src/stdlib/SDL_iconv.c
author Sam Lantinga <slouken@libsdl.org>
Thu, 28 Jun 2007 06:57:08 +0000
changeset 2135 0313af081a84
parent 1895 c121d94672cb
child 2136 5a1b6fe6e90f
permissions -rw-r--r--
Merge iconv const changes from 1.2
I'm not entirely happy with them. Maybe the right way to go is to leave
SDL_iconv() taking a non-const inbuf? How often are we converting const
strings anyway?
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@1895
    33
size_t
slouken@1895
    34
SDL_iconv(SDL_iconv_t cd,
slouken@2135
    35
          const char **inbuf, size_t * inbytesleft,
slouken@1895
    36
          char **outbuf, size_t * outbytesleft)
slouken@1501
    37
{
slouken@2135
    38
    size_t retCode;
slouken@2135
    39
#ifdef ICONV_REALLY_MODIFIES_INBUF
slouken@2135
    40
    if (inbuf && *inbuf && inbytesleft) {
slouken@2135
    41
        char *tmp = SDL_stack_alloc(char, *inbytesleft);
slouken@2135
    42
        char *ptr = tmp;
slouken@2135
    43
        SDL_memcpy(tmp, inbuf, *inbytesleft);
slouken@2135
    44
        retCode = iconv(cd, &ptr, inbytesleft, outbuf, outbytesleft);
slouken@2135
    45
        inbuf += (ptr - tmp);
slouken@2135
    46
        SDL_stack_free(tmp);
slouken@2135
    47
    } else {
slouken@2135
    48
        retCode = iconv(cd, NULL, inbytesleft, outbuf, outbytesleft);
slouken@2135
    49
    }
slouken@2135
    50
#else
slouken@2135
    51
    retCode = iconv(cd, (char **) 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@1895
   113
    {
slouken@1895
   114
    "ASCII", ENCODING_ASCII}, {
slouken@1895
   115
    "US-ASCII", ENCODING_ASCII}, {
slouken@1895
   116
    "LATIN1", ENCODING_LATIN1}, {
slouken@1895
   117
    "ISO-8859-1", ENCODING_LATIN1}, {
slouken@1895
   118
    "UTF8", ENCODING_UTF8}, {
slouken@1895
   119
    "UTF-8", ENCODING_UTF8}, {
slouken@1895
   120
    "UTF16", ENCODING_UTF16}, {
slouken@1895
   121
    "UTF-16", ENCODING_UTF16}, {
slouken@1895
   122
    "UTF16BE", ENCODING_UTF16BE}, {
slouken@1895
   123
    "UTF-16BE", ENCODING_UTF16BE}, {
slouken@1895
   124
    "UTF16LE", ENCODING_UTF16LE}, {
slouken@1895
   125
    "UTF-16LE", ENCODING_UTF16LE}, {
slouken@1895
   126
    "UTF32", ENCODING_UTF32}, {
slouken@1895
   127
    "UTF-32", ENCODING_UTF32}, {
slouken@1895
   128
    "UTF32BE", ENCODING_UTF32BE}, {
slouken@1895
   129
    "UTF-32BE", ENCODING_UTF32BE}, {
slouken@1895
   130
    "UTF32LE", ENCODING_UTF32LE}, {
slouken@1895
   131
    "UTF-32LE", ENCODING_UTF32LE}, {
slouken@1895
   132
    "UCS2", ENCODING_UCS2}, {
slouken@1895
   133
    "UCS-2", ENCODING_UCS2}, {
slouken@1895
   134
    "UCS4", ENCODING_UCS4}, {
slouken@1895
   135
"UCS-4", ENCODING_UCS4},};
slouken@1501
   136
slouken@1895
   137
SDL_iconv_t
slouken@1895
   138
SDL_iconv_open(const char *tocode, const char *fromcode)
slouken@1501
   139
{
slouken@1895
   140
    int src_fmt = ENCODING_UNKNOWN;
slouken@1895
   141
    int dst_fmt = ENCODING_UNKNOWN;
slouken@1895
   142
    int i;
slouken@1501
   143
slouken@1895
   144
    for (i = 0; i < SDL_arraysize(encodings); ++i) {
slouken@1895
   145
        if (SDL_strcasecmp(fromcode, encodings[i].name) == 0) {
slouken@1895
   146
            src_fmt = encodings[i].format;
slouken@1895
   147
            if (dst_fmt != ENCODING_UNKNOWN) {
slouken@1895
   148
                break;
slouken@1895
   149
            }
slouken@1895
   150
        }
slouken@1895
   151
        if (SDL_strcasecmp(tocode, encodings[i].name) == 0) {
slouken@1895
   152
            dst_fmt = encodings[i].format;
slouken@1895
   153
            if (src_fmt != ENCODING_UNKNOWN) {
slouken@1895
   154
                break;
slouken@1895
   155
            }
slouken@1895
   156
        }
slouken@1895
   157
    }
slouken@1895
   158
    if (src_fmt != ENCODING_UNKNOWN && dst_fmt != ENCODING_UNKNOWN) {
slouken@1895
   159
        SDL_iconv_t cd = (SDL_iconv_t) SDL_malloc(sizeof(*cd));
