Mercurial > sdl-ios-xcode

annotate src/stdlib/SDL_iconv.c @ 4013:5804139bdf0b SDL-1.2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Use GL direct updates when possible
author Sam Lantinga <slouken@libsdl.org>
date Sat, 07 Jul 2007 19:22:27 +0000
parents 098ac044cd2f
children 0aadbc81c497
rev   line source
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
1 /*
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
2 SDL - Simple DirectMedia Layer
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
3 Copyright (C) 1997-2006 Sam Lantinga
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
4
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
5 This library is free software; you can redistribute it and/or
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
6 modify it under the terms of the GNU Lesser General Public
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
7 License as published by the Free Software Foundation; either
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
8 version 2.1 of the License, or (at your option) any later version.
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
9
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
10 This library is distributed in the hope that it will be useful,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
13 Lesser General Public License for more details.
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
14
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
15 You should have received a copy of the GNU Lesser General Public
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
16 License along with this library; if not, write to the Free Software
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
17 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
18
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
19 Sam Lantinga
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
20 slouken@libsdl.org
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
21 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
22 #include "SDL_config.h"
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
23
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
24 /* This file contains portable iconv functions for SDL */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
25
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
26 #include "SDL_stdinc.h"
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
27 #include "SDL_endian.h"
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
28
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
29 #ifdef HAVE_ICONV
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
30
3987
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
31 /* Depending on which standard the iconv() was implemented with,
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
32 iconv() may or may not use const char ** for the inbuf param.
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
33 If we get this wrong, it's just a warning, so no big deal.
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
34 */
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
35 #if defined(_XGP6) || \
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
36 defined(__GLIBC__) && ((__GLIBC__ > 2) || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2))
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
37 #define ICONV_INBUF_NONCONST
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
38 #endif
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
39
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
40 #include <errno.h>
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
41
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
42 size_t SDL_iconv(SDL_iconv_t cd,
3984
b74530a1dad6 Better fix for bug 419
Sam Lantinga <slouken@libsdl.org>
parents: 3918
diff changeset
43 const char **inbuf, size_t *inbytesleft,
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
44 char **outbuf, size_t *outbytesleft)
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
45 {
3985
2f8efcf14c83 Okay, apparently the newer standard specifies char** for the inbuf
Sam Lantinga <slouken@libsdl.org>
parents: 3984
diff changeset
46 size_t retCode;
3987
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
47 #ifdef ICONV_INBUF_NONCONST
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
48 retCode = iconv(cd, (char **)inbuf, inbytesleft, outbuf, outbytesleft);
3985
2f8efcf14c83 Okay, apparently the newer standard specifies char** for the inbuf
Sam Lantinga <slouken@libsdl.org>
parents: 3984
diff changeset
49 #else
3987
00486a9c2893 iconv() doesn't write to the data, just make compilers happy
Sam Lantinga <slouken@libsdl.org>
parents: 3986
diff changeset
50 retCode = iconv(cd, inbuf, inbytesleft, outbuf, outbytesleft);
3985
2f8efcf14c83 Okay, apparently the newer standard specifies char** for the inbuf
Sam Lantinga <slouken@libsdl.org>
parents: 3984
diff changeset
51 #endif
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
52 if ( retCode == (size_t)-1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
53 switch(errno) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
54 case E2BIG:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
55 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
56 case EILSEQ:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
57 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
58 case EINVAL:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
59 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
60 default:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
61 return SDL_ICONV_ERROR;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
62 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
63 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
64 return retCode;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
65 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
66
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
67 #else
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
68
1503
5e4dad24a5de props yo
Sam Lantinga <slouken@libsdl.org>
parents: 1502
diff changeset
69 /* Lots of useful information on Unicode at:
5e4dad24a5de props yo
Sam Lantinga <slouken@libsdl.org>
parents: 1502
diff changeset
70 http://www.cl.cam.ac.uk/~mgk25/unicode.html
5e4dad24a5de props yo
Sam Lantinga <slouken@libsdl.org>
parents: 1502
diff changeset
71 */
5e4dad24a5de props yo
Sam Lantinga <slouken@libsdl.org>
parents: 1502
diff changeset
72
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
73 #define UNICODE_BOM 0xFEFF
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
74
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
75 #define UNKNOWN_ASCII '?'
