode_to_mb (unsigned int code,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
121 long (*success) PARAMS ((const char *buf, size_t buflen,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
122 void *callback_arg)),
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
123 long (*failure) PARAMS ((unsigned int code, const char *msg,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
124 void *callback_arg)),
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
125 void *callback_arg)
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
126 {
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
127 static int initialized;
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
128 static int is_utf8;
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
129 #if HAVE_ICONV
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
130 static iconv_t utf8_to_local;
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
131 #endif
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
132
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
133 char inbuf[6];
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
134 int count;
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
135
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
136 if (!initialized)
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
137 {
2650
ba5bbff44a5b (print_unicode_char): Work around ansi2knr deficiency.
Jim Meyering <jim@meyering.net>
parents: 2412
diff changeset
138 extern const char *locale_charset PARAMS ((void));
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
139 const char *charset = locale_charset ();
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
140
3202
283d780ab032 (print_unicode_char): Simplify accordingly.
Jim Meyering <jim@meyering.net>
parents: 3123
diff changeset
141 is_utf8 = !strcmp (charset, UTF8_NAME);
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
142 #if HAVE_ICONV
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
143 if (!is_utf8)
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
144 {
3202
283d780ab032 (print_unicode_char): Simplify accordingly.
Jim Meyering <jim@meyering.net>
parents: 3123
diff changeset
145 utf8_to_local = iconv_open (charset, UTF8_NAME);
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
146 if (utf8_to_local == (iconv_t)(-1))
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
147 /* For an unknown encoding, assume ASCII. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
148 utf8_to_local = iconv_open ("ASCII", UTF8_NAME);
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
149 }
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
150 #endif
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
151 initialized = 1;
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
152 }
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
153
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
154 /* Test whether the utf8_to_local converter is available at all. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
155 if (!is_utf8)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
156 {
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
157 #if HAVE_ICONV
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
158 if (utf8_to_local == (iconv_t)(-1))
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
159 return failure (code, N_("iconv function not usable"), callback_arg);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
160 #else
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
161 return failure (code, N_("iconv function not available"), callback_arg);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
162 #endif
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
163 }
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
164
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
165 /* Convert the character to UTF-8. */
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
166 count = utf8_wctomb ((unsigned char *) inbuf, code);
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
167 if (count < 0)
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
168 return failure (code, N_("character out of range"), callback_arg);
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
169
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
170 #if HAVE_ICONV
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
171 if (!is_utf8)
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
172 {
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
173 char outbuf[25];
3123
f65bb2158462 (print_unicode_char): Cast the second iconv() arg,
Jim Meyering <jim@meyering.net>
parents: 3090
diff changeset
174 const char *inptr;
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
175 size_t inbytesleft;
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
176 char *outptr;
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
177 size_t outbytesleft;
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
178 size_t res;
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
179
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
180 inptr = inbuf;
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
181 inbytesleft = count;
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
182 outptr = outbuf;
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
183 outbytesleft = sizeof (outbuf);
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
184
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
185 /* Convert the character from UTF-8 to the locale's charset. */
3123
f65bb2158462 (print_unicode_char): Cast the second iconv() arg,
Jim Meyering <jim@meyering.net>
parents: 3090
diff changeset
186 res = iconv (utf8_to_local,
f65bb2158462 (print_unicode_char): Cast the second iconv() arg,
Jim Meyering <jim@meyering.net>
parents: 3090
diff changeset
187 (ICONV_CONST char **)&inptr, &inbytesleft,
f65bb2158462 (print_unicode_char): Cast the second iconv() arg,
Jim Meyering <jim@meyering.net>
parents: 3090
diff changeset
188 &outptr, &outbytesleft);
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
189 if (inbytesleft > 0 || res == (size_t)(-1)
2412
00595b9008f3 (print_unicode_char): Avoid triggering Solaris iconv bug.
Jim Meyering <jim@meyering.net>
parents: 2352
diff changeset
190 /* Irix iconv() inserts a NUL byte if it cannot convert. */
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
191 # if !defined _LIBICONV_VERSION && (defined sgi || defined __sgi)
2412
00595b9008f3 (print_unicode_char): Avoid triggering Solaris iconv bug.