slouken@1895
   160
        if (cd) {
slouken@1895
   161
            cd->src_fmt = src_fmt;
slouken@1895
   162
            cd->dst_fmt = dst_fmt;
slouken@1895
   163
            return cd;
slouken@1895
   164
        }
slouken@1895
   165
    }
slouken@1895
   166
    return (SDL_iconv_t) - 1;
slouken@1501
   167
}
slouken@1501
   168
slouken@1895
   169
size_t
slouken@1895
   170
SDL_iconv(SDL_iconv_t cd,
slouken@2135
   171
          const char **inbuf, size_t * inbytesleft,
slouken@1895
   172
          char **outbuf, size_t * outbytesleft)
slouken@1501
   173
{
slouken@1895
   174
    /* For simplicity, we'll convert everything to and from UCS-4 */
slouken@2135
   175
    const char *src;
slouken@2135
   176
    char *dst;
slouken@1895
   177
    size_t srclen, dstlen;
slouken@2135
   178
    Uint32 ch = 0;
slouken@1895
   179
    size_t total;
slouken@1501
   180
slouken@1895
   181
    if (!inbuf || !*inbuf) {
slouken@1895
   182
        /* Reset the context */
slouken@1895
   183
        return 0;
slouken@1895
   184
    }
slouken@1895
   185
    if (!outbuf || !*outbuf || !outbytesleft || !*outbytesleft) {
slouken@1895
   186
        return SDL_ICONV_E2BIG;
slouken@1895
   187
    }
slouken@1895
   188
    src = *inbuf;
slouken@1895
   189
    srclen = (inbytesleft ? *inbytesleft : 0);
slouken@1895
   190
    dst = *outbuf;
slouken@1895
   191
    dstlen = *outbytesleft;
slouken@1501
   192
slouken@1895
   193
    switch (cd->src_fmt) {
slouken@1895
   194
    case ENCODING_UTF16:
slouken@1895
   195
        /* Scan for a byte order marker */
slouken@1895
   196
        {
slouken@1895
   197
            Uint8 *p = (Uint8 *) src;
slouken@1895
   198
            size_t n = srclen / 2;
slouken@1895
   199
            while (n) {
slouken@1895
   200
                if (p[0] == 0xFF && p[1] == 0xFE) {
slouken@1895
   201
                    cd->src_fmt = ENCODING_UTF16BE;
slouken@1895
   202
                    break;
slouken@1895
   203
                } else if (p[0] == 0xFE && p[1] == 0xFF) {
slouken@1895
   204
                    cd->src_fmt = ENCODING_UTF16LE;
slouken@1895
   205
                    break;
slouken@1895
   206
                }
slouken@1895
   207
                p += 2;
slouken@1895
   208
                --n;
slouken@1895
   209
            }
slouken@1895
   210
            if (n == 0) {
slouken@1895
   211
                /* We can't tell, default to host order */
slouken@1895
   212
                cd->src_fmt = ENCODING_UTF16NATIVE;
slouken@1895
   213
            }
slouken@1895
   214
        }
slouken@1895
   215
        break;
slouken@1895
   216
    case ENCODING_UTF32:
slouken@1895
   217
        /* Scan for a byte order marker */
slouken@1895
   218
        {
slouken@1895
   219
            Uint8 *p = (Uint8 *) src;
slouken@1895
   220
            size_t n = srclen / 4;
slouken@1895
   221
            while (n) {
slouken@1895
   222
                if (p[0] == 0xFF && p[1] == 0xFE &&
slouken@1895
   223
                    p[2] == 0x00 && p[3] == 0x00) {
slouken@1895
   224
                    cd->src_fmt = ENCODING_UTF32BE;
slouken@1895
   225
                    break;
slouken@1895
   226
                } else if (p[0] == 0x00 && p[1] == 0x00 &&
slouken@1895
   227
                           p[2] == 0xFE && p[3] == 0xFF) {
slouken@1895
   228
                    cd->src_fmt = ENCODING_UTF32LE;
slouken@1895
   229
                    break;
slouken@1895
   230
                }
slouken@1895
   231
                p += 4;
slouken@1895
   232
                --n;
slouken@1895
   233
            }
slouken@1895
   234
            if (n == 0) {
slouken@1895
   235
                /* We can't tell, default to host order */
slouken@1895
   236
                cd->src_fmt = ENCODING_UTF32NATIVE;
slouken@1895
   237
            }
slouken@1895
   238
        }
slouken@1895
   239