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
76 #define UNKNOWN_UNICODE 0xFFFD
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
77
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
78 enum {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
79 ENCODING_UNKNOWN,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
80 ENCODING_ASCII,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
81 ENCODING_LATIN1,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
82 ENCODING_UTF8,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
83 ENCODING_UTF16, /* Needs byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
84 ENCODING_UTF16BE,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
85 ENCODING_UTF16LE,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
86 ENCODING_UTF32, /* Needs byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
87 ENCODING_UTF32BE,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
88 ENCODING_UTF32LE,
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
89 ENCODING_UCS2, /* Native byte order assumed */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
90 ENCODING_UCS4, /* Native byte order assumed */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
91 };
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
92 #if SDL_BYTEORDER == SDL_BIG_ENDIAN
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
93 #define ENCODING_UTF16NATIVE ENCODING_UTF16BE
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
94 #define ENCODING_UTF32NATIVE ENCODING_UTF32BE
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
95 #else
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
96 #define ENCODING_UTF16NATIVE ENCODING_UTF16LE
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
97 #define ENCODING_UTF32NATIVE ENCODING_UTF32LE
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
98 #endif
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
99
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
100 struct _SDL_iconv_t
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
101 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
102 int src_fmt;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
103 int dst_fmt;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
104 };
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
105
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
106 static struct {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
107 const char *name;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
108 int format;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
109 } encodings[] = {
3997
6a4f3a32c2e6 Fixed bug #349
Sam Lantinga <slouken@libsdl.org>
parents: 3987
diff changeset
110 { "646", ENCODING_ASCII },
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
111 { "ASCII", ENCODING_ASCII },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
112 { "US-ASCII", ENCODING_ASCII },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
113 { "LATIN1", ENCODING_LATIN1 },
3997
6a4f3a32c2e6 Fixed bug #349
Sam Lantinga <slouken@libsdl.org>
parents: 3987
diff changeset
114 { "8859-1", ENCODING_LATIN1 },
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
115 { "ISO-8859-1", ENCODING_LATIN1 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
116 { "UTF8", ENCODING_UTF8 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
117 { "UTF-8", ENCODING_UTF8 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
118 { "UTF16", ENCODING_UTF16 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
119 { "UTF-16", ENCODING_UTF16 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
120 { "UTF16BE", ENCODING_UTF16BE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
121 { "UTF-16BE", ENCODING_UTF16BE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
122 { "UTF16LE", ENCODING_UTF16LE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
123 { "UTF-16LE", ENCODING_UTF16LE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
124 { "UTF32", ENCODING_UTF32 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
125 { "UTF-32", ENCODING_UTF32 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
126 { "UTF32BE", ENCODING_UTF32BE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
127 { "UTF-32BE", ENCODING_UTF32BE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
128 { "UTF32LE", ENCODING_UTF32LE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
129 { "UTF-32LE", ENCODING_UTF32LE },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
130 { "UCS2", ENCODING_UCS2 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
131 { "UCS-2", ENCODING_UCS2 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
132 { "UCS4", ENCODING_UCS4 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
133 { "UCS-4", ENCODING_UCS4 },
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
134 };
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
135
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
136 SDL_iconv_t SDL_iconv_open(const char *tocode, const char *fromcode)
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
137 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
138 int src_fmt = ENCODING_UNKNOWN;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
139 int dst_fmt = ENCODING_UNKNOWN;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
140 int i;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
141
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
142 for ( i = 0; i < SDL_arraysize(encodings); ++i ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
143 if ( SDL_strcasecmp(fromcode, encodings[i].name) == 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
144 src_fmt = encodings[i].format;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
145 if ( dst_fmt != ENCODING_UNKNOWN ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
146 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
147 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
148 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
149 if ( SDL_strcasecmp(tocode, encodings[i].name) == 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
150 dst_fmt = encodings[i].format;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
151 if ( src_fmt != ENCODING_UNKNOWN ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
152 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
153 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
154 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
155 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
156 if ( src_fmt != ENCODING_UNKNOWN && dst_fmt != ENCODING_UNKNOWN ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
157 SDL_iconv_t cd = (SDL_iconv_t)SDL_malloc(sizeof(*cd));
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
158 if ( cd ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
159 cd->src_fmt = src_fmt;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
160 cd->dst_fmt = dst_fmt;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
161 return cd;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
162 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
163 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
164 return (SDL_iconv_t)-1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
165 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
166
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
167 size_t SDL_iconv(SDL_iconv_t cd,
3918
f16c15f3bc2b Minor const correctness patch to SDL_iconv.
