src/stdlib/SDL_iconv.c
author Sam Lantinga <slouken@libsdl.org>
Mon, 13 Mar 2006 01:08:00 +0000
changeset 1501 73dc5d39bbf8
child 1502 d403a39389da
permissions -rw-r--r--
Added UTF-8 <-> UTF-16 <-> UTF-32 <-> UCS-2 <-> UCS-4 conversion capability
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@1501
    31
#include <errno.h>
slouken@1501
    32
slouken@1501
    33
size_t SDL_iconv(SDL_iconv_t cd,
slouken@1501
    34
                 char **inbuf, size_t *inbytesleft,
slouken@1501
    35
                 char **outbuf, size_t *outbytesleft)
slouken@1501
    36
{
slouken@1501
    37
	size_t retCode = iconv(cd, inbuf, inbytesleft, outbuf, outbytesleft);
slouken@1501
    38
	if ( retCode == (size_t)-1 ) {
slouken@1501
    39
		switch(errno) {
slouken@1501
    40
		    case E2BIG:
slouken@1501
    41
			return SDL_ICONV_E2BIG;
slouken@1501
    42
		    case EILSEQ:
slouken@1501
    43
			return SDL_ICONV_EILSEQ;
slouken@1501
    44
		    case EINVAL:
slouken@1501
    45
			return SDL_ICONV_EINVAL;
slouken@1501
    46
		    default:
slouken@1501
    47
			return SDL_ICONV_ERROR;
slouken@1501
    48
		}
slouken@1501
    49
	}
slouken@1501
    50
	return retCode;
slouken@1501
    51
}
slouken@1501
    52
slouken@1501
    53
#else
slouken@1501
    54
slouken@1501
    55
#define UNICODE_BOM	0xFEFF
slouken@1501
    56
slouken@1501
    57
#define UNKNOWN_ASCII	'?'
slouken@1501
    58
#define UNKNOWN_UNICODE	0xFFFD
slouken@1501
    59
slouken@1501
    60
enum {
slouken@1501
    61
	ENCODING_UNKNOWN,
slouken@1501
    62
	ENCODING_ASCII,
slouken@1501
    63
	ENCODING_LATIN1,
slouken@1501
    64
	ENCODING_UTF8,
slouken@1501
    65
	ENCODING_UTF16,		/* Needs byte order marker */
slouken@1501
    66
	ENCODING_UTF16BE,
slouken@1501
    67
	ENCODING_UTF16LE,
slouken@1501
    68
	ENCODING_UTF32,		/* Needs byte order marker */
slouken@1501
    69
	ENCODING_UTF32BE,
slouken@1501
    70
	ENCODING_UTF32LE,
slouken@1501
    71
	ENCODING_UCS2,		/* Native byte order assumed */
slouken@1501
    72
	ENCODING_UCS4,		/* Native byte order assumed */
slouken@1501
    73
};
slouken@1501
    74
#if SDL_BYTEORDER == SDL_BIG_ENDIAN
slouken@1501
    75
#define ENCODING_UTF16NATIVE	ENCODING_UTF16BE
slouken@1501
    76
#define ENCODING_UTF32NATIVE	ENCODING_UTF32BE
slouken@1501
    77
#else
slouken@1501
    78
#define ENCODING_UTF16NATIVE	ENCODING_UTF16LE
slouken@1501
    79
#define ENCODING_UTF32NATIVE	ENCODING_UTF32LE
slouken@1501
    80
#endif
slouken@1501
    81
slouken@1501
    82
struct _SDL_iconv_t
slouken@1501
    83
{
slouken@1501
    84
	int src_fmt;
slouken@1501
    85
	int dst_fmt;
slouken@1501
    86
};
slouken@1501
    87
slouken@1501
    88
static struct {
slouken@1501
    89
	const char *name;
slouken@1501
    90
	int format;
slouken@1501
    91
} encodings[] = {
slouken@1501
    92
	{ "ASCII",	ENCODING_ASCII },
slouken@1501
    93
	{ "US-ASCII",	ENCODING_ASCII },
slouken@1501
    94
	{ "LATIN1",	ENCODING_LATIN1 },
slouken@1501
    95
	{ "ISO-8859-1",	ENCODING_LATIN1 },
slouken@1501
    96
	{ "UTF8",	ENCODING_UTF8 },
slouken@1501
    97
	{ "UTF-8",	ENCODING_UTF8 },
slouken@1501
    98
	{ "UTF16",	ENCODING_UTF16 },
slouken@1501
    99
	{ "UTF-16",	ENCODING_UTF16 },
slouken@1501
   100
	{ "UTF16BE",	ENCODING_UTF16BE },
slouken@1501
   101
	{ "UTF-16BE",	ENCODING_UTF16BE },
slouken@1501
   102
	{ "UTF16LE",	ENCODING_UTF16LE },
slouken@1501
   103
	{ "UTF-16LE",	ENCODING_UTF16LE },
slouken@1501
   104
	{ "UTF32",	ENCODING_UTF32 },
slouken@1501
   105
	{ "UTF-32",	ENCODING_UTF32 },
slouken@1501
   106
	{ "UTF32BE",	ENCODING_UTF32BE },
slouken@1501
   107
	{ "UTF-32BE",	ENCODING_UTF32BE },
slouken@1501
   108
	{ "UTF32LE",	ENCODING_UTF32LE },
slouken@1501
   109
	{ "UTF-32LE",	ENCODING_UTF32LE },
slouken@1501
   110
	{ "UCS2",	ENCODING_UCS2 },
slouken@1501
   111
	{ "UCS-2",	ENCODING_UCS2 },
slouken@1501
   112
	{ "UCS4",	ENCODING_UCS4 },
slouken@1501
   113
	{ "UCS-4",	ENCODING_UCS4 },
slouken@1501
   114
};
slouken@1501
   115
slouken@1501
   116
SDL_iconv_t SDL_iconv_open(const char *tocode, const char *fromcode)
slouken@1501
   117
{
slouken@1501
   118