Jim Meyering <jim@meyering.net>
parents: 2352
diff changeset
192 || (res > 0 && code != 0 && outptr - outbuf == 1 && *outbuf == '\0')
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
193 # endif
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
194 )
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
195 return failure (code, NULL, callback_arg);
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
196
2412
00595b9008f3 (print_unicode_char): Avoid triggering Solaris iconv bug.
Jim Meyering <jim@meyering.net>
parents: 2352
diff changeset
197 /* Avoid glibc-2.1 bug and Solaris 2.7 bug. */
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
198 # if defined _LIBICONV_VERSION \
2412
00595b9008f3 (print_unicode_char): Avoid triggering Solaris iconv bug.
Jim Meyering <jim@meyering.net>
parents: 2352
diff changeset
199 || !((__GLIBC__ - 0 == 2 && __GLIBC_MINOR__ - 0 <= 1) || defined __sun)
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
200
2351
f0f61b4d7d83 (utf8_wctomb): New function.
Jim Meyering <jim@meyering.net>
parents: 2329
diff changeset
201 /* Get back to the initial shift state. */
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
202 res = iconv (utf8_to_local, NULL, NULL, &outptr, &outbytesleft);
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
203 if (res == (size_t)(-1))
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
204 return failure (code, NULL, callback_arg);
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
205 # endif
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
206
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
207 return success (outbuf, outptr - outbuf, callback_arg);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
208 }
2983
4dbece6c259b Back out Paul's changes while we wait for him and Bruno to reach agreement.
Jim Meyering <jim@meyering.net>
parents: 2977
diff changeset
209 #endif
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
210
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
211 /* At this point, is_utf8 is true, so no conversion is needed. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
212 return success (inbuf, count, callback_arg);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
213 }
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
214
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
215 /* Simple success callback that outputs the converted string.
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
216 The STREAM is passed as callback_arg. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
217 long
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
218 fwrite_success_callback (const char *buf, size_t buflen, void *callback_arg)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
219 {
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
220 FILE *stream = (FILE *) callback_arg;
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
221
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
222 fwrite (buf, 1, buflen, stream);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
223 return 0;
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
224 }
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
225
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
226 /* Simple failure callback that displays an error and exits. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
227 static long
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
228 exit_failure_callback (unsigned int code, const char *msg, void *callback_arg)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
229 {
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
230 if (msg == NULL)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
231 error (1, 0, _("cannot convert U+%04X to local character set"), code);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
232 else
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
233 error (1, 0, _("cannot convert U+%04X to local character set: %s"), code,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
234 gettext (msg));
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
235 return -1;
2305
463a686cf218 New file from Bruno.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
236 }
3727
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
237
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
238 /* Simple failure callback that displays a fallback representation in plain
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
239 ASCII, using the same notation as ISO C99 strings. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
240 static long
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
241 fallback_failure_callback (unsigned int code, const char *msg, void *callback_arg)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
242 {
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
243 FILE *stream = (FILE *) callback_arg;
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
244
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
245 if (code < 0x10000)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
246 fprintf (stream, "\\u%04X", code);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
247 else
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
248 fprintf (stream, "\\U%08X", code);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
249 return -1;
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
250 }
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
251
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
252 /* Outputs the Unicode character CODE to the output stream STREAM.
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
253 Upon failure, exit if exit_on_error is true, otherwise output a fallback
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
254 notation. */
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
255 void
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
256 print_unicode_char (FILE *stream, unsigned int code, int exit_on_error)
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
257 {
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
258 unicode_to_mb (code, fwrite_success_callback,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
259 exit_on_error
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
260 ? exit_failure_callback
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
261 : fallback_failure_callback,
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
262 stream);
f80e972daf0f 2002-02-02 Paul Eggert <eggert@twinsun.com>
Jim Meyering <jim@meyering.net>
parents: 3202
diff changeset
263 }