        break;
slouken@1895
   240
    }
slouken@1501
   241
slouken@1895
   242
    switch (cd->dst_fmt) {
slouken@1895
   243
    case ENCODING_UTF16:
slouken@1895
   244
        /* Default to host order, need to add byte order marker */
slouken@1895
   245
        if (dstlen < 2) {
slouken@1895
   246
            return SDL_ICONV_E2BIG;
slouken@1895
   247
        }
slouken@1895
   248
        *(Uint16 *) dst = UNICODE_BOM;
slouken@1895
   249
        dst += 2;
slouken@1895
   250
        dstlen -= 2;
slouken@1895
   251
        cd->dst_fmt = ENCODING_UTF16NATIVE;
slouken@1895
   252
        break;
slouken@1895
   253
    case ENCODING_UTF32:
slouken@1895
   254
        /* Default to host order, need to add byte order marker */
slouken@1895
   255
        if (dstlen < 4) {
slouken@1895
   256
            return SDL_ICONV_E2BIG;
slouken@1895
   257
        }
slouken@1895
   258
        *(Uint32 *) dst = UNICODE_BOM;
slouken@1895
   259
        dst += 4;
slouken@1895
   260
        dstlen -= 4;
slouken@1895
   261
        cd->dst_fmt = ENCODING_UTF32NATIVE;
slouken@1895
   262
        break;
slouken@1895
   263
    }
slouken@1501
   264
slouken@1895
   265
    total = 0;
slouken@1895
   266
    while (srclen > 0) {
slouken@1895
   267
        /* Decode a character */
slouken@1895
   268
        switch (cd->src_fmt) {
slouken@1895
   269
        case ENCODING_ASCII:
slouken@1895
   270
            {
slouken@1895
   271
                Uint8 *p = (Uint8 *) src;
slouken@1895
   272
                ch = (Uint32) (p[0] & 0x7F);
slouken@1895
   273
                ++src;
slouken@1895
   274
                --srclen;
slouken@1895
   275
            }
slouken@1895
   276
            break;
slouken@1895
   277
        case ENCODING_LATIN1:
slouken@1895
   278
            {
slouken@1895
   279
                Uint8 *p = (Uint8 *) src;
slouken@1895
   280
                ch = (Uint32) p[0];
slouken@1895
   281
                ++src;
slouken@1895
   282
                --srclen;
slouken@1895
   283
            }
slouken@1895
   284
            break;
slouken@1895
   285
        case ENCODING_UTF8:    /* RFC 3629 */
slouken@1895
   286
            {
slouken@1895
   287
                Uint8 *p = (Uint8 *) src;
slouken@1895
   288
                size_t left = 0;
slouken@1895
   289
                SDL_bool overlong = SDL_FALSE;
slouken@1895
   290
                if (p[0] >= 0xFC) {
slouken@1895
   291
                    if ((p[0] & 0xFE) != 0xFC) {
slouken@1895
   292
                        /* Skip illegal sequences
slouken@1895
   293
                           return SDL_ICONV_EILSEQ;
slouken@1895
   294
                         */
slouken@1895
   295
                        ch = UNKNOWN_UNICODE;
slouken@1895
   296
                    } else {
slouken@1895
   297
                        if (p[0] == 0xFC) {
slouken@1895
   298
                            overlong = SDL_TRUE;
slouken@1895
   299
                        }
slouken@1895
   300
                        ch = (Uint32) (p[0] & 0x01);
slouken@1895
   301
                        left = 5;
slouken@1895
   302
                    }
slouken@1895
   303
                } else if (p[0] >= 0xF8) {
slouken@1895
   304
                    if ((p[0] & 0xFC) != 0xF8) {
slouken@1895
   305
                        /* Skip illegal sequences
slouken@1895
   306
                           return SDL_ICONV_EILSEQ;
slouken@1895
   307
                         */
slouken@1895
   308
                        ch = UNKNOWN_UNICODE;
slouken@1895
   309
                    } else {
slouken@1895
   310
                        if (p[0] == 0xF8) {
slouken@1895
   311
                            overlong = SDL_TRUE;
slouken@1895
   312
                        }
slouken@1895
   313
                        ch = (Uint32) (p[0] & 0x03);
slouken@1895
   314
                        left = 4;
slouken@1895
   315