Ryan C. Gordon <icculus@icculus.org>
parents: 1849
diff changeset
168 const char **inbuf, size_t *inbytesleft,
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
169 char **outbuf, size_t *outbytesleft)
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
170 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
171 /* For simplicity, we'll convert everything to and from UCS-4 */
3918
f16c15f3bc2b Minor const correctness patch to SDL_iconv.
Ryan C. Gordon <icculus@icculus.org>
parents: 1849
diff changeset
172 const char *src;
f16c15f3bc2b Minor const correctness patch to SDL_iconv.
Ryan C. Gordon <icculus@icculus.org>
parents: 1849
diff changeset
173 char *dst;
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
174 size_t srclen, dstlen;
1849
b5a4ac87b98c Fixed uninitialized variable warnings
Sam Lantinga <slouken@libsdl.org>
parents: 1519
diff changeset
175 Uint32 ch = 0;
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
176 size_t total;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
177
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
178 if ( !inbuf || !*inbuf ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
179 /* Reset the context */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
180 return 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
181 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
182 if ( !outbuf || !*outbuf || !outbytesleft || !*outbytesleft ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
183 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
184 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
185 src = *inbuf;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
186 srclen = (inbytesleft ? *inbytesleft : 0);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
187 dst = *outbuf;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
188 dstlen = *outbytesleft;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
189
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
190 switch ( cd->src_fmt ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
191 case ENCODING_UTF16:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
192 /* Scan for a byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
193 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
194 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
195 size_t n = srclen / 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
196 while ( n ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
197 if ( p[0] == 0xFF && p[1] == 0xFE ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
198 cd->src_fmt = ENCODING_UTF16BE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
199 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
200 } else if ( p[0] == 0xFE && p[1] == 0xFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
201 cd->src_fmt = ENCODING_UTF16LE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
202 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
203 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
204 p += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
205 --n;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
206 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
207 if ( n == 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
208 /* We can't tell, default to host order */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
209 cd->src_fmt = ENCODING_UTF16NATIVE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
210 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
211 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
212 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
213 case ENCODING_UTF32:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
214 /* Scan for a byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
215 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
216 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
217 size_t n = srclen / 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
218 while ( n ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
219 if ( p[0] == 0xFF && p[1] == 0xFE &&
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
220 p[2] == 0x00 && p[3] == 0x00 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
221 cd->src_fmt = ENCODING_UTF32BE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
222 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
223 } else if ( p[0] == 0x00 && p[1] == 0x00 &&
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
224 p[2] == 0xFE && p[3] == 0xFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
225 cd->src_fmt = ENCODING_UTF32LE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
226 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
227 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
228 p += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
229 --n;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
230 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
231 if ( n == 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
232 /* We can't tell, default to host order */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
233 cd->src_fmt = ENCODING_UTF32NATIVE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
234 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
235 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
236 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
237 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
238
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
239 switch ( cd->dst_fmt ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
240 case ENCODING_UTF16:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
241 /* Default to host order, need to add byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
242 if ( dstlen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
243 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
244 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
245 *(Uint16 *)dst = UNICODE_BOM;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
246 dst += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
247 dstlen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
248 cd->dst_fmt = ENCODING_UTF16NATIVE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
249 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
250 case ENCODING_UTF32:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
251 /* Default to host order, need to add byte order marker */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
252 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
253 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
254 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
255 *(Uint32 *)dst = UNICODE_BOM;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
256 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
257 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
258 cd->dst_fmt = ENCODING_UTF32NATIVE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
259 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
260 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
261
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
262 total = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
263 while ( srclen > 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
264 /* Decode a character */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
265 switch ( cd->src_fmt ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
266 case ENCODING_ASCII:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
267 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
268 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
269 ch = (Uint32)(p[0] & 0x7F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
270 ++src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
271 --srclen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
272 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
273 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
274 case ENCODING_LATIN1:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
275 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
276 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
277 ch = (Uint32)p[0];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
278 ++src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
279 --srclen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
280 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
281 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
282 case ENCODING_UTF8: /* RFC 3629 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
283 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
284 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
285 size_t left = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
286 SDL_bool overlong = SDL_FALSE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
287 if ( p[0] >= 0xFC ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
288 if ( (p[0] & 0xFE) != 0xFC ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
289 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
290 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
291 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
292 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
293 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
294 if ( p[0] == 0xFC ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
295 overlong = SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
296 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
297 ch = (Uint32)(p[0] & 0x01);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
298 left = 5;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
299 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