	int src_fmt = ENCODING_UNKNOWN;
slouken@1501
   119
	int dst_fmt = ENCODING_UNKNOWN;
slouken@1501
   120
	int i;
slouken@1501
   121
slouken@1501
   122
	for ( i = 0; i < SDL_arraysize(encodings); ++i ) {
slouken@1501
   123
		if ( SDL_strcasecmp(fromcode, encodings[i].name) == 0 ) {
slouken@1501
   124
			src_fmt = encodings[i].format;
slouken@1501
   125
			if ( dst_fmt != ENCODING_UNKNOWN ) {
slouken@1501
   126
				break;
slouken@1501
   127
			}
slouken@1501
   128
		}
slouken@1501
   129
		if ( SDL_strcasecmp(tocode, encodings[i].name) == 0 ) {
slouken@1501
   130
			dst_fmt = encodings[i].format;
slouken@1501
   131
			if ( src_fmt != ENCODING_UNKNOWN ) {
slouken@1501
   132
				break;
slouken@1501
   133
			}
slouken@1501
   134
		}
slouken@1501
   135
	}
slouken@1501
   136
	if ( src_fmt != ENCODING_UNKNOWN && dst_fmt != ENCODING_UNKNOWN ) {
slouken@1501
   137
		SDL_iconv_t cd = (SDL_iconv_t)SDL_malloc(sizeof(*cd));
slouken@1501
   138
		if ( cd ) {
slouken@1501
   139
			cd->src_fmt = src_fmt;
slouken@1501
   140
			cd->dst_fmt = dst_fmt;
slouken@1501
   141
			return cd;
slouken@1501
   142
		}
slouken@1501
   143
	}
slouken@1501
   144
	return (SDL_iconv_t)-1;
slouken@1501
   145
}
slouken@1501
   146
slouken@1501
   147
size_t SDL_iconv(SDL_iconv_t cd,
slouken@1501
   148
                 char **inbuf, size_t *inbytesleft,
slouken@1501
   149
                 char **outbuf, size_t *outbytesleft)
slouken@1501
   150
{
slouken@1501
   151
	/* For simplicity, we'll convert everything to and from UCS-4 */
slouken@1501
   152
	char *src, *dst;
slouken@1501
   153
	size_t srclen, dstlen;
slouken@1501
   154
	Uint32 ch;
slouken@1501
   155
	size_t total;
slouken@1501
   156
slouken@1501
   157
	if ( !inbuf || !*inbuf ) {
slouken@1501
   158
		/* Reset the context */
slouken@1501
   159
		return 0;
slouken@1501
   160
	}
slouken@1501
   161
	if ( !outbuf || !*outbuf || !outbytesleft || !*outbytesleft ) {
slouken@1501
   162
		return SDL_ICONV_E2BIG;
slouken@1501
   163
	}
slouken@1501
   164
	src = *inbuf;
slouken@1501
   165
	srclen = (inbytesleft ? *inbytesleft : 0);
slouken@1501
   166
	dst = *outbuf;
slouken@1501
   167
	dstlen = *outbytesleft;
slouken@1501
   168
slouken@1501
   169
	switch ( cd->src_fmt ) {
slouken@1501
   170
	    case ENCODING_UTF16:
slouken@1501
   171
		/* Scan for a byte order marker */
slouken@1501
   172
		{
slouken@1501
   173
			Uint8 *p = (Uint8 *)src;
slouken@1501
   174
			size_t n = srclen / 2;
slouken@1501
   175
			while ( n ) {
slouken@1501
   176
				if ( p[0] == 0xFF && p[1] == 0xFE ) {
slouken@1501
   177
					cd->src_fmt = ENCODING_UTF16BE;
slouken@1501
   178
					break;
slouken@1501
   179
				} else if ( p[0] == 0xFE && p[1] == 0xFF ) {
slouken@1501
   180
					cd->src_fmt = ENCODING_UTF16LE;
slouken@1501
   181
					break;
slouken@1501
   182
				}
slouken@1501
   183
				p += 2;
slouken@1501
   184
				--n;
slouken@1501
   185
			}
slouken@1501
   186
			if ( n == 0 ) {
slouken@1501
   187
				/* We can't tell, default to host order */
slouken@1501
   188
				cd->src_fmt = ENCODING_UTF16NATIVE;
slouken@1501
   189
			}
slouken@1501
   190
		}
slouken@1501
   191
		break;
slouken@1501
   192
	    case ENCODING_UTF32:
slouken@1501
   193
		/* Scan for a byte order marker */
slouken@1501
   194
		{
slouken@1501
   195
			Uint8 *p = (Uint8 *)src;
slouken@1501
   196
			size_t n = srclen / 4;
slouken@1501
   197
			while ( n ) {
slouken@1501
   198
				if ( p[0] == 0xFF && p[1] == 0xFE &&
slouken@1501
   199
				     p[2] == 0x00 && p[3] == 0x00 ) {
slouken@1501
   200
					cd->src_fmt = ENCODING_UTF32BE;
slouken@1501
   201
					break;
slouken@1501
   202
				} else if ( p[0] == 0x00 && p[1] == 0x00 &&
slouken@1501
   203
				            p[2] == 0xFE && p[3] == 0xFF ) {
slouken@1501
   204
					cd->src_fmt = ENCODING_UTF32LE;
slouken@1501
   205
					break;
slouken@1501
   206
				}
slouken@1501
   207
				p += 4;
slouken@1501
   208
				--n;
slouken@1501
   209
			}
slouken@1501
   210
			if ( n == 0 ) {
slouken@1501
   211
				/* We can't tell, default to host order */
slouken@1501
   212
