src/stdlib/SDL_iconv.c
author Sam Lantinga <slouken@libsdl.org>
Wed, 04 Jul 2007 08:01:04 +0000
changeset 2143 e906da4414a3
parent 2142 ca80c942e69c
child 2182 cc2597da0840
permissions -rw-r--r--
Fix for bug #447 merged from SDL 1.2
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@2136
    31
/* Depending on which standard the iconv() was implemented with,
slouken@2136
    32
   iconv() may or may not use const char ** for the inbuf param.
slouken@2136
    33
   If we get this wrong, it's just a warning, so no big deal.
slouken@2136
    34
*/
slouken@2136
    35
#if defined(_XGP6) || \
slouken@2136
    36
    defined(__GLIBC__) && ((__GLIBC__ > 2) || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 2))
slouken@2136
    37
#define ICONV_INBUF_NONCONST
slouken@2136
    38
#endif
slouken@2136
    39
slouken@1501
    40
#include <errno.h>
slouken@1501
    41
slouken@1895
    42
size_t
slouken@1895
    43
SDL_iconv(SDL_iconv_t cd,
slouken@2135
    44
          const char **inbuf, size_t * inbytesleft,
slouken@1895
    45
          char **outbuf, size_t * outbytesleft)
slouken@1501
    46
{
slouken@2135
    47
    size_t retCode;
slouken@2136
    48
#ifdef ICONV_INBUF_NONCONST
slouken@2136
    49
    retCode = iconv(cd, (char **) inbuf, inbytesleft, outbuf, outbytesleft);
slouken@2135
    50
#else
slouken@2136
    51
    retCode = iconv(cd, inbuf, inbytesleft, outbuf, outbytesleft);
slouken@2135
    52
#endif
slouken@1895
    53
    if (retCode == (size_t) - 1) {
slouken@1895
    54
        switch (errno) {
slouken@1895
    55
        case E2BIG:
slouken@1895
    56
            return SDL_ICONV_E2BIG;
slouken@1895
    57
        case EILSEQ:
slouken@1895
    58
            return SDL_ICONV_EILSEQ;
slouken@1895
    59
        case EINVAL:
slouken@1895
    60
            return SDL_ICONV_EINVAL;
slouken@1895
    61
        default:
slouken@1895
    62
            return SDL_ICONV_ERROR;
slouken@1895
    63
        }
slouken@1895
    64
    }
slouken@1895
    65
    return retCode;
slouken@1501
    66
}
slouken@1501
    67
slouken@1501
    68
#else
slouken@1501
    69
slouken@1503
    70
/* Lots of useful information on Unicode at:
slouken@1503
    71
	http://www.cl.cam.ac.uk/~mgk25/unicode.html
slouken@1503
    72
*/
slouken@1503
    73
slouken@1501
    74
#define UNICODE_BOM	0xFEFF
slouken@1501
    75
slouken@1501
    76
#define UNKNOWN_ASCII	'?'
slouken@1501
    77
#define UNKNOWN_UNICODE	0xFFFD
slouken@1501
    78
slouken@1895
    79
enum
slouken@1895
    80
{
slouken@1895
    81
    ENCODING_UNKNOWN,
slouken@1895
    82
    ENCODING_ASCII,
slouken@1895
    83
    ENCODING_LATIN1,
slouken@1895
    84
    ENCODING_UTF8,
slouken@1895
    85
    ENCODING_UTF16,             /* Needs byte order marker */
slouken@1895
    86
    ENCODING_UTF16BE,
slouken@1895
    87
    ENCODING_UTF16LE,
slouken@1895
    88
    ENCODING_UTF32,             /* Needs byte order marker */
slouken@1895
    89
    ENCODING_UTF32BE,
slouken@1895
    90
    ENCODING_UTF32LE,
slouken@1895
    91
    ENCODING_UCS2,              /* Native byte order assumed */
slouken@1895
    92
    ENCODING_UCS4,              /* Native byte order assumed */
slouken@1501
    93
};
slouken@1501
    94
#if SDL_BYTEORDER == SDL_BIG_ENDIAN
slouken@1501
    95
#define ENCODING_UTF16NATIVE	ENCODING_UTF16BE
slouken@1501
    96
#define ENCODING_UTF32NATIVE	ENCODING_UTF32BE
slouken@1501
    97
#else
slouken@1501
    98
#define ENCODING_UTF16NATIVE	ENCODING_UTF16LE
slouken@1501
    99
#define ENCODING_UTF32NATIVE	ENCODING_UTF32LE
slouken@1501
   100
#endif
slouken@1501
   101
slouken@1501
   102
struct _SDL_iconv_t
slouken@1501
   103
{
slouken@1895
   104
    int src_fmt;
slouken@1895
   105
    int dst_fmt;
slouken@1501
   106
};
slouken@1501
   107
slouken@1895
   108
static struct
slouken@1895
   109
{
slouken@1895
   110
    const char *name;
slouken@1895
   111
    int format;
slouken@1501
   112
} encodings[] = {
slouken@2142
   113
/* *INDENT-OFF* */
slouken@2142
   114
    { "646", ENCODING_ASCII },
slouken@2142
   115
    { "ASCII", ENCODING_ASCII },
slouken@2142
   116
    { "US-ASCII", ENCODING_ASCII },
slouken@2142
   117
    { "LATIN1", ENCODING_LATIN1 },
slouken@2142
   118
    { "8859-1", ENCODING_LATIN1 },
slouken@2142
   119
    { "ISO-8859-1", ENCODING_LATIN1 },
slouken@2142
   120
    { "UTF8", ENCODING_UTF8 },
slouken@2142
   121
    { "UTF-8", ENCODING_UTF8 },
slouken@2142
   122
    { "UTF16", ENCODING_UTF16 },
slouken@2142
   123
    { "UTF-16", ENCODING_UTF16 },
slouken@2142
   124
    { "UTF16BE", ENCODING_UTF16BE },
slouken@2142
   125
    { "UTF-16BE", ENCODING_UTF16BE },
slouken@2142
   126
    { "UTF16LE", ENCODING_UTF16LE },
slouken@2142
   127
    { "UTF-16LE", ENCODING_UTF16LE },
slouken@2142
   128
    { "UTF32", ENCODING_UTF32 },
slouken@2142
   129
    { "UTF-32", ENCODING_UTF32 },
slouken@2142
   130
    { "UTF32BE", ENCODING_UTF32BE },
slouken@2142
   131
    { "UTF-32BE", ENCODING_UTF32BE },
slouken@2142
   132
    { "UTF32LE", ENCODING_UTF32LE },
slouken@2142
   133
    { "UTF-32LE", ENCODING_UTF32LE },
slouken@2142
   134
    { "UCS2", ENCODING_UCS2 },
slouken@2142
   135
    { "UCS-2", ENCODING_UCS2 },
slouken@2142
   136
    { "UCS4", ENCODING_UCS4 },
slouken@2142
   137
    { "UCS-4", ENCODING_UCS4 },
slouken@2142
   138
/* *INDENT-ON* */
slouken@2142
   139
};