300 } else if ( p[0] >= 0xF8 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
301 if ( (p[0] & 0xFC) != 0xF8 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
302 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
303 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
304 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
305 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
306 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
307 if ( p[0] == 0xF8 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
308 overlong = SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
309 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
310 ch = (Uint32)(p[0] & 0x03);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
311 left = 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
312 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
313 } else if ( p[0] >= 0xF0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
314 if ( (p[0] & 0xF8) != 0xF0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
315 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
316 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
317 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
318 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
319 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
320 if ( p[0] == 0xF0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
321 overlong = SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
322 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
323 ch = (Uint32)(p[0] & 0x07);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
324 left = 3;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
325 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
326 } else if ( p[0] >= 0xE0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
327 if ( (p[0] & 0xF0) != 0xE0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
328 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
329 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
330 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
331 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
332 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
333 if ( p[0] == 0xE0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
334 overlong = SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
335 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
336 ch = (Uint32)(p[0] & 0x0F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
337 left = 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
338 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
339 } else if ( p[0] >= 0xC0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
340 if ( (p[0] & 0xE0) != 0xC0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
341 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
342 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
343 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
344 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
345 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
346 if ( (p[0] & 0xCE) == 0xC0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
347 overlong = SDL_TRUE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
348 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
349 ch = (Uint32)(p[0] & 0x1F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
350 left = 1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
351 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
352 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
353 if ( (p[0] & 0x80) != 0x00 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
354 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
355 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
356 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
357 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
358 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
359 ch = (Uint32)p[0];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
360 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
361 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
362 ++src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
363 --srclen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
364 if ( srclen < left ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
365 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
366 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
367 while ( left-- ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
368 ++p;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
369 if ( (p[0] & 0xC0) != 0x80 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
370 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
371 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
372 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
373 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
374 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
375 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
376 ch <<= 6;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
377 ch |= (p[0] & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
378 ++src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
379 --srclen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
380 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
381 if ( overlong ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
382 /* Potential security risk
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
383 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
384 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
385 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
386 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
387 if ( (ch >= 0xD800 && ch <= 0xDFFF) ||
1519
c99e7a9c9bc9 RFC 3629 restricted the range of characters encoded with UTF-8 to 0000-10FFFF (the UTF-16 accessible range)
Sam Lantinga <slouken@libsdl.org>
parents: 1513
diff changeset
388 (ch == 0xFFFE || ch == 0xFFFF) ||
c99e7a9c9bc9 RFC 3629 restricted the range of characters encoded with UTF-8 to 0000-10FFFF (the UTF-16 accessible range)
Sam Lantinga <slouken@libsdl.org>
parents: 1513
diff changeset
389 ch > 0x10FFFF ) {
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
390 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
391 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
392 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
393 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
394 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
395 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
396 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
397 case ENCODING_UTF16BE: /* RFC 2781 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
398 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
399 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
400 Uint16 W1, W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
401 if ( srclen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
402 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
403 }
1513
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
404 W1 = ((Uint16)p[0] << 8) |
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
405 (Uint16)p[1];
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
406 src += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
407 srclen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
408 if ( W1 < 0xD800 || W1 > 0xDFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
409 ch = (Uint32)W1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
410 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
411 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
412 if ( W1 > 0xDBFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
413 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
414 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
415 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
416 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
417 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
418 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
419 if ( srclen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
420 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
421 }
1510
720f8bb49d7d Win32 fixes
Sam Lantinga <slouken@libsdl.org>
parents: 1503
diff changeset
422 p = (Uint8 *)src;
1513
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
423 W2 = ((Uint16)p[0] << 8) |
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
424 (Uint16)p[1];
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
425 src += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
426 srclen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
427 if ( W2 < 0xDC00 || W2 > 0xDFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
428 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
429 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
430 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
431 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
432 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
433 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
434 ch = (((Uint32)(W1 & 0x3FF) << 10) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
435 (Uint32)(W2 & 0x3FF)) + 0x10000;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
436 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
437 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
438 case ENCODING_UTF16LE: /* RFC 2781 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
439 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
440 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
441 Uint16 W1, W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