				cd->src_fmt = ENCODING_UTF32NATIVE;
slouken@1501
   213
			}
slouken@1501
   214
		}
slouken@1501
   215
		break;
slouken@1501
   216
	}
slouken@1501
   217
slouken@1501
   218
	switch ( cd->dst_fmt ) {
slouken@1501
   219
	    case ENCODING_UTF16:
slouken@1501
   220
		/* Default to host order, need to add byte order marker */
slouken@1501
   221
		if ( dstlen < 2 ) {
slouken@1501
   222
			return SDL_ICONV_E2BIG;
slouken@1501
   223
		}
slouken@1501
   224
		*(Uint16 *)dst = UNICODE_BOM;
slouken@1501
   225
		dst += 2;
slouken@1501
   226
		dstlen -= 2;
slouken@1501
   227
		cd->dst_fmt = ENCODING_UTF16NATIVE;
slouken@1501
   228
		break;
slouken@1501
   229
	    case ENCODING_UTF32:
slouken@1501
   230
		/* Default to host order, need to add byte order marker */
slouken@1501
   231
		if ( dstlen < 4 ) {
slouken@1501
   232
			return SDL_ICONV_E2BIG;
slouken@1501
   233
		}
slouken@1501
   234
		*(Uint32 *)dst = UNICODE_BOM;
slouken@1501
   235
		dst += 4;
slouken@1501
   236
		dstlen -= 4;
slouken@1501
   237
		cd->dst_fmt = ENCODING_UTF32NATIVE;
slouken@1501
   238
		break;
slouken@1501
   239
	}
slouken@1501
   240
slouken@1501
   241
	total = 0;
slouken@1501
   242
	while ( srclen > 0 ) {
slouken@1501
   243
		/* Decode a character */
slouken@1501
   244
		switch ( cd->src_fmt ) {
slouken@1501
   245
		    case ENCODING_ASCII:
slouken@1501
   246
			{
slouken@1501
   247
				Uint8 *p = (Uint8 *)src;
slouken@1501
   248
				ch = (Uint32)(p[0] & 0x7F);
slouken@1501
   249
				++src;
slouken@1501
   250
				--srclen;
slouken@1501
   251
			}
slouken@1501
   252
			break;
slouken@1501
   253
		    case ENCODING_LATIN1:
slouken@1501
   254
			{
slouken@1501
   255
				Uint8 *p = (Uint8 *)src;
slouken@1501
   256
				ch = (Uint32)p[0];
slouken@1501
   257
				++src;
slouken@1501
   258
				--srclen;
slouken@1501
   259
			}
slouken@1501
   260
			break;
slouken@1501
   261
		    case ENCODING_UTF8: /* RFC 3629 */
slouken@1501
   262
			{
slouken@1501
   263
				Uint8 *p = (Uint8 *)src;
slouken@1501
   264
				size_t left = 0;
slouken@1501
   265
				SDL_bool overlong = SDL_FALSE;
slouken@1501
   266
				if ( p[0] >= 0xFC ) {
slouken@1501
   267
					if ( (p[0] & 0xFE) != 0xFC ) {
slouken@1501
   268
						/* Skip illegal sequences
slouken@1501
   269
						return SDL_ICONV_EILSEQ;
slouken@1501
   270
						*/
slouken@1501
   271
						ch = UNKNOWN_UNICODE;
slouken@1501
   272
					} else {
slouken@1501
   273
						if ( p[0] == 0xFC ) {
slouken@1501
   274
							overlong = SDL_TRUE;
slouken@1501
   275
						}
slouken@1501
   276
						ch = (Uint32)(p[0] & 0x01);
slouken@1501
   277
						left = 5;
slouken@1501
   278
					}
slouken@1501
   279
				} else if ( p[0] >= 0xF8 ) {
slouken@1501
   280
					if ( (p[0] & 0xFC) != 0xF8 ) {
slouken@1501
   281
						/* Skip illegal sequences
slouken@1501
   282
						return SDL_ICONV_EILSEQ;
slouken@1501
   283
						*/
slouken@1501
   284
						ch = UNKNOWN_UNICODE;
slouken@1501
   285
					} else {
slouken@1501
   286
						if ( p[0] == 0xF8 ) {
slouken@1501
   287
							overlong = SDL_TRUE;
slouken@1501
   288
						}
slouken@1501
   289
						ch = (Uint32)(p[0] & 0x03);
slouken@1501
   290
						left = 4;
slouken@1501
   291
					}
slouken@1501
   292
				} else if ( p[0] >= 0xF0 ) {
slouken@1501
   293
					if ( (p[0] & 0xF8) != 0xF0 ) {
slouken@1501
   294
						/* Skip illegal sequences
slouken@1501
   295
						return SDL_ICONV_EILSEQ;
slouken@1501
   296
						*/
slouken@1501
   297
						ch = UNKNOWN_UNICODE;
slouken@1501
   298
					} else {
slouken@1501
   299
						if ( p[0] == 0xF0 ) {
slouken@1501
   300
							overlong = SDL_TRUE;
slouken@1501
   301
						}
slouken@1501
   302
						ch = (Uint32)(p[0] & 0x07);
slouken@1501
   303
						left = 3;
slouken@1501
   304
					}
slouken@1501
   305
				} else if ( p[0] >= 0xE0 ) {
slouken@1501
   306
					if ( (p[0] & 0xF0) != 0xE0 ) {
slouken@1501
   307
						/* Skip illegal sequences
slouken@1501
   308
						return SDL_ICONV_EILSEQ;
slouken@1501
   309
						*/
slouken@1501
   310
						ch = UNKNOWN_UNICODE;
slouken@1501
   311
					} else {
slouken@1501