slouken@1501
   140
slouken@1895
   141
SDL_iconv_t
slouken@1895
   142
SDL_iconv_open(const char *tocode, const char *fromcode)
slouken@1501
   143
{
slouken@1895
   144
    int src_fmt = ENCODING_UNKNOWN;
slouken@1895
   145
    int dst_fmt = ENCODING_UNKNOWN;
slouken@1895
   146
    int i;
slouken@1501
   147
slouken@1895
   148
    for (i = 0; i < SDL_arraysize(encodings); ++i) {
slouken@1895
   149
        if (SDL_strcasecmp(fromcode, encodings[i].name) == 0) {
slouken@1895
   150
            src_fmt = encodings[i].format;
slouken@1895
   151
            if (dst_fmt != ENCODING_UNKNOWN) {
slouken@1895
   152
                break;
slouken@1895
   153
            }
slouken@1895
   154
        }
slouken@1895
   155
        if (SDL_strcasecmp(tocode, encodings[i].name) == 0) {
slouken@1895
   156
            dst_fmt = encodings[i].format;
slouken@1895
   157
            if (src_fmt != ENCODING_UNKNOWN) {
slouken@1895
   158
                break;
slouken@1895
   159
            }
slouken@1895
   160
        }
slouken@1895
   161
    }
slouken@1895
   162
    if (src_fmt != ENCODING_UNKNOWN && dst_fmt != ENCODING_UNKNOWN) {
slouken@1895
   163
        SDL_iconv_t cd = (SDL_iconv_t) SDL_malloc(sizeof(*cd));
slouken@1895
   164
        if (cd) {
slouken@1895
   165
            cd->src_fmt = src_fmt;
slouken@1895
   166
            cd->dst_fmt = dst_fmt;
slouken@1895
   167
            return cd;
slouken@1895
   168
        }
slouken@1895
   169
    }
slouken@1895
   170
    return (SDL_iconv_t) - 1;
slouken@1501
   171
}
slouken@1501
   172
slouken@1895
   173
size_t
slouken@1895
   174
SDL_iconv(SDL_iconv_t cd,
slouken@2135
   175
          const char **inbuf, size_t * inbytesleft,
slouken@1895
   176
          char **outbuf, size_t * outbytesleft)
slouken@1501
   177
{
slouken@1895
   178
    /* For simplicity, we'll convert everything to and from UCS-4 */
slouken@2135
   179
    const char *src;
slouken@2135
   180
    char *dst;
slouken@1895
   181
    size_t srclen, dstlen;
slouken@2135
   182
    Uint32 ch = 0;
slouken@1895
   183
    size_t total;
slouken@1501
   184
slouken@1895
   185
    if (!inbuf || !*inbuf) {
slouken@1895
   186
        /* Reset the context */
slouken@1895
   187
        return 0;
slouken@1895
   188
    }
slouken@1895
   189
    if (!outbuf || !*outbuf || !outbytesleft || !*outbytesleft) {
slouken@1895
   190
        return SDL_ICONV_E2BIG;
slouken@1895
   191
    }
slouken@1895
   192
    src = *inbuf;
slouken@1895
   193
    srclen = (inbytesleft ? *inbytesleft : 0);
slouken@1895
   194
    dst = *outbuf;
slouken@1895
   195
    dstlen = *outbytesleft;
slouken@1501
   196
slouken@1895
   197
    switch (cd->src_fmt) {
slouken@1895
   198
    case ENCODING_UTF16:
slouken@1895
   199
        /* Scan for a byte order marker */
slouken@1895
   200
        {
slouken@1895
   201
            Uint8 *p = (Uint8 *) src;
slouken@1895
   202
            size_t n = srclen / 2;
slouken@1895
   203
            while (n) {
slouken@1895
   204
                if (p[0] == 0xFF && p[1] == 0xFE) {
slouken@1895
   205
                    cd->src_fmt = ENCODING_UTF16BE;
slouken@1895
   206
                    break;
slouken@1895
   207
                } else if (p[0] == 0xFE && p[1] == 0xFF) {
slouken@1895
   208
                    cd->src_fmt = ENCODING_UTF16LE;
slouken@1895
   209
                    break;
slouken@1895
   210
                }
slouken@1895
   211
                p += 2;
slouken@1895
   212
                --n;
slouken@1895
   213
            }
slouken@1895
   214
            if (n == 0) {
slouken@1895
   215
                /* We can't tell, default to host order */
slouken@1895
   216
                cd->src_fmt = ENCODING_UTF16NATIVE;
slouken@1895
   217
            }
slouken@1895
   218
        }
slouken@1895
   219
        break;
slouken@1895
   220
    case ENCODING_UTF32:
slouken@1895
   221
        /* Scan for a byte order marker */
slouken@1895
   222
        {
slouken@1895
   223
            Uint8 *p = (Uint8 *) src;
slouken@1895
   224
            size_t n = srclen / 4;
slouken@1895
   225
            while (n) {
slouken@1895
   226
                if (p[0] == 0xFF && p[1] == 0xFE &&
slouken@1895
   227
                    p[2] == 0x00 && p[3] == 0x00) {
slouken@1895
   228
                    cd->src_fmt = ENCODING_UTF32BE;
slouken@1895
   229
                    break;
slouken@1895
   230
                } else if (p[0] == 0x00 && p[1] == 0x00 &&
slouken@1895
   231
                           p[2] == 0xFE && p[3] == 0xFF) {
slouken@1895
   232
                    cd->src_fmt = ENCODING_UTF32LE;
slouken@1895
   233
                    break;
slouken@1895
   234
                }
slouken@1895
   235
                p += 4;
slouken@1895
   236
                --n;
slouken@1895
   237
            }
slouken@1895
   238
            if (n == 0) {
slouken@1895
   239
                /* We can't tell, default to host order */
slouken@1895
   240
                cd->src_fmt = ENCODING_UTF32NATIVE;
slouken@1895
   241
            }
slouken@1895
   242
        }
slouken@1895
   243
        break;
slouken@1895
   244
    }
slouken@1501
   245
slouken@1895
   246
    switch (cd->dst_fmt) {
slouken@1895
   247
    case ENCODING_UTF16:
slouken@1895
   248
        /* Default to host order, need to add byte order marker */
slouken@1895