442 if ( srclen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
443 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
444 }
1513
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
445 W1 = ((Uint16)p[1] << 8) |
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
446 (Uint16)p[0];
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
447 src += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
448 srclen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
449 if ( W1 < 0xD800 || W1 > 0xDFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
450 ch = (Uint32)W1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
451 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
452 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
453 if ( W1 > 0xDBFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
454 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
455 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
456 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
457 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
458 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
459 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
460 if ( srclen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
461 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
462 }
1510
720f8bb49d7d Win32 fixes
Sam Lantinga <slouken@libsdl.org>
parents: 1503
diff changeset
463 p = (Uint8 *)src;
1513
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
464 W2 = ((Uint16)p[1] << 8) |
13a3520a13f9 *** empty log message ***
Sam Lantinga <slouken@libsdl.org>
parents: 1510
diff changeset
465 (Uint16)p[0];
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
466 src += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
467 srclen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
468 if ( W2 < 0xDC00 || W2 > 0xDFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
469 /* Skip illegal sequences
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
470 return SDL_ICONV_EILSEQ;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
471 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
472 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
473 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
474 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
475 ch = (((Uint32)(W1 & 0x3FF) << 10) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
476 (Uint32)(W2 & 0x3FF)) + 0x10000;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
477 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
478 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
479 case ENCODING_UTF32BE:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
480 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
481 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
482 if ( srclen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
483 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
484 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
485 ch = ((Uint32)p[0] << 24) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
486 ((Uint32)p[1] << 16) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
487 ((Uint32)p[2] << 8) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
488 (Uint32)p[3];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
489 src += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
490 srclen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
491 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
492 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
493 case ENCODING_UTF32LE:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
494 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
495 Uint8 *p = (Uint8 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
496 if ( srclen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
497 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
498 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
499 ch = ((Uint32)p[3] << 24) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
500 ((Uint32)p[2] << 16) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
501 ((Uint32)p[1] << 8) |
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
502 (Uint32)p[0];
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
503 src += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
504 srclen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
505 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
506 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
507 case ENCODING_UCS2:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
508 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
509 Uint16 *p = (Uint16 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
510 if ( srclen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
511 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
512 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
513 ch = *p;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
514 src += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
515 srclen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
516 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
517 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
518 case ENCODING_UCS4:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
519 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
520 Uint32 *p = (Uint32 *)src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
521 if ( srclen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
522 return SDL_ICONV_EINVAL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
523 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
524 ch = *p;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
525 src += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
526 srclen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
527 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
528 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
529 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
530
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
531 /* Encode a character */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
532 switch ( cd->dst_fmt ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
533 case ENCODING_ASCII:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
534 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
535 Uint8 *p = (Uint8 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
536 if ( dstlen < 1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
537 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
538 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
539 if ( ch > 0x7F ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
540 *p = UNKNOWN_ASCII;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
541 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
542 *p = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
543 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
544 ++dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
545 --dstlen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
546 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
547 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
548 case ENCODING_LATIN1:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
549 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
550 Uint8 *p = (Uint8 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
551 if ( dstlen < 1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
552 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
553 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
554 if ( ch > 0xFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
555 *p = UNKNOWN_ASCII;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
556 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
557 *p = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
558 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
559 ++dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
560 --dstlen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
561 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
562 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
563 case ENCODING_UTF8: /* RFC 3629 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
564 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
565 Uint8 *p = (Uint8 *)dst;
1519
c99e7a9c9bc9 RFC 3629 restricted the range of characters encoded with UTF-8 to 0000-10FFFF (the UTF-16 accessible range)
Sam Lantinga <slouken@libsdl.org>
parents: 1513
diff changeset
566 if ( ch > 0x10FFFF ) {
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
567 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
568 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
569 if ( ch <= 0x7F ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
570 if ( dstlen < 1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
571 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
572 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
573 *p = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
574 ++dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
575 --dstlen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
576 } else if ( ch <= 0x7FF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
577 if ( dstlen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
578 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
579 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
580 p[0] = 0xC0 | (Uint8)((ch >> 6) & 0x1F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
581 p[1] = 0x80 | (Uint8)(ch & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
582 dst += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
583 dstlen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