   312
						if ( p[0] == 0xE0 ) {
slouken@1501
   313
							overlong = SDL_TRUE;
slouken@1501
   314
						}
slouken@1501
   315
						ch = (Uint32)(p[0] & 0x0F);
slouken@1501
   316
						left = 2;
slouken@1501
   317
					}
slouken@1501
   318
				} else if ( p[0] >= 0xC0 ) {
slouken@1501
   319
					if ( (p[0] & 0xE0) != 0xC0 ) {
slouken@1501
   320
						/* Skip illegal sequences
slouken@1501
   321
						return SDL_ICONV_EILSEQ;
slouken@1501
   322
						*/
slouken@1501
   323
						ch = UNKNOWN_UNICODE;
slouken@1501
   324
					} else {
slouken@1501
   325
						if ( (p[0] & 0xCE) == 0xC0 ) {
slouken@1501
   326
							overlong = SDL_TRUE;
slouken@1501
   327
						}
slouken@1501
   328
						ch = (Uint32)(p[0] & 0x1F);
slouken@1501
   329
						left = 1;
slouken@1501
   330
					}
slouken@1501
   331
				} else {
slouken@1501
   332
					if ( (p[0] & 0x80) != 0x00 ) {
slouken@1501
   333
						/* Skip illegal sequences
slouken@1501
   334
						return SDL_ICONV_EILSEQ;
slouken@1501
   335
						*/
slouken@1501
   336
						ch = UNKNOWN_UNICODE;
slouken@1501
   337
					} else {
slouken@1501
   338
						ch = (Uint32)p[0];
slouken@1501
   339
					}
slouken@1501
   340
				}
slouken@1501
   341
				++src;
slouken@1501
   342
				--srclen;
slouken@1501
   343
				if ( srclen < left ) {
slouken@1501
   344
					return SDL_ICONV_EINVAL;
slouken@1501
   345
				}
slouken@1501
   346
				while ( left-- ) {
slouken@1501
   347
					++p;
slouken@1501
   348
					if ( (p[0] & 0xC0) != 0x80 ) {
slouken@1501
   349
						/* Skip illegal sequences
slouken@1501
   350
						return SDL_ICONV_EILSEQ;
slouken@1501
   351
						*/
slouken@1501
   352
						ch = UNKNOWN_UNICODE;
slouken@1501
   353
						break;
slouken@1501
   354
					}
slouken@1501
   355
					ch <<= 6;
slouken@1501
   356
					ch |= (p[0] & 0x3F);
slouken@1501
   357
					++src;
slouken@1501
   358
					--srclen;
slouken@1501
   359
				}
slouken@1501
   360
				if ( overlong ) {
slouken@1501
   361
					/* Potential security risk
slouken@1501
   362
					return SDL_ICONV_EILSEQ;
slouken@1501
   363
					*/
slouken@1501
   364
					ch = UNKNOWN_UNICODE;
slouken@1501
   365
				}
slouken@1501
   366
				if ( (ch >= 0xD800 && ch <= 0xDFFF) ||
slouken@1501
   367
				     (ch == 0xFFFE || ch == 0xFFFF) ) {
slouken@1501
   368
					/* Skip illegal sequences
slouken@1501
   369
					return SDL_ICONV_EILSEQ;
slouken@1501
   370
					*/
slouken@1501
   371
					ch = UNKNOWN_UNICODE;
slouken@1501
   372
				}
slouken@1501
   373
			}
slouken@1501
   374
			break;
slouken@1501
   375
		    case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1501
   376
			{
slouken@1501
   377
				Uint8 *p = (Uint8 *)src;
slouken@1501
   378
				Uint16 W1, W2;
slouken@1501
   379
				if ( srclen < 2 ) {
slouken@1501
   380
					return SDL_ICONV_EINVAL;
slouken@1501
   381
				}
slouken@1501
   382
				W1 = ((Uint32)p[0] << 8) |
slouken@1501
   383
				      (Uint32)p[1];
slouken@1501
   384
				src += 2;
slouken@1501
   385
				srclen -= 2;
slouken@1501
   386
				if ( W1 < 0xD800 || W1 > 0xDFFF ) {
slouken@1501
   387
					ch = (Uint32)W1;
slouken@1501
   388
					break;
slouken@1501
   389
				}
slouken@1501
   390
				if ( W1 > 0xDBFF ) {
slouken@1501
   391
					/* Skip illegal sequences
slouken@1501
   392
					return SDL_ICONV_EILSEQ;
slouken@1501
   393
					*/
slouken@1501
   394
					ch = UNKNOWN_UNICODE;
slouken@1501
   395
					break;
slouken@1501
   396
				}
slouken@1501
   397
				if ( srclen < 2 ) {
slouken@1501
   398
					return SDL_ICONV_EINVAL;
slouken@1501
   399
				}
slouken@1501
   400
				p = src;
slouken@1501
   401
				W2 = ((Uint32)p[0] << 8) |
slouken@1501
   402
				      (Uint32)p[1];
slouken@1501
   403
				src += 2;
slouken@1501
   404
				srclen -= 2;
slouken@1501
   405
				if ( W2 < 0xDC00 || W2 > 0xDFFF ) {
slouken@1501
   406
					/* Skip illegal sequences
slouken@1501
   407
					return SDL_ICONV_EILSEQ;
slouken@1501
   408
					*/
slouken@1501
   409
					ch = UNKNOWN_UNICODE;
slouken@1501
   410
					break;
slouken@1501
   411
				}
slouken@1501
   412
