src/stdlib/SDL_iconv.c
author Sam Lantinga
Thu, 28 Jun 2007 06:47:35 +0000
branchSDL-1.2
changeset 3985 2f8efcf14c83
parent 3984 b74530a1dad6
child 3986 4f73308bbb32
permissions -rw-r--r--
Okay, apparently the newer standard specifies char** for the inbuf
parameter to iconv()
(See http://lists.debian.org/debian-glibc/2004/05/msg00006.html)

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