   249
        if (dstlen < 2) {
slouken@1895
   250
            return SDL_ICONV_E2BIG;
slouken@1895
   251
        }
slouken@1895
   252
        *(Uint16 *) dst = UNICODE_BOM;
slouken@1895
   253
        dst += 2;
slouken@1895
   254
        dstlen -= 2;
slouken@1895
   255
        cd->dst_fmt = ENCODING_UTF16NATIVE;
slouken@1895
   256
        break;
slouken@1895
   257
    case ENCODING_UTF32:
slouken@1895
   258
        /* Default to host order, need to add byte order marker */
slouken@1895
   259
        if (dstlen < 4) {
slouken@1895
   260
            return SDL_ICONV_E2BIG;
slouken@1895
   261
        }
slouken@1895
   262
        *(Uint32 *) dst = UNICODE_BOM;
slouken@1895
   263
        dst += 4;
slouken@1895
   264
        dstlen -= 4;
slouken@1895
   265
        cd->dst_fmt = ENCODING_UTF32NATIVE;
slouken@1895
   266
        break;
slouken@1895
   267
    }
slouken@1501
   268
slouken@1895
   269
    total = 0;
slouken@1895
   270
    while (srclen > 0) {
slouken@1895
   271
        /* Decode a character */
slouken@1895
   272
        switch (cd->src_fmt) {
slouken@1895
   273
        case ENCODING_ASCII:
slouken@1895
   274
            {
slouken@1895
   275
                Uint8 *p = (Uint8 *) src;
slouken@1895
   276
                ch = (Uint32) (p[0] & 0x7F);
slouken@1895
   277
                ++src;
slouken@1895
   278
                --srclen;
slouken@1895
   279
            }
slouken@1895
   280
            break;
slouken@1895
   281
        case ENCODING_LATIN1:
slouken@1895
   282
            {
slouken@1895
   283
                Uint8 *p = (Uint8 *) src;
slouken@1895
   284
                ch = (Uint32) p[0];
slouken@1895
   285
                ++src;
slouken@1895
   286
                --srclen;
slouken@1895
   287
            }
slouken@1895
   288
            break;
slouken@1895
   289
        case ENCODING_UTF8:    /* RFC 3629 */
slouken@1895
   290
            {
slouken@1895
   291
                Uint8 *p = (Uint8 *) src;
slouken@1895
   292
                size_t left = 0;
slouken@1895
   293
                SDL_bool overlong = SDL_FALSE;
slouken@1895
   294
                if (p[0] >= 0xFC) {
slouken@1895
   295
                    if ((p[0] & 0xFE) != 0xFC) {
slouken@1895
   296
                        /* Skip illegal sequences
slouken@1895
   297
                           return SDL_ICONV_EILSEQ;
slouken@1895
   298
                         */
slouken@1895
   299
                        ch = UNKNOWN_UNICODE;
slouken@1895
   300
                    } else {
slouken@1895
   301
                        if (p[0] == 0xFC) {
slouken@1895
   302
                            overlong = SDL_TRUE;
slouken@1895
   303
                        }
slouken@1895
   304
                        ch = (Uint32) (p[0] & 0x01);
slouken@1895
   305
                        left = 5;
slouken@1895
   306
                    }
slouken@1895
   307
                } else if (p[0] >= 0xF8) {
slouken@1895
   308
                    if ((p[0] & 0xFC) != 0xF8) {
slouken@1895
   309
                        /* Skip illegal sequences
slouken@1895
   310
                           return SDL_ICONV_EILSEQ;
slouken@1895
   311
                         */
slouken@1895
   312
                        ch = UNKNOWN_UNICODE;
slouken@1895
   313
                    } else {
slouken@1895
   314
                        if (p[0] == 0xF8) {
slouken@1895
   315
                            overlong = SDL_TRUE;
slouken@1895
   316
                        }
slouken@1895
   317
                        ch = (Uint32) (p[0] & 0x03);
slouken@1895
   318
                        left = 4;
slouken@1895
   319
                    }
slouken@1895
   320
                } else if (p[0] >= 0xF0) {
slouken@1895
   321
                    if ((p[0] & 0xF8) != 0xF0) {
slouken@1895
   322
                        /* Skip illegal sequences
slouken@1895
   323
                           return SDL_ICONV_EILSEQ;
slouken@1895
   324
                         */
slouken@1895
   325
                        ch = UNKNOWN_UNICODE;
slouken@1895
   326
                    } else {
slouken@1895
   327
                        if (p[0] == 0xF0) {
slouken@1895
   328
                            overlong = SDL_TRUE;
slouken@1895
   329
                        }
slouken@1895
   330
                        ch = (Uint32) (p[0] & 0x07);
slouken@1895
   331
                        left = 3;
slouken@1895
   332
                    }
slouken@1895
   333
                } else if (p[0] >= 0xE0) {
slouken@1895
   334
                    if ((p[0] & 0xF0) != 0xE0) {
slouken@1895
   335
                        /* Skip illegal sequences
slouken@1895
   336
                           return SDL_ICONV_EILSEQ;
slouken@1895
   337
                         */
slouken@1895
   338
                        ch = UNKNOWN_UNICODE;
slouken@1895
   339
                    } else {
slouken@1895
   340
                        if (p[0] == 0xE0) {
slouken@1895
   341
                            overlong = SDL_TRUE;
slouken@1895
   342
                        }
slouken@1895
   343
                        ch = (Uint32) (p[0] & 0x0F);
slouken@1895
   344
                        left = 2;
slouken@1895
   345
                    }
slouken@1895
   346
                } else if (p[0] >= 0xC0) {
slouken@1895
   347
                    if ((p[0] & 0xE0) != 0xC0) {