                    }
slouken@1895
   316
                } else if (p[0] >= 0xF0) {
slouken@1895
   317
                    if ((p[0] & 0xF8) != 0xF0) {
slouken@1895
   318
                        /* Skip illegal sequences
slouken@1895
   319
                           return SDL_ICONV_EILSEQ;
slouken@1895
   320
                         */
slouken@1895
   321
                        ch = UNKNOWN_UNICODE;
slouken@1895
   322
                    } else {
slouken@1895
   323
                        if (p[0] == 0xF0) {
slouken@1895
   324
                            overlong = SDL_TRUE;
slouken@1895
   325
                        }
slouken@1895
   326
                        ch = (Uint32) (p[0] & 0x07);
slouken@1895
   327
                        left = 3;
slouken@1895
   328
                    }
slouken@1895
   329
                } else if (p[0] >= 0xE0) {
slouken@1895
   330
                    if ((p[0] & 0xF0) != 0xE0) {
slouken@1895
   331
                        /* Skip illegal sequences
slouken@1895
   332
                           return SDL_ICONV_EILSEQ;
slouken@1895
   333
                         */
slouken@1895
   334
                        ch = UNKNOWN_UNICODE;
slouken@1895
   335
                    } else {
slouken@1895
   336
                        if (p[0] == 0xE0) {
slouken@1895
   337
                            overlong = SDL_TRUE;
slouken@1895
   338
                        }
slouken@1895
   339
                        ch = (Uint32) (p[0] & 0x0F);
slouken@1895
   340
                        left = 2;
slouken@1895
   341
                    }
slouken@1895
   342
                } else if (p[0] >= 0xC0) {
slouken@1895
   343
                    if ((p[0] & 0xE0) != 0xC0) {
slouken@1895
   344
                        /* Skip illegal sequences
slouken@1895
   345
                           return SDL_ICONV_EILSEQ;
slouken@1895
   346
                         */
slouken@1895
   347
                        ch = UNKNOWN_UNICODE;
slouken@1895
   348
                    } else {
slouken@1895
   349
                        if ((p[0] & 0xCE) == 0xC0) {
slouken@1895
   350
                            overlong = SDL_TRUE;
slouken@1895
   351
                        }
slouken@1895
   352
                        ch = (Uint32) (p[0] & 0x1F);
slouken@1895
   353
                        left = 1;
slouken@1895
   354
                    }
slouken@1895
   355
                } else {
slouken@1895
   356
                    if ((p[0] & 0x80) != 0x00) {
slouken@1895
   357
                        /* Skip illegal sequences
slouken@1895
   358
                           return SDL_ICONV_EILSEQ;
slouken@1895
   359
                         */
slouken@1895
   360
                        ch = UNKNOWN_UNICODE;
slouken@1895
   361
                    } else {
slouken@1895
   362
                        ch = (Uint32) p[0];
slouken@1895
   363
                    }
slouken@1895
   364
                }
slouken@1895
   365
                ++src;
slouken@1895
   366
                --srclen;
slouken@1895
   367
                if (srclen < left) {
slouken@1895
   368
                    return SDL_ICONV_EINVAL;
slouken@1895
   369
                }
slouken@1895
   370
                while (left--) {
slouken@1895
   371
                    ++p;
slouken@1895
   372
                    if ((p[0] & 0xC0) != 0x80) {
slouken@1895
   373
                        /* Skip illegal sequences
slouken@1895
   374
                           return SDL_ICONV_EILSEQ;
slouken@1895
   375
                         */
slouken@1895
   376
                        ch = UNKNOWN_UNICODE;
slouken@1895
   377
                        break;
slouken@1895
   378
                    }
slouken@1895
   379
                    ch <<= 6;
slouken@1895
   380
                    ch |= (p[0] & 0x3F);
slouken@1895
   381
                    ++src;
slouken@1895
   382