				ch = (((Uint32)(W1 & 0x3FF) << 10) |
slouken@1501
   413
				      (Uint32)(W2 & 0x3FF)) + 0x10000;
slouken@1501
   414
			}
slouken@1501
   415
			break;
slouken@1501
   416
		    case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1501
   417
			{
slouken@1501
   418
				Uint8 *p = (Uint8 *)src;
slouken@1501
   419
				Uint16 W1, W2;
slouken@1501
   420
				if ( srclen < 2 ) {
slouken@1501
   421
					return SDL_ICONV_EINVAL;
slouken@1501
   422
				}
slouken@1501
   423
				W1 = ((Uint32)p[1] << 8) |
slouken@1501
   424
				      (Uint32)p[0];
slouken@1501
   425
				src += 2;
slouken@1501
   426
				srclen -= 2;
slouken@1501
   427
				if ( W1 < 0xD800 || W1 > 0xDFFF ) {
slouken@1501
   428
					ch = (Uint32)W1;
slouken@1501
   429
					break;
slouken@1501
   430
				}
slouken@1501
   431
				if ( W1 > 0xDBFF ) {
slouken@1501
   432
					/* Skip illegal sequences
slouken@1501
   433
					return SDL_ICONV_EILSEQ;
slouken@1501
   434
					*/
slouken@1501
   435
					ch = UNKNOWN_UNICODE;
slouken@1501
   436
					break;
slouken@1501
   437
				}
slouken@1501
   438
				if ( srclen < 2 ) {
slouken@1501
   439
					return SDL_ICONV_EINVAL;
slouken@1501
   440
				}
slouken@1501
   441
				p = src;
slouken@1501
   442
				W2 = ((Uint32)p[1] << 8) |
slouken@1501
   443
				      (Uint32)p[0];
slouken@1501
   444
				src += 2;
slouken@1501
   445
				srclen -= 2;
slouken@1501
   446
				if ( W2 < 0xDC00 || W2 > 0xDFFF ) {
slouken@1501
   447
					/* Skip illegal sequences
slouken@1501
   448
					return SDL_ICONV_EILSEQ;
slouken@1501
   449
					*/
slouken@1501
   450
					ch = UNKNOWN_UNICODE;
slouken@1501
   451
					break;
slouken@1501
   452
				}
slouken@1501
   453
				ch = (((Uint32)(W1 & 0x3FF) << 10) |
slouken@1501
   454
				      (Uint32)(W2 & 0x3FF)) + 0x10000;
slouken@1501
   455
			}
slouken@1501
   456
			break;
slouken@1501
   457
		    case ENCODING_UTF32BE:
slouken@1501
   458
			{
slouken@1501
   459
				Uint8 *p = (Uint8 *)src;
slouken@1501
   460
				if ( srclen < 4 ) {
slouken@1501
   461
					return SDL_ICONV_EINVAL;
slouken@1501
   462
				}
slouken@1501
   463
				ch = ((Uint32)p[0] << 24) |
slouken@1501
   464
				     ((Uint32)p[1] << 16) |
slouken@1501
   465
				     ((Uint32)p[2] << 8) |
slouken@1501
   466
				      (Uint32)p[3];
slouken@1501
   467
				src += 4;
slouken@1501
   468
				srclen -= 4;
slouken@1501
   469
			}
slouken@1501
   470
			break;
slouken@1501
   471
		    case ENCODING_UTF32LE:
slouken@1501
   472
			{
slouken@1501
   473
				Uint8 *p = (Uint8 *)src;
slouken@1501
   474
				if ( srclen < 4 ) {
slouken@1501
   475
					return SDL_ICONV_EINVAL;
slouken@1501
   476
				}
slouken@1501
   477
				ch = ((Uint32)p[3] << 24) |
slouken@1501
   478
				     ((Uint32)p[2] << 16) |
slouken@1501
   479
				     ((Uint32)p[1] << 8) |
slouken@1501
   480
				      (Uint32)p[0];
slouken@1501
   481
				src += 4;
slouken@1501
   482
				srclen -= 4;
slouken@1501
   483
			}
slouken@1501
   484
			break;
slouken@1501
   485
		    case ENCODING_UCS2:
slouken@1501
   486
			{
slouken@1501
   487
				Uint16 *p = (Uint16 *)src;
slouken@1501
   488
				if ( srclen < 2 ) {
slouken@1501
   489
					return SDL_ICONV_EINVAL;
slouken@1501
   490
				}
slouken@1501
   491
				ch = *p;
slouken@1501
   492
				src += 2;
slouken@1501
   493
				srclen -= 2;
slouken@1501
   494
			}
slouken@1501
   495
			break;
slouken@1501
   496
		    case ENCODING_UCS4:
slouken@1501
   497
			{
slouken@1501
   498
				Uint32 *p = (Uint32 *)src;
slouken@1501
   499
				if ( srclen < 4 ) {
slouken@1501
   500
					return SDL_ICONV_EINVAL;
slouken@1501
   501
				}
slouken@1501
   502
				ch = *p;
slouken@1501
   503
				src += 4;
slouken@1501
   504
				srclen -= 4;
slouken@1501
   505
			}
slouken@1501
   506
			break;
slouken@1501
   507
		}
slouken@1501
   508
slouken@1501
   509
		/* Encode a character */
slouken@1501
   510
		switch ( cd->dst_fmt ) {
slouken@1501
   511
		    case ENCODING_ASCII:
slouken@1501
   512
			{
slouken@1501
   513
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   514
				if ( dstlen < 1 ) {
slouken@1501