slouken@1895
   348
                        /* Skip illegal sequences
slouken@1895
   349
                           return SDL_ICONV_EILSEQ;
slouken@1895
   350
                         */
slouken@1895
   351
                        ch = UNKNOWN_UNICODE;
slouken@1895
   352
                    } else {
slouken@1895
   353
                        if ((p[0] & 0xCE) == 0xC0) {
slouken@1895
   354
                            overlong = SDL_TRUE;
slouken@1895
   355
                        }
slouken@1895
   356
                        ch = (Uint32) (p[0] & 0x1F);
slouken@1895
   357
                        left = 1;
slouken@1895
   358
                    }
slouken@1895
   359
                } else {
slouken@1895
   360
                    if ((p[0] & 0x80) != 0x00) {
slouken@1895
   361
                        /* Skip illegal sequences
slouken@1895
   362
                           return SDL_ICONV_EILSEQ;
slouken@1895
   363
                         */
slouken@1895
   364
                        ch = UNKNOWN_UNICODE;
slouken@1895
   365
                    } else {
slouken@1895
   366
                        ch = (Uint32) p[0];
slouken@1895
   367
                    }
slouken@1895
   368
                }
slouken@1895
   369
                ++src;
slouken@1895
   370
                --srclen;
slouken@1895
   371
                if (srclen < left) {
slouken@1895
   372
                    return SDL_ICONV_EINVAL;
slouken@1895
   373
                }
slouken@1895
   374
                while (left--) {
slouken@1895
   375
                    ++p;
slouken@1895
   376
                    if ((p[0] & 0xC0) != 0x80) {
slouken@1895
   377
                        /* Skip illegal sequences
slouken@1895
   378
                           return SDL_ICONV_EILSEQ;
slouken@1895
   379
                         */
slouken@1895
   380
                        ch = UNKNOWN_UNICODE;
slouken@1895
   381
                        break;
slouken@1895
   382
                    }
slouken@1895
   383
                    ch <<= 6;
slouken@1895
   384
                    ch |= (p[0] & 0x3F);
slouken@1895
   385
                    ++src;
slouken@1895
   386
                    --srclen;
slouken@1895
   387
                }
slouken@1895
   388
                if (overlong) {
slouken@1895
   389
                    /* Potential security risk
slouken@1895
   390
                       return SDL_ICONV_EILSEQ;
slouken@1895
   391
                     */
slouken@1895
   392
                    ch = UNKNOWN_UNICODE;
slouken@1895
   393
                }
slouken@1895
   394
                if ((ch >= 0xD800 && ch <= 0xDFFF) ||
slouken@1895
   395
                    (ch == 0xFFFE || ch == 0xFFFF) || ch > 0x10FFFF) {
slouken@1895
   396
                    /* Skip illegal sequences
slouken@1895
   397
                       return SDL_ICONV_EILSEQ;
slouken@1895
   398
                     */
slouken@1895
   399
                    ch = UNKNOWN_UNICODE;
slouken@1895
   400
                }
slouken@1895
   401
            }
slouken@1895
   402
            break;
slouken@1895
   403
        case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1895
   404
            {
slouken@1895
   405
                Uint8 *p = (Uint8 *) src;
slouken@1895
   406
                Uint16 W1, W2;
slouken@1895
   407
                if (srclen < 2) {
slouken@1895
   408
                    return SDL_ICONV_EINVAL;
slouken@1895
   409
                }
slouken@1895
   410
                W1 = ((Uint16) p[0] << 8) | (Uint16) p[1];
slouken@1895
   411
                src += 2;
slouken@1895
   412
                srclen -= 2;
slouken@1895
   413
                if (W1 < 0xD800 || W1 > 0xDFFF) {
slouken@1895
   414
                    ch = (Uint32) W1;
slouken@1895
   415
                    break;
slouken@1895
   416
                }
slouken@1895
   417
                if (W1 > 0xDBFF) {
slouken@1895
   418
                    /* Skip illegal sequences
slouken@1895
   419
                       return SDL_ICONV_EILSEQ;
slouken@1895
   420
                     */
slouken@1895
   421
                    ch = UNKNOWN_UNICODE;
slouken@1895
   422
                    break;
slouken@1895
   423
                }
slouken@1895
   424
                if (srclen < 2) {
slouken@1895
   425
                    return SDL_ICONV_EINVAL;
slouken@1895
   426
                }
slouken@1895
   427
                p = (Uint8 *) src;
slouken@1895
   428
                W2 = ((Uint16) p[0] << 8) | (Uint16) p[1];
slouken@1895
   429
                src += 2;
slouken@1895
   430
                srclen -= 2;
slouken@1895
   431
                if (W2 < 0xDC00 || W2 > 0xDFFF) {
slouken@1895
   432
                    /* Skip illegal sequences
slouken@1895
   433
                       return SDL_ICONV_EILSEQ;
slouken@1895
   434
                     */
slouken@1895
   435
                    ch = UNKNOWN_UNICODE;
slouken@1895
   436
                    break;
slouken@1895
   437
                }
slouken@1895
   438
                ch = (((Uint32) (W1 & 0x3FF) << 10) |
slouken@1895
   439
                      (Uint32) (W2 & 0x3FF)) + 0x10000;
slouken@1895
   440
            }
slouken@1895
   441
            break;
slouken@1895
   442
        case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1895
   443