584 } else if ( ch <= 0xFFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
585 if ( dstlen < 3 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
586 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
587 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
588 p[0] = 0xE0 | (Uint8)((ch >> 12) & 0x0F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
589 p[1] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
590 p[2] = 0x80 | (Uint8)(ch & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
591 dst += 3;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
592 dstlen -= 3;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
593 } else if ( ch <= 0x1FFFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
594 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
595 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
596 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
597 p[0] = 0xF0 | (Uint8)((ch >> 18) & 0x07);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
598 p[1] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
599 p[2] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
600 p[3] = 0x80 | (Uint8)(ch & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
601 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
602 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
603 } else if ( ch <= 0x3FFFFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
604 if ( dstlen < 5 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
605 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
606 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
607 p[0] = 0xF8 | (Uint8)((ch >> 24) & 0x03);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
608 p[1] = 0x80 | (Uint8)((ch >> 18) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
609 p[2] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
610 p[3] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
611 p[4] = 0x80 | (Uint8)(ch & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
612 dst += 5;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
613 dstlen -= 5;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
614 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
615 if ( dstlen < 6 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
616 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
617 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
618 p[0] = 0xFC | (Uint8)((ch >> 30) & 0x01);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
619 p[1] = 0x80 | (Uint8)((ch >> 24) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
620 p[2] = 0x80 | (Uint8)((ch >> 18) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
621 p[3] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
622 p[4] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
623 p[5] = 0x80 | (Uint8)(ch & 0x3F);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
624 dst += 6;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
625 dstlen -= 6;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
626 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
627 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
628 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
629 case ENCODING_UTF16BE: /* RFC 2781 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
630 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
631 Uint8 *p = (Uint8 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
632 if ( ch > 0x10FFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
633 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
634 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
635 if ( ch < 0x10000 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
636 if ( dstlen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
637 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
638 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
639 p[0] = (Uint8)(ch >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
640 p[1] = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
641 dst += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
642 dstlen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
643 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
644 Uint16 W1, W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
645 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
646 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
647 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
648 ch = ch - 0x10000;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
649 W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
650 W2 = 0xDC00 | (Uint16)(ch & 0x3FF);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
651 p[0] = (Uint8)(W1 >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
652 p[1] = (Uint8)W1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
653 p[2] = (Uint8)(W2 >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
654 p[3] = (Uint8)W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
655 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
656 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
657 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
658 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
659 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
660 case ENCODING_UTF16LE: /* RFC 2781 */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
661 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
662 Uint8 *p = (Uint8 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
663 if ( ch > 0x10FFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
664 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
665 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
666 if ( ch < 0x10000 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
667 if ( dstlen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
668 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
669 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
670 p[1] = (Uint8)(ch >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
671 p[0] = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
672 dst += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
673 dstlen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
674 } else {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
675 Uint16 W1, W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
676 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
677 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
678 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
679 ch = ch - 0x10000;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
680 W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
681 W2 = 0xDC00 | (Uint16)(ch & 0x3FF);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
682 p[1] = (Uint8)(W1 >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
683 p[0] = (Uint8)W1;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
684 p[3] = (Uint8)(W2 >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
685 p[2] = (Uint8)W2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
686 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
687 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
688 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
689 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
690 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
691 case ENCODING_UTF32BE:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
692 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
693 Uint8 *p = (Uint8 *)dst;
1502
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
694 if ( ch > 0x10FFFF ) {
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
695 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
696 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
697 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
698 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
699 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
700 p[0] = (Uint8)(ch >> 24);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
701 p[1] = (Uint8)(ch >> 16);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
702 p[2] = (Uint8)(ch >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
703 p[3] = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
704 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
705 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
706 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
707 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
708 case ENCODING_UTF32LE:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
709 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
710 Uint8 *p = (Uint8 *)dst;
1502
d403a39389da UTF-32 by definition can't handle characters higher than 0x10FFFF
Sam Lantinga <slouken@libsdl.org>
parents: 1501
diff changeset
711 if ( ch > 0x10FFFF ) {
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
712 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
713 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
714 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
715 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
716 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
717 p[3] = (Uint8)(ch >> 24);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
718 p[2] = (Uint8)(ch >> 16);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
719 p[1] = (Uint8)(ch >> 8);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
720 p[0] = (Uint8)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
721 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
722 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
723 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
724 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
