src/stdlib/SDL_iconv.c
author Sam Lantinga
Wed, 04 Jul 2007 07:54:06 +0000
branchSDL-1.2
changeset 3998 098ac044cd2f
parent 3997 6a4f3a32c2e6
child 4053 0aadbc81c497
permissions -rw-r--r--
Fixed bug #447

Xlib uses the native locale, not latin1

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