            {
slouken@1895
   444
                Uint8 *p = (Uint8 *) src;
slouken@1895
   445
                Uint16 W1, W2;
slouken@1895
   446
                if (srclen < 2) {
slouken@1895
   447
                    return SDL_ICONV_EINVAL;
slouken@1895
   448
                }
slouken@1895
   449
                W1 = ((Uint16) p[1] << 8) | (Uint16) p[0];
slouken@1895
   450
                src += 2;
slouken@1895
   451
                srclen -= 2;
slouken@1895
   452
                if (W1 < 0xD800 || W1 > 0xDFFF) {
slouken@1895
   453
                    ch = (Uint32) W1;
slouken@1895
   454
                    break;
slouken@1895
   455
                }
slouken@1895
   456
                if (W1 > 0xDBFF) {
slouken@1895
   457
                    /* Skip illegal sequences
slouken@1895
   458
                       return SDL_ICONV_EILSEQ;
slouken@1895
   459
                     */
slouken@1895
   460
                    ch = UNKNOWN_UNICODE;
slouken@1895
   461
                    break;
slouken@1895
   462
                }
slouken@1895
   463
                if (srclen < 2) {
slouken@1895
   464
                    return SDL_ICONV_EINVAL;
slouken@1895
   465
                }
slouken@1895
   466
                p = (Uint8 *) src;
slouken@1895
   467
                W2 = ((Uint16) p[1] << 8) | (Uint16) p[0];
slouken@1895
   468
                src += 2;
slouken@1895
   469
                srclen -= 2;
slouken@1895
   470
                if (W2 < 0xDC00 || W2 > 0xDFFF) {
slouken@1895
   471
                    /* Skip illegal sequences
slouken@1895
   472
                       return SDL_ICONV_EILSEQ;
slouken@1895
   473
                     */
slouken@1895
   474
                    ch = UNKNOWN_UNICODE;
slouken@1895
   475
                    break;
slouken@1895
   476
                }
slouken@1895
   477
                ch = (((Uint32) (W1 & 0x3FF) << 10) |
slouken@1895
   478
                      (Uint32) (W2 & 0x3FF)) + 0x10000;
slouken@1895
   479
            }
slouken@1895
   480
            break;
slouken@1895
   481
        case ENCODING_UTF32BE:
slouken@1895
   482
            {
slouken@1895
   483
                Uint8 *p = (Uint8 *) src;
slouken@1895
   484
                if (srclen < 4) {
slouken@1895
   485
                    return SDL_ICONV_EINVAL;
slouken@1895
   486
                }
slouken@1895
   487
                ch = ((Uint32) p[0] << 24) |
slouken@1895
   488
                    ((Uint32) p[1] << 16) |
slouken@1895
   489
                    ((Uint32) p[2] << 8) | (Uint32) p[3];
slouken@1895
   490
                src += 4;
slouken@1895
   491
                srclen -= 4;
slouken@1895
   492
            }
slouken@1895
   493
            break;
slouken@1895
   494
        case ENCODING_UTF32LE:
slouken@1895
   495
            {
slouken@1895
   496
                Uint8 *p = (Uint8 *) src;
slouken@1895
   497
                if (srclen < 4) {
slouken@1895
   498
                    return SDL_ICONV_EINVAL;
slouken@1895
   499
                }
slouken@1895
   500
                ch = ((Uint32) p[3] << 24) |
slouken@1895
   501
                    ((Uint32) p[2] << 16) |
slouken@1895
   502
                    ((Uint32) p[1] << 8) | (Uint32) p[0];
slouken@1895
   503
                src += 4;
slouken@1895
   504
                srclen -= 4;
slouken@1895
   505
            }
slouken@1895
   506
            break;
slouken@1895
   507
        case ENCODING_UCS2:
slouken@1895
   508
            {
slouken@1895
   509
                Uint16 *p = (Uint16 *) src;
slouken@1895
   510
                if (srclen < 2) {
slouken@1895
   511
                    return SDL_ICONV_EINVAL;
slouken@1895
   512
                }
slouken@1895
   513
                ch = *p;
slouken@1895
   514
                src += 2;
slouken@1895
   515
                srclen -= 2;
slouken@1895
   516
            }
slouken@1895
   517
            break;
slouken@1895
   518
        case ENCODING_UCS4:
slouken@1895
   519
            {
slouken@1895
   520
                Uint32 *p = (Uint32 *) src;
slouken@1895
   521
                if (srclen < 4) {
slouken@1895
   522
                    return SDL_ICONV_EINVAL;
slouken@1895
   523
                }
slouken@1895
   524
                ch = *p;
slouken@1895
   525
                src += 4;
slouken@1895
   526
                srclen -= 4;
slouken@1895
   527
            }
slouken@1895
   528
            break;
slouken@1895
   529
        }
slouken@1501
   530
slouken@1895
   531
        /* Encode a character */
slouken@1895
   532
        switch (cd->dst_fmt) {
slouken@1895
   533
        case ENCODING_ASCII:
slouken@1895
   534
            {
slouken@1895
   535
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   536
                if (dstlen < 1) {
slouken@1895
   537
                    return SDL_ICONV_E2BIG;
slouken@1895
   538
                }
slouken@1895
   539
                if (ch > 0x7F) {
slouken@1895
   540
                    *p = UNKNOWN_ASCII;
slouken@1895
   541
                } else {
slouken@1895
   542
                    *p = (Uint8) ch;
slouken@1895
   543
                }
slouken@1895
   544
                ++dst;
slouken@1895
   545
                --dstlen;
slouken@1895
   546
            }
slouken@1895
   547
            break;
slouken@1895
   548
        case ENCODING_LATIN1:
slouken@1895
   549
            {
slouken@1895
   550
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   551
                if (dstlen < 1) {
slouken@1895
   552
                    return SDL_ICONV_E2BIG;
slouken@1895
   553
                }
slouken@1895
   554
                if (ch > 0xFF) {
slouken@1895
   555
                    *p = UNKNOWN_ASCII;
slouken@1895
   556
                } else {
slouken@1895
   557
                    *p = (Uint8) ch;
slouken@1895
   558
                }
slouken@1895
   559
                ++dst;
slouken@1895
   560
                --dstlen;
slouken@1895
   561
            }
slouken@1895
   562
            break;
slouken@1895
   563
        case ENCODING_UTF8:    /* RFC 3629 */
slouken@1895
   564
            {
slouken@1895
   565
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   566
                if (ch > 0x10FFFF) {
slouken@1895
   567
                    ch = UNKNOWN_UNICODE;
slouken@1895
   568
                }
slouken@1895
   569
                if (ch <= 0x7F) {
slouken@1895
   570
                    if (dstlen < 1) {
slouken@1895
   571
                        return SDL_ICONV_E2BIG;
slouken@1895
   572
                    }
slouken@1895
   573
                    *p = (Uint8) ch;
slouken@1895
   574
                    ++dst;
slouken@1895
   575
                    --dstlen;
slouken@1895
   576
                } else if (ch <= 0x7FF) {
slouken@1895
   577
                    if (dstlen < 2) {
slouken@1895
   578
                        return SDL_ICONV_E2BIG;
slouken@1895
   579
                    }
slouken@1895
   580
                    p[0] = 0xC0 | (Uint8) ((ch >> 6) & 0x1F);
slouken@1895
   581
                    p[1] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   582
                    dst += 2;
slouken@1895
   583
                    dstlen -= 2;
slouken@1895
   584
                } else if (ch <= 0xFFFF) {
slouken@1895
   585
                    if (dstlen < 3) {
slouken@1895
   586
                        return SDL_ICONV_E2BIG;
slouken@1895
   587
                    }
slouken@1895
   588
                    p[0] = 0xE0 | (Uint8) ((ch >> 12) & 0x0F);
slouken@1895
   589
                    p[1] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   590
                    p[2] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   591
                    dst += 3;
slouken@1895
   592
                    dstlen -= 3;
slouken@1895
   593
                } else if (ch <= 0x1FFFFF) {
slouken@1895
   594
                    if (dstlen < 4) {
slouken@1895
   595
                        return SDL_ICONV_E2BIG;
slouken@1895
   596
                    }
slouken@1895
   597
                    p[0] = 0xF0 | (Uint8) ((ch >> 18) & 0x07);
slouken@1895
   598
                    p[1] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   599
                    p[2] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   600
                    p[3] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   601
                    dst += 4;
slouken@1895
   602
                    dstlen -= 4;
slouken@1895
   603
                } else if (ch <= 0x3FFFFFF) {
slouken@1895
   604
                    if (dstlen < 5) {
slouken@1895
   605
                        return SDL_ICONV_E2BIG;
slouken@1895
   606
                    }
slouken@1895
   607
                    p[0] = 0xF8 | (Uint8) ((ch >> 24) & 0x03);
slouken@1895
   608
                    p[1] = 0x80 | (Uint8) ((ch >> 18) & 0x3F);
slouken@1895
   609
                    p[2] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   610
                    p[3] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   611
                    p[4] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   612
                    dst += 5;
slouken@1895
   613
                    dstlen -= 5;
slouken@1895
   614
                } else {
slouken@1895
   615
                    if (dstlen < 6) {
slouken@1895
   616
                        return SDL_ICONV_E2BIG;
slouken@1895
   617
                    }
slouken@1895
   618
                    p[0] = 0xFC | (Uint8) ((ch >> 30) & 0x01);
slouken@1895
   619
                    p[1] = 0x80 | (Uint8) ((ch >> 24) & 0x3F);
slouken@1895
   620
                    p[2] = 0x80 | (Uint8) ((ch >> 18) & 0x3F);
slouken@1895
   621
                    p[3] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   622
                    p[4] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   623
                    p[5] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   624
                    dst += 6;
slouken@1895
   625
                    dstlen -= 6;
slouken@1895
   626
                }
slouken@1895
   627
            }
slouken@1895
   628
            break;
slouken@1895
   629
        case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1895
   630
            {
slouken@1895
   631
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   632
                if (ch > 0x10FFFF) {
slouken@1895
   633
                    ch = UNKNOWN_UNICODE;
slouken@1895
   634
                }
slouken@1895
   635
                if (ch < 0x10000) {
slouken@1895
   636
                    if (dstlen < 2) {
slouken@1895
   637
                        return SDL_ICONV_E2BIG;
slouken@1895
   638
                    }