725 case ENCODING_UCS2:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
726 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
727 Uint16 *p = (Uint16 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
728 if ( ch > 0xFFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
729 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
730 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
731 if ( dstlen < 2 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
732 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
733 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
734 *p = (Uint16)ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
735 dst += 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
736 dstlen -= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
737 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
738 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
739 case ENCODING_UCS4:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
740 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
741 Uint32 *p = (Uint32 *)dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
742 if ( ch > 0x7FFFFFFF ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
743 ch = UNKNOWN_UNICODE;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
744 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
745 if ( dstlen < 4 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
746 return SDL_ICONV_E2BIG;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
747 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
748 *p = ch;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
749 dst += 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
750 dstlen -= 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
751 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
752 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
753 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
754
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
755 /* Update state */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
756 *inbuf = src;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
757 *inbytesleft = srclen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
758 *outbuf = dst;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
759 *outbytesleft = dstlen;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
760 ++total;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
761 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
762 return total;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
763 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
764
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
765 int SDL_iconv_close(SDL_iconv_t cd)
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
766 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
767 if ( cd && cd != (SDL_iconv_t)-1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
768 SDL_free(cd);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
769 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
770 return 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
771 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
772
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
773 #endif /* !HAVE_ICONV */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
774
3998
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
775 static const char *getlocale()
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
776 {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
777 const char *lang;
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
778
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
779 lang = SDL_getenv("LC_ALL");
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
780 if ( !lang ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
781 lang = SDL_getenv("LC_CTYPE");
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
782 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
783 if ( !lang ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
784 lang = SDL_getenv("LC_MESSAGES");
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
785 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
786 if ( !lang ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
787 lang = SDL_getenv("LANG");
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
788 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
789 if ( !lang || SDL_strcmp(lang, "C") == 0 ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
790 lang = "ASCII";
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
791 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
792 return lang;
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
793 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
794
3918
f16c15f3bc2b Minor const correctness patch to SDL_iconv.
Ryan C. Gordon <icculus@icculus.org>
parents: 1849
diff changeset
795 char *SDL_iconv_string(const char *tocode, const char *fromcode, const char *inbuf, size_t inbytesleft)
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
796 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
797 SDL_iconv_t cd;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
798 char *string;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
799 size_t stringsize;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
800 char *outbuf;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
801 size_t outbytesleft;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
802 size_t retCode = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
803
3998
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
804 if ( !fromcode || !*fromcode ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
805 fromcode = getlocale();
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
806 }
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
807 if ( !tocode || !*tocode ) {
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
808 tocode = getlocale();
098ac044cd2f Fixed bug #447
Sam Lantinga <slouken@libsdl.org>
parents: 3997
diff changeset
809 }
1501
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
810 cd = SDL_iconv_open(tocode, fromcode);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
811 if ( cd == (SDL_iconv_t)-1 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
812 return NULL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
813 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
814
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
815 stringsize = inbytesleft > 4 ? inbytesleft : 4;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
816 string = SDL_malloc(stringsize);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
817 if ( !string ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
818 SDL_iconv_close(cd);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
819 return NULL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
820 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
821 outbuf = string;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
822 outbytesleft = stringsize;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
823 SDL_memset(outbuf, 0, 4);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
824
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
825 while ( inbytesleft > 0 ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
826 retCode = SDL_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
827 switch (retCode) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
828 case SDL_ICONV_E2BIG:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
829 {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
830 char *oldstring = string;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
831 stringsize *= 2;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
832 string = SDL_realloc(string, stringsize);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
833 if ( !string ) {
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
834 SDL_iconv_close(cd);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
835 return NULL;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
836 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
837 outbuf = string + (outbuf - oldstring);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
838 outbytesleft = stringsize - (outbuf - string);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
839 SDL_memset(outbuf, 0, 4);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
840 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
841 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
842 case SDL_ICONV_EILSEQ:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
843 /* Try skipping some input data - not perfect, but... */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
844 ++inbuf;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
845 --inbytesleft;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
846 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
847 case SDL_ICONV_EINVAL:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
848 case SDL_ICONV_ERROR:
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
849 /* We can't continue... */
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
850 inbytesleft = 0;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
851 break;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
852 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
853 }
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
854 SDL_iconv_close(cd);
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
855
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
856 return string;
73dc5d39bbf8 Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
Sam Lantinga <slouken@libsdl.org>
parents:
diff changeset
857 }