   515
					return SDL_ICONV_E2BIG;
slouken@1501
   516
				}
slouken@1501
   517
				if ( ch > 0x7F ) {
slouken@1501
   518
					*p = UNKNOWN_ASCII;
slouken@1501
   519
				} else {
slouken@1501
   520
					*p = (Uint8)ch;
slouken@1501
   521
				}
slouken@1501
   522
				++dst;
slouken@1501
   523
				--dstlen;
slouken@1501
   524
			}
slouken@1501
   525
			break;
slouken@1501
   526
		    case ENCODING_LATIN1:
slouken@1501
   527
			{
slouken@1501
   528
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   529
				if ( dstlen < 1 ) {
slouken@1501
   530
					return SDL_ICONV_E2BIG;
slouken@1501
   531
				}
slouken@1501
   532
				if ( ch > 0xFF ) {
slouken@1501
   533
					*p = UNKNOWN_ASCII;
slouken@1501
   534
				} else {
slouken@1501
   535
					*p = (Uint8)ch;
slouken@1501
   536
				}
slouken@1501
   537
				++dst;
slouken@1501
   538
				--dstlen;
slouken@1501
   539
			}
slouken@1501
   540
			break;
slouken@1501
   541
		    case ENCODING_UTF8: /* RFC 3629 */
slouken@1501
   542
			{
slouken@1501
   543
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   544
				if ( ch > 0x7FFFFFFF ) {
slouken@1501
   545
					ch = UNKNOWN_UNICODE;
slouken@1501
   546
				}
slouken@1501
   547
				if ( ch <= 0x7F ) {
slouken@1501
   548
					if ( dstlen < 1 ) {
slouken@1501
   549
						return SDL_ICONV_E2BIG;
slouken@1501
   550
					}
slouken@1501
   551
					*p = (Uint8)ch;
slouken@1501
   552
					++dst;
slouken@1501
   553
					--dstlen;
slouken@1501
   554
				} else if ( ch <= 0x7FF ) {
slouken@1501
   555
					if ( dstlen < 2 ) {
slouken@1501
   556
						return SDL_ICONV_E2BIG;
slouken@1501
   557
					}
slouken@1501
   558
					p[0] = 0xC0 | (Uint8)((ch >> 6) & 0x1F);
slouken@1501
   559
					p[1] = 0x80 | (Uint8)(ch & 0x3F);
slouken@1501
   560
					dst += 2;
slouken@1501
   561
					dstlen -= 2;
slouken@1501
   562
				} else if ( ch <= 0xFFFF ) {
slouken@1501
   563
					if ( dstlen < 3 ) {
slouken@1501
   564
						return SDL_ICONV_E2BIG;
slouken@1501
   565
					}
slouken@1501
   566
					p[0] = 0xE0 | (Uint8)((ch >> 12) & 0x0F);
slouken@1501
   567
					p[1] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
slouken@1501
   568
					p[2] = 0x80 | (Uint8)(ch & 0x3F);
slouken@1501
   569
					dst += 3;
slouken@1501
   570
					dstlen -= 3;
slouken@1501
   571
				} else if ( ch <= 0x1FFFFF ) {
slouken@1501
   572
					if ( dstlen < 4 ) {
slouken@1501
   573
						return SDL_ICONV_E2BIG;
slouken@1501
   574
					}
slouken@1501
   575
					p[0] = 0xF0 | (Uint8)((ch >> 18) & 0x07);
slouken@1501
   576
					p[1] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
slouken@1501
   577
					p[2] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
slouken@1501
   578
					p[3] = 0x80 | (Uint8)(ch & 0x3F);
slouken@1501
   579
					dst += 4;
slouken@1501
   580
					dstlen -= 4;
slouken@1501
   581
				} else if ( ch <= 0x3FFFFFF ) {
slouken@1501
   582
					if ( dstlen < 5 ) {
slouken@1501
   583
						return SDL_ICONV_E2BIG;
slouken@1501
   584
					}
slouken@1501
   585
					p[0] = 0xF8 | (Uint8)((ch >> 24) & 0x03);
slouken@1501
   586
					p[1] = 0x80 | (Uint8)((ch >> 18) & 0x3F);
slouken@1501
   587
					p[2] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
slouken@1501
   588
					p[3] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
slouken@1501
   589
					p[4] = 0x80 | (Uint8)(ch & 0x3F);
slouken@1501
   590
					dst += 5;
slouken@1501
   591
					dstlen -= 5;
slouken@1501
   592
				} else {
slouken@1501
   593
					if ( dstlen < 6 ) {
slouken@1501
   594
						return SDL_ICONV_E2BIG;
slouken@1501
   595
					}
slouken@1501
   596
					p[0] = 0xFC | (Uint8)((ch >> 30) & 0x01);
slouken@1501
   597
					p[1] = 0x80 | (Uint8)((ch >> 24) & 0x3F);
slouken@1501
   598
					p[2] = 0x80 | (Uint8)((ch >> 18) & 0x3F);
slouken@1501
   599
					p[3] = 0x80 | (Uint8)((ch >> 12) & 0x3F);
slouken@1501
   600
					p[4] = 0x80 | (Uint8)((ch >> 6) & 0x3F);
slouken@1501
   601
					p[5] = 0x80 | (Uint8)(ch & 0x3F);
slouken@1501
   602
					dst += 6;
slouken@1501
   603
					dstlen -= 6;
slouken@1501
   604
				}
slouken@1501
   605