slouken@1895
   639
                    p[0] = (Uint8) (ch >> 8);
slouken@1895
   640
                    p[1] = (Uint8) ch;
slouken@1895
   641
                    dst += 2;
slouken@1895
   642
                    dstlen -= 2;
slouken@1895
   643
                } else {
slouken@1895
   644
                    Uint16 W1, W2;
slouken@1895
   645
                    if (dstlen < 4) {
slouken@1895
   646
                        return SDL_ICONV_E2BIG;
slouken@1895
   647
                    }
slouken@1895
   648
                    ch = ch - 0x10000;
slouken@1895
   649
                    W1 = 0xD800 | (Uint16) ((ch >> 10) & 0x3FF);
slouken@1895
   650
                    W2 = 0xDC00 | (Uint16) (ch & 0x3FF);
slouken@1895
   651
                    p[0] = (Uint8) (W1 >> 8);
slouken@1895
   652
                    p[1] = (Uint8) W1;
slouken@1895
   653
                    p[2] = (Uint8) (W2 >> 8);
slouken@1895
   654
                    p[3] = (Uint8) W2;
slouken@1895
   655
                    dst += 4;
slouken@1895
   656
                    dstlen -= 4;
slouken@1895
   657
                }
slouken@1895
   658
            }
slouken@1895
   659
            break;
slouken@1895
   660
        case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1895
   661
            {
slouken@1895
   662
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   663
                if (ch > 0x10FFFF) {
slouken@1895
   664
                    ch = UNKNOWN_UNICODE;
slouken@1895
   665
                }
slouken@1895
   666
                if (ch < 0x10000) {
slouken@1895
   667
                    if (dstlen < 2) {
slouken@1895
   668
                        return SDL_ICONV_E2BIG;
slouken@1895
   669
                    }
slouken@1895
   670
                    p[1] = (Uint8) (ch >> 8);
slouken@1895
   671
                    p[0] = (Uint8) ch;
slouken@1895
   672
                    dst += 2;
slouken@1895
   673
                    dstlen -= 2;
slouken@1895
   674
                } else {
slouken@1895
   675
                    Uint16 W1, W2;
slouken@1895
   676
                    if (dstlen < 4) {
slouken@1895
   677
                        return SDL_ICONV_E2BIG;
slouken@1895
   678
                    }
slouken@1895
   679
                    ch = ch - 0x10000;
slouken@1895
   680
                    W1 = 0xD800 | (Uint16) ((ch >> 10) & 0x3FF);
slouken@1895
   681
                    W2 = 0xDC00 | (Uint16) (ch & 0x3FF);
slouken@1895
   682
                    p[1] = (Uint8) (W1 >> 8);
slouken@1895
   683
                    p[0] = (Uint8) W1;
slouken@1895
   684
                    p[3] = (Uint8) (W2 >> 8);
slouken@1895
   685
                    p[2] = (Uint8) W2;
slouken@1895
   686
                    dst += 4;
slouken@1895
   687
                    dstlen -= 4;
slouken@1895
   688
                }
slouken@1895
   689
            }
slouken@1895
   690
            break;
slouken@1895
   691
        case ENCODING_UTF32BE:
slouken@1895
   692
            {
slouken@1895
   693
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   694
                if (ch > 0x10FFFF) {
slouken@1895
   695
                    ch = UNKNOWN_UNICODE;
slouken@1895
   696
                }
slouken@1895
   697
                if (dstlen < 4) {
slouken@1895
   698
                    return SDL_ICONV_E2BIG;
slouken@1895
   699
                }
slouken@1895
   700
                p[0] = (Uint8) (ch >> 24);
slouken@1895
   701
                p[1] = (Uint8) (ch >> 16);
slouken@1895
   702
                p[2] = (Uint8) (ch >> 8);
slouken@1895
   703
                p[3] = (Uint8) ch;
slouken@1895
   704
                dst += 4;
slouken@1895
   705
                dstlen -= 4;
slouken@1895
   706
            }
slouken@1895
   707
            break;
slouken@1895
   708
        case ENCODING_UTF32LE:
slouken@1895
   709
            {
slouken@1895
   710
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   711
                if (ch > 0x10FFFF) {
slouken@1895
   712
                    ch = UNKNOWN_UNICODE;
slouken@1895
   713
                }
slouken@1895
   714
                if (dstlen < 4) {
slouken@1895
   715
                    return SDL_ICONV_E2BIG;
slouken@1895
   716
                }
slouken@1895
   717
                p[3] = (Uint8) (ch >> 24);
slouken@1895
   718
                p[2] = (Uint8) (ch >> 16);
slouken@1895
   719
                p[1] = (Uint8) (ch >> 8);
slouken@1895
   720
                p[0] = (Uint8) ch;
slouken@1895
   721
                dst += 4;
slouken@1895
   722
                dstlen -= 4;
slouken@1895
   723
            }
slouken@1895
   724
            break;
slouken@1895
   725
        case ENCODING_UCS2:
slouken@1895
   726
            {
slouken@1895
   727
                Uint16 *p = (Uint16 *) dst;
slouken@1895
   728
                if (ch > 0xFFFF) {
slouken@1895
   729
                    ch = UNKNOWN_UNICODE;
slouken@1895
   730
                }
slouken@1895
   731
                if (dstlen < 2) {
slouken@1895
   732
                    return SDL_ICONV_E2BIG;
slouken@1895
   733
                }
slouken@1895
   734
                *p = (Uint16) ch;
slouken@1895
   735
                dst += 2;
slouken@1895
   736
                dstlen -= 2;
slouken@1895
   737
            }
slouken@1895
   738
            break;
slouken@1895
   739