                    --srclen;
slouken@1895
   383
                }
slouken@1895
   384
                if (overlong) {
slouken@1895
   385
                    /* Potential security risk
slouken@1895
   386
                       return SDL_ICONV_EILSEQ;
slouken@1895
   387
                     */
slouken@1895
   388
                    ch = UNKNOWN_UNICODE;
slouken@1895
   389
                }
slouken@1895
   390
                if ((ch >= 0xD800 && ch <= 0xDFFF) ||
slouken@1895
   391
                    (ch == 0xFFFE || ch == 0xFFFF) || ch > 0x10FFFF) {
slouken@1895
   392
                    /* Skip illegal sequences
slouken@1895
   393
                       return SDL_ICONV_EILSEQ;
slouken@1895
   394
                     */
slouken@1895
   395
                    ch = UNKNOWN_UNICODE;
slouken@1895
   396
                }
slouken@1895
   397
            }
slouken@1895
   398
            break;
slouken@1895
   399
        case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1895
   400
            {
slouken@1895
   401
                Uint8 *p = (Uint8 *) src;
slouken@1895
   402
                Uint16 W1, W2;
slouken@1895
   403
                if (srclen < 2) {
slouken@1895
   404
                    return SDL_ICONV_EINVAL;
slouken@1895
   405
                }
slouken@1895
   406
                W1 = ((Uint16) p[0] << 8) | (Uint16) p[1];
slouken@1895
   407
                src += 2;
slouken@1895
   408
                srclen -= 2;
slouken@1895
   409
                if (W1 < 0xD800 || W1 > 0xDFFF) {
slouken@1895
   410
                    ch = (Uint32) W1;
slouken@1895
   411
                    break;
slouken@1895
   412
                }
slouken@1895
   413
                if (W1 > 0xDBFF) {
slouken@1895
   414
                    /* Skip illegal sequences
slouken@1895
   415
                       return SDL_ICONV_EILSEQ;
slouken@1895
   416
                     */
slouken@1895
   417
                    ch = UNKNOWN_UNICODE;
slouken@1895
   418
                    break;
slouken@1895
   419
                }
slouken@1895
   420
                if (srclen < 2) {
slouken@1895
   421
                    return SDL_ICONV_EINVAL;
slouken@1895
   422
                }
slouken@1895
   423
                p = (Uint8 *) src;
slouken@1895
   424
                W2 = ((Uint16) p[0] << 8) | (Uint16) p[1];
slouken@1895
   425
                src += 2;
slouken@1895
   426
                srclen -= 2;
slouken@1895
   427
                if (W2 < 0xDC00 || W2 > 0xDFFF) {
slouken@1895
   428
                    /* Skip illegal sequences
slouken@1895
   429
                       return SDL_ICONV_EILSEQ;
slouken@1895
   430
                     */
slouken@1895
   431
                    ch = UNKNOWN_UNICODE;
slouken@1895
   432
                    break;
slouken@1895
   433
                }
slouken@1895
   434
                ch = (((Uint32) (W1 & 0x3FF) << 10) |
slouken@1895
   435
                      (Uint32) (W2 & 0x3FF)) + 0x10000;
slouken@1895
   436
            }
slouken@1895
   437
            break;
slouken@1895
   438
        case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1895
   439
            {
slouken@1895
   440
                Uint8 *p = (Uint8 *) src;
slouken@1895
   441
                Uint16 W1, W2;
slouken@1895
   442
                if (srclen < 2) {
slouken@1895
   443
                    return SDL_ICONV_EINVAL;
slouken@1895
   444
                }
slouken@1895
   445
                W1 = ((Uint16) p[1] << 8) | (Uint16) p[0];
slouken@1895
   446
                src += 2;
slouken@1895
   447
                srclen -= 2;
slouken@1895
   448
                if (W1 < 0xD800 || W1 > 0xDFFF) {
slouken@1895
   449
                    ch = (Uint32) W1;
slouken@1895
   450
                    break;
slouken@1895
   451
                }
slouken@1895
   452
                if (W1 > 0xDBFF) {
slouken@1895
   453