			}
slouken@1501
   606
			break;
slouken@1501
   607
		    case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1501
   608
			{
slouken@1501
   609
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   610
				if ( ch > 0x10FFFF ) {
slouken@1501
   611
					ch = UNKNOWN_UNICODE;
slouken@1501
   612
				}
slouken@1501
   613
				if ( ch < 0x10000 ) {
slouken@1501
   614
					if ( dstlen < 2 ) {
slouken@1501
   615
						return SDL_ICONV_E2BIG;
slouken@1501
   616
					}
slouken@1501
   617
					p[0] = (Uint8)(ch >> 8);
slouken@1501
   618
					p[1] = (Uint8)ch;
slouken@1501
   619
					dst += 2;
slouken@1501
   620
					dstlen -= 2;
slouken@1501
   621
				} else {
slouken@1501
   622
					Uint16 W1, W2;
slouken@1501
   623
					if ( dstlen < 4 ) {
slouken@1501
   624
						return SDL_ICONV_E2BIG;
slouken@1501
   625
					}
slouken@1501
   626
					ch = ch - 0x10000;
slouken@1501
   627
					W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF);
slouken@1501
   628
					W2 = 0xDC00 | (Uint16)(ch & 0x3FF);
slouken@1501
   629
					p[0] = (Uint8)(W1 >> 8);
slouken@1501
   630
					p[1] = (Uint8)W1;
slouken@1501
   631
					p[2] = (Uint8)(W2 >> 8);
slouken@1501
   632
					p[3] = (Uint8)W2;
slouken@1501
   633
					dst += 4;
slouken@1501
   634
					dstlen -= 4;
slouken@1501
   635
				}
slouken@1501
   636
			}
slouken@1501
   637
			break;
slouken@1501
   638
		    case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1501
   639
			{
slouken@1501
   640
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   641
				if ( ch > 0x10FFFF ) {
slouken@1501
   642
					ch = UNKNOWN_UNICODE;
slouken@1501
   643
				}
slouken@1501
   644
				if ( ch < 0x10000 ) {
slouken@1501
   645
					if ( dstlen < 2 ) {
slouken@1501
   646
						return SDL_ICONV_E2BIG;
slouken@1501
   647
					}
slouken@1501
   648
					p[1] = (Uint8)(ch >> 8);
slouken@1501
   649
					p[0] = (Uint8)ch;
slouken@1501
   650
					dst += 2;
slouken@1501
   651
					dstlen -= 2;
slouken@1501
   652
				} else {
slouken@1501
   653
					Uint16 W1, W2;
slouken@1501
   654
					if ( dstlen < 4 ) {
slouken@1501
   655
						return SDL_ICONV_E2BIG;
slouken@1501
   656
					}
slouken@1501
   657
					ch = ch - 0x10000;
slouken@1501
   658
					W1 = 0xD800 | (Uint16)((ch >> 10) & 0x3FF);
slouken@1501
   659
					W2 = 0xDC00 | (Uint16)(ch & 0x3FF);
slouken@1501
   660
					p[1] = (Uint8)(W1 >> 8);
slouken@1501
   661
					p[0] = (Uint8)W1;
slouken@1501
   662
					p[3] = (Uint8)(W2 >> 8);
slouken@1501
   663
					p[2] = (Uint8)W2;
slouken@1501
   664
					dst += 4;
slouken@1501
   665
					dstlen -= 4;
slouken@1501
   666
				}
slouken@1501
   667
			}
slouken@1501
   668
			break;
slouken@1501
   669
		    case ENCODING_UTF32BE:
slouken@1501
   670
			{
slouken@1501
   671
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   672
				if ( ch > 0x7FFFFFFF ) {
slouken@1501
   673
					ch = UNKNOWN_UNICODE;
slouken@1501
   674
				}
slouken@1501
   675
				if ( dstlen < 4 ) {
slouken@1501
   676
					return SDL_ICONV_E2BIG;
slouken@1501
   677
				}
slouken@1501
   678
				p[0] = (Uint8)(ch >> 24);
slouken@1501
   679
				p[1] = (Uint8)(ch >> 16);
slouken@1501
   680
				p[2] = (Uint8)(ch >> 8);
slouken@1501
   681
				p[3] = (Uint8)ch;
slouken@1501
   682
				dst += 4;
slouken@1501
   683
				dstlen -= 4;
slouken@1501
   684
			}
slouken@1501
   685
			break;
slouken@1501
   686
		    case ENCODING_UTF32LE:
slouken@1501
   687
			{
slouken@1501
   688
				Uint8 *p = (Uint8 *)dst;
slouken@1501
   689
				if ( ch > 0x7FFFFFFF ) {
slouken@1501
   690
					ch = UNKNOWN_UNICODE;
slouken@1501
   691
				}
slouken@1501
   692
				if ( dstlen < 4 ) {
slouken@1501
   693
					return SDL_ICONV_E2BIG;
slouken@1501
   694
				}
slouken@1501
   695
				p[3] = (Uint8)(ch >> 24);
slouken@1501
   696
				p[2] = (Uint8)(ch >> 16);
slouken@1501
   697
				p[1] = (Uint8)(ch >> 8);
slouken@1501
   698
				p[0] = (Uint8)ch;
slouken@1501
   699
				dst += 4;
slouken@1501
   700
				dstlen -= 4;
slouken@1501
   701
			}
slouken@1501
   702
			break;
slouken@1501
   703
		    case ENCODING_UCS2:
slouken@1501
   704
			{
slouken@1501
   705
				Uint16 *p = (Uint16 *)dst;
slouken@1501
   706
				if ( ch > 0xFFFF ) {
slouken@1501
   707
					ch = UNKNOWN_UNICODE;
slouken@1501
   708
				}
slouken@1501
   709
				if ( dstlen < 2 ) {
slouken@1501
   710
					return SDL_ICONV_E2BIG;
slouken@1501
   711
				}
slouken@1501
   712
				*p = (Uint16)ch;
slouken@1501
   713
				dst += 2;
slouken@1501
   714
				dstlen -= 2;
slouken@1501
   715
			}
slouken@1501
   716
			break;
slouken@1501
   717
		    case ENCODING_UCS4:
slouken@1501
   718
			{
slouken@1501
   719
				Uint32 *p = (Uint32 *)dst;
slouken@1501
   720
				if ( ch > 0x7FFFFFFF ) {
slouken@1501
   721
					ch = UNKNOWN_UNICODE;
slouken@1501
   722
				}
slouken@1501
   723
				if ( dstlen < 4 ) {
slouken@1501
   724
					return SDL_ICONV_E2BIG;
slouken@1501
   725
				}
slouken@1501
   726
				*p = ch;
slouken@1501
   727
				dst += 4;
slouken@1501
   728
				dstlen -= 4;
slouken@1501
   729
			}
slouken@1501
   730
			break;
slouken@1501
   731
		}
slouken@1501
   732
slouken@1501
   733
		/* Update state */
slouken@1501
   734
		*inbuf = src;
slouken@1501
   735
		*inbytesleft = srclen;
slouken@1501
   736
		*outbuf = dst;
slouken@1501
   737
		*outbytesleft = dstlen;
slouken@1501
   738
		++total;
slouken@1501
   739
	}
slouken@1501
   740
	return total;
slouken@1501
   741
}
slouken@1501
   742
slouken@1501
   743
int SDL_iconv_close(SDL_iconv_t cd)
slouken@1501
   744
{
slouken@1501
   745
	if ( cd && cd != (SDL_iconv_t)-1 ) {
slouken@1501
   746
		SDL_free(cd);
slouken@1501
   747
	}
slouken@1501
   748
	return 0;
slouken@1501
   749
}
slouken@1501
   750
slouken@1501
   751
#endif /* !HAVE_ICONV */
slouken@1501
   752
slouken@1501
   753
char *SDL_iconv_string(const char *tocode, const char *fromcode, char *inbuf, size_t inbytesleft)
slouken@1501
   754
{
slouken@1501
   755
	SDL_iconv_t cd;
slouken@1501
   756
	char *string;
slouken@1501
   757
	size_t stringsize;
slouken@1501
   758
	char *outbuf;
slouken@1501
   759
	size_t outbytesleft;
slouken@1501
   760
	size_t retCode = 0;
slouken@1501
   761
slouken@1501
   762
	cd = SDL_iconv_open(tocode, fromcode);
slouken@1501
   763
	if ( cd == (SDL_iconv_t)-1 ) {
slouken@1501
   764
		return NULL;
slouken@1501
   765
	}
slouken@1501
   766
slouken@1501
   767
	stringsize = inbytesleft > 4 ? inbytesleft : 4;
slouken@1501
   768
	string = SDL_malloc(stringsize);
slouken@1501
   769
	if ( !string ) {
slouken@1501
   770
		SDL_iconv_close(cd);
slouken@1501
   771
		return NULL;
slouken@1501
   772
	}
slouken@1501
   773
	outbuf = string;
slouken@1501
   774
	outbytesleft = stringsize;
slouken@1501
   775
	SDL_memset(outbuf, 0, 4);
slouken@1501
   776
slouken@1501
   777
	while ( inbytesleft > 0 ) {
slouken@1501
   778
		retCode = SDL_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
slouken@1501
   779
		switch (retCode) {
slouken@1501
   780
		    case SDL_ICONV_E2BIG:
slouken@1501
   781
			{
slouken@1501
   782
				char *oldstring = string;
slouken@1501
   783
				stringsize *= 2;
slouken@1501
   784
				string = SDL_realloc(string, stringsize);
slouken@1501
   785
				if ( !string ) {
slouken@1501
   786
					SDL_iconv_close(cd);
slouken@1501
   787
					return NULL;
slouken@1501
   788
				}
slouken@1501
   789
				outbuf = string + (outbuf - oldstring);
slouken@1501
   790
				outbytesleft = stringsize - (outbuf - string);
slouken@1501
   791
				SDL_memset(outbuf, 0, 4);
slouken@1501
   792
			}
slouken@1501
   793
			break;
slouken@1501
   794
		    case SDL_ICONV_EILSEQ:
slouken@1501
   795
			/* Try skipping some input data - not perfect, but... */
slouken@1501
   796
			++inbuf;
slouken@1501
   797
			--inbytesleft;
slouken@1501
   798
			break;
slouken@1501
   799
		    case SDL_ICONV_EINVAL:
slouken@1501
   800
		    case SDL_ICONV_ERROR:
slouken@1501
   801
			/* We can't continue... */
slouken@1501
   802
			inbytesleft = 0;
slouken@1501
   803
			break;
slouken@1501
   804
		}
slouken@1501
   805
	}
slouken@1501
   806
	SDL_iconv_close(cd);
slouken@1501
   807
slouken@1501
   808
	return string;
slouken@1501
   809
}