        case ENCODING_UCS4:
slouken@1895
   740
            {
slouken@1895
   741
                Uint32 *p = (Uint32 *) dst;
slouken@1895
   742
                if (ch > 0x7FFFFFFF) {
slouken@1895
   743
                    ch = UNKNOWN_UNICODE;
slouken@1895
   744
                }
slouken@1895
   745
                if (dstlen < 4) {
slouken@1895
   746
                    return SDL_ICONV_E2BIG;
slouken@1895
   747
                }
slouken@1895
   748
                *p = ch;
slouken@1895
   749
                dst += 4;
slouken@1895
   750
                dstlen -= 4;
slouken@1895
   751
            }
slouken@1895
   752
            break;
slouken@1895
   753
        }
slouken@1501
   754
slouken@1895
   755
        /* Update state */
slouken@1895
   756
        *inbuf = src;
slouken@1895
   757
        *inbytesleft = srclen;
slouken@1895
   758
        *outbuf = dst;
slouken@1895
   759
        *outbytesleft = dstlen;
slouken@1895
   760
        ++total;
slouken@1895
   761
    }
slouken@1895
   762
    return total;
slouken@1501
   763
}
slouken@1501
   764
slouken@1895
   765
int
slouken@1895
   766
SDL_iconv_close(SDL_iconv_t cd)
slouken@1501
   767
{
slouken@1895
   768
    if (cd && cd != (SDL_iconv_t) - 1) {
slouken@1895
   769
        SDL_free(cd);
slouken@1895
   770
    }
slouken@1895
   771
    return 0;
slouken@1501
   772
}
slouken@1501
   773
slouken@1501
   774
#endif /* !HAVE_ICONV */
slouken@1501
   775
slouken@2143
   776
static const char *
slouken@2143
   777
getlocale()
slouken@2143
   778
{
slouken@2143
   779
    const char *lang;
slouken@2143
   780
slouken@2143
   781
    lang = SDL_getenv("LC_ALL");
slouken@2143
   782
    if (!lang) {
slouken@2143
   783
        lang = SDL_getenv("LC_CTYPE");
slouken@2143
   784
    }
slouken@2143
   785
    if (!lang) {
slouken@2143
   786
        lang = SDL_getenv("LC_MESSAGES");
slouken@2143
   787
    }
slouken@2143
   788
    if (!lang) {
slouken@2143
   789
        lang = SDL_getenv("LANG");
slouken@2143
   790
    }
slouken@2143
   791
    if (!lang || SDL_strcmp(lang, "C") == 0) {
slouken@2143
   792
        lang = "ASCII";
slouken@2143
   793
    }
slouken@2143
   794
    return lang;
slouken@2143
   795
}
slouken@2143
   796
slouken@1895
   797
char *
slouken@2135
   798
SDL_iconv_string(const char *tocode, const char *fromcode, const char *inbuf,
slouken@1895
   799
                 size_t inbytesleft)
slouken@1501
   800
{
slouken@1895
   801
    SDL_iconv_t cd;
slouken@1895
   802
    char *string;
slouken@1895
   803
    size_t stringsize;
slouken@1895
   804
    char *outbuf;
slouken@1895
   805
    size_t outbytesleft;
slouken@1895
   806
    size_t retCode = 0;
slouken@1501
   807
slouken@2143
   808
    if (!fromcode || !*fromcode) {
slouken@2143
   809
        fromcode = getlocale();
slouken@2143
   810
    }
slouken@2143
   811
    if (!tocode || !*tocode) {
slouken@2143
   812
        tocode = getlocale();
slouken@2143
   813
    }
slouken@1895
   814
    cd = SDL_iconv_open(tocode, fromcode);
slouken@1895
   815
    if (cd == (SDL_iconv_t) - 1) {
slouken@1895
   816
        return NULL;
slouken@1895
   817
    }
slouken@1501
   818
slouken@1895
   819
    stringsize = inbytesleft > 4 ? inbytesleft : 4;
slouken@1895
   820
    string = SDL_malloc(stringsize);
slouken@1895
   821
    if (!string) {
slouken@1895
   822
        SDL_iconv_close(cd);
slouken@1895
   823
        return NULL;
slouken@1895
   824
    }
slouken@1895
   825
    outbuf = string;
slouken@1895
   826
    outbytesleft = stringsize;
slouken@1895
   827
    SDL_memset(outbuf, 0, 4);
slouken@1501
   828
slouken@1895
   829
    while (inbytesleft > 0) {
slouken@1895
   830
        retCode = SDL_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
slouken@1895
   831
        switch (retCode) {
slouken@1895
   832
        case SDL_ICONV_E2BIG:
slouken@1895
   833
            {
slouken@1895
   834
                char *oldstring = string;
slouken@1895
   835
                stringsize *= 2;
slouken@1895
   836
                string = SDL_realloc(string, stringsize);
slouken@1895
   837
                if (!string) {
slouken@1895
   838
                    SDL_iconv_close(cd);
slouken@1895
   839
                    return NULL;
slouken@1895
   840
                }
slouken@1895
   841
                outbuf = string + (outbuf - oldstring);
slouken@1895
   842
                outbytesleft = stringsize - (outbuf - string);
slouken@1895
   843
                SDL_memset(outbuf, 0, 4);
slouken@1895
   844
            }
slouken@1895
   845
            break;
slouken@1895
   846
        case SDL_ICONV_EILSEQ:
slouken@1895
   847
            /* Try skipping some input data - not perfect, but... */
slouken@1895
   848
            ++inbuf;
slouken@1895
   849
            --inbytesleft;
slouken@1895
   850
            break;
slouken@1895
   851
        case SDL_ICONV_EINVAL:
slouken@1895
   852
        case SDL_ICONV_ERROR:
slouken@1895
   853
            /* We can't continue... */
slouken@1895
   854
            inbytesleft = 0;
slouken@1895
   855
            break;
slouken@1895
   856
        }
slouken@1895
   857
    }
slouken@1895
   858
    SDL_iconv_close(cd);
slouken@1501
   859
slouken@1895
   860
    return string;
slouken@1501
   861
}
slouken@1895
   862
slouken@1895
   863
/* vi: set ts=4 sw=4 expandtab: */