                    /* Skip illegal sequences
slouken@1895
   454
                       return SDL_ICONV_EILSEQ;
slouken@1895
   455
                     */
slouken@1895
   456
                    ch = UNKNOWN_UNICODE;
slouken@1895
   457
                    break;
slouken@1895
   458
                }
slouken@1895
   459
                if (srclen < 2) {
slouken@1895
   460
                    return SDL_ICONV_EINVAL;
slouken@1895
   461
                }
slouken@1895
   462
                p = (Uint8 *) src;
slouken@1895
   463
                W2 = ((Uint16) p[1] << 8) | (Uint16) p[0];
slouken@1895
   464
                src += 2;
slouken@1895
   465
                srclen -= 2;
slouken@1895
   466
                if (W2 < 0xDC00 || W2 > 0xDFFF) {
slouken@1895
   467
                    /* Skip illegal sequences
slouken@1895
   468
                       return SDL_ICONV_EILSEQ;
slouken@1895
   469
                     */
slouken@1895
   470
                    ch = UNKNOWN_UNICODE;
slouken@1895
   471
                    break;
slouken@1895
   472
                }
slouken@1895
   473
                ch = (((Uint32) (W1 & 0x3FF) << 10) |
slouken@1895
   474
                      (Uint32) (W2 & 0x3FF)) + 0x10000;
slouken@1895
   475
            }
slouken@1895
   476
            break;
slouken@1895
   477
        case ENCODING_UTF32BE:
slouken@1895
   478
            {
slouken@1895
   479
                Uint8 *p = (Uint8 *) src;
slouken@1895
   480
                if (srclen < 4) {
slouken@1895
   481
                    return SDL_ICONV_EINVAL;
slouken@1895
   482
                }
slouken@1895
   483
                ch = ((Uint32) p[0] << 24) |
slouken@1895
   484
                    ((Uint32) p[1] << 16) |
slouken@1895
   485
                    ((Uint32) p[2] << 8) | (Uint32) p[3];
slouken@1895
   486
                src += 4;
slouken@1895
   487
                srclen -= 4;
slouken@1895
   488
            }
slouken@1895
   489
            break;
slouken@1895
   490
        case ENCODING_UTF32LE:
slouken@1895
   491
            {
slouken@1895
   492
                Uint8 *p = (Uint8 *) src;
slouken@1895
   493
                if (srclen < 4) {
slouken@1895
   494
                    return SDL_ICONV_EINVAL;
slouken@1895
   495
                }
slouken@1895
   496
                ch = ((Uint32) p[3] << 24) |
slouken@1895
   497
                    ((Uint32) p[2] << 16) |
slouken@1895
   498
                    ((Uint32) p[1] << 8) | (Uint32) p[0];
slouken@1895
   499
                src += 4;
slouken@1895
   500
                srclen -= 4;
slouken@1895
   501
            }
slouken@1895
   502
            break;
slouken@1895
   503
        case ENCODING_UCS2:
slouken@1895
   504
            {
slouken@1895
   505
                Uint16 *p = (Uint16 *) src;
slouken@1895
   506
                if (srclen < 2) {
slouken@1895
   507
                    return SDL_ICONV_EINVAL;
slouken@1895
   508
                }
slouken@1895
   509
                ch = *p;
slouken@1895
   510
                src += 2;
slouken@1895
   511
                srclen -= 2;
slouken@1895
   512
            }
slouken@1895
   513
            break;
slouken@1895
   514
        case ENCODING_UCS4:
slouken@1895
   515
            {
slouken@1895
   516
                Uint32 *p = (Uint32 *) src;
slouken@1895
   517
                if (srclen < 4) {
slouken@1895
   518
                    return SDL_ICONV_EINVAL;
slouken@1895
   519
                }
slouken@1895
   520
                ch = *p;
slouken@1895
   521
                src += 4;
slouken@1895
   522
                srclen -= 4;
slouken@1895
   523
            }
slouken@1895
   524
            break;
slouken@1895
   525
        }
slouken@1501
   526
slouken@1895
   527
        /* Encode a character */
slouken@1895
   528
        switch (cd->dst_fmt) {
slouken@1895
   529
        case ENCODING_ASCII:
slouken@1895
   530
            {
slouken@1895
   531
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   532
                if (dstlen < 1) {
slouken@1895
   533
                    return SDL_ICONV_E2BIG;
slouken@1895
   534
                }
slouken@1895
   535
                if (ch > 0x7F) {
slouken@1895
   536
                    *p = UNKNOWN_ASCII;
slouken@1895
   537
                } else {
slouken@1895
   538
                    *p = (Uint8) ch;
slouken@1895
   539
                }
slouken@1895
   540
                ++dst;
slouken@1895
   541
                --dstlen;
slouken@1895
   542
            }
slouken@1895
   543
            break;
slouken@1895
   544
        case ENCODING_LATIN1:
slouken@1895
   545
            {
slouken@1895
   546
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   547
                if (dstlen < 1) {
slouken@1895
   548
                    return SDL_ICONV_E2BIG;
slouken@1895
   549
                }
slouken@1895
   550
                if (ch > 0xFF) {
slouken@1895
   551
                    *p = UNKNOWN_ASCII;
slouken@1895
   552
                } else {
slouken@1895
   553
                    *p = (Uint8) ch;
slouken@1895
   554
                }
slouken@1895
   555
                ++dst;
slouken@1895
   556
                --dstlen;
slouken@1895
   557
            }
slouken@1895
   558
            break;
slouken@1895
   559
        case ENCODING_UTF8:    /* RFC 3629 */
slouken@1895
   560
            {
slouken@1895
   561
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   562
                if (ch > 0x10FFFF) {
slouken@1895
   563
                    ch = UNKNOWN_UNICODE;
slouken@1895
   564
                }
slouken@1895
   565
                if (ch <= 0x7F) {
slouken@1895
   566
                    if (dstlen < 1) {
slouken@1895
   567
                        return SDL_ICONV_E2BIG;
slouken@1895
   568
                    }
slouken@1895
   569
                    *p = (Uint8) ch;
slouken@1895
   570
                    ++dst;
slouken@1895
   571
                    --dstlen;
slouken@1895
   572
                } else if (ch <= 0x7FF) {
slouken@1895
   573
                    if (dstlen < 2) {
slouken@1895
   574
                        return SDL_ICONV_E2BIG;
slouken@1895
   575
                    }
slouken@1895
   576
                    p[0] = 0xC0 | (Uint8) ((ch >> 6) & 0x1F);
slouken@1895
   577
                    p[1] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   578
                    dst += 2;
slouken@1895
   579
                    dstlen -= 2;
slouken@1895
   580
                } else if (ch <= 0xFFFF) {
slouken@1895
   581
                    if (dstlen < 3) {
slouken@1895
   582
                        return SDL_ICONV_E2BIG;
slouken@1895
   583
                    }
slouken@1895
   584
                    p[0] = 0xE0 | (Uint8) ((ch >> 12) & 0x0F);
slouken@1895
   585
                    p[1] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   586
                    p[2] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   587
                    dst += 3;
slouken@1895
   588
                    dstlen -= 3;
slouken@1895
   589
                } else if (ch <= 0x1FFFFF) {
slouken@1895
   590
                    if (dstlen < 4) {
slouken@1895
   591
                        return SDL_ICONV_E2BIG;
slouken@1895
   592
                    }
slouken@1895
   593
                    p[0] = 0xF0 | (Uint8) ((ch >> 18) & 0x07);
slouken@1895
   594
                    p[1] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   595
                    p[2] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   596
                    p[3] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   597
                    dst += 4;
slouken@1895
   598
                    dstlen -= 4;
slouken@1895
   599
                } else if (ch <= 0x3FFFFFF) {
slouken@1895
   600
                    if (dstlen < 5) {
slouken@1895
   601
                        return SDL_ICONV_E2BIG;
slouken@1895
   602
                    }
slouken@1895
   603
                    p[0] = 0xF8 | (Uint8) ((ch >> 24) & 0x03);
slouken@1895
   604
                    p[1] = 0x80 | (Uint8) ((ch >> 18) & 0x3F);
slouken@1895
   605
                    p[2] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   606
                    p[3] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   607
                    p[4] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   608
                    dst += 5;
slouken@1895
   609
                    dstlen -= 5;
slouken@1895
   610
                } else {
slouken@1895
   611
                    if (dstlen < 6) {
slouken@1895
   612
                        return SDL_ICONV_E2BIG;
slouken@1895
   613
                    }
slouken@1895
   614
                    p[0] = 0xFC | (Uint8) ((ch >> 30) & 0x01);
slouken@1895
   615
                    p[1] = 0x80 | (Uint8) ((ch >> 24) & 0x3F);
slouken@1895
   616
                    p[2] = 0x80 | (Uint8) ((ch >> 18) & 0x3F);
slouken@1895
   617
                    p[3] = 0x80 | (Uint8) ((ch >> 12) & 0x3F);
slouken@1895
   618
                    p[4] = 0x80 | (Uint8) ((ch >> 6) & 0x3F);
slouken@1895
   619
                    p[5] = 0x80 | (Uint8) (ch & 0x3F);
slouken@1895
   620
                    dst += 6;
slouken@1895
   621
                    dstlen -= 6;
slouken@1895
   622
                }
slouken@1895
   623
            }
slouken@1895
   624
            break;
slouken@1895
   625
        case ENCODING_UTF16BE: /* RFC 2781 */
slouken@1895
   626
            {
slouken@1895
   627
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   628
                if (ch > 0x10FFFF) {
slouken@1895
   629
                    ch = UNKNOWN_UNICODE;
slouken@1895
   630
                }
slouken@1895
   631
                if (ch < 0x10000) {
slouken@1895
   632
                    if (dstlen < 2) {
slouken@1895
   633
                        return SDL_ICONV_E2BIG;
slouken@1895
   634
                    }
slouken@1895
   635
                    p[0] = (Uint8) (ch >> 8);
slouken@1895
   636
                    p[1] = (Uint8) ch;
slouken@1895
   637
                    dst += 2;
slouken@1895
   638
                    dstlen -= 2;
slouken@1895
   639
                } else {
slouken@1895
   640
                    Uint16 W1, W2;
slouken@1895
   641
                    if (dstlen < 4) {
slouken@1895
   642
                        return SDL_ICONV_E2BIG;
slouken@1895
   643
                    }
slouken@1895
   644
                    ch = ch - 0x10000;
slouken@1895
   645
                    W1 = 0xD800 | (Uint16) ((ch >> 10) & 0x3FF);
slouken@1895
   646
                    W2 = 0xDC00 | (Uint16) (ch & 0x3FF);
slouken@1895
   647
                    p[0] = (Uint8) (W1 >> 8);
slouken@1895
   648
                    p[1] = (Uint8) W1;
slouken@1895
   649
                    p[2] = (Uint8) (W2 >> 8);
slouken@1895
   650
                    p[3] = (Uint8) W2;
slouken@1895
   651
                    dst += 4;
slouken@1895
   652
                    dstlen -= 4;
slouken@1895
   653
                }
slouken@1895
   654
            }
slouken@1895
   655
            break;
slouken@1895
   656
        case ENCODING_UTF16LE: /* RFC 2781 */
slouken@1895
   657
            {
slouken@1895
   658
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   659
                if (ch > 0x10FFFF) {
slouken@1895
   660
                    ch = UNKNOWN_UNICODE;
slouken@1895
   661
                }
slouken@1895
   662
                if (ch < 0x10000) {
slouken@1895
   663
                    if (dstlen < 2) {
slouken@1895
   664
                        return SDL_ICONV_E2BIG;
slouken@1895
   665
                    }
slouken@1895
   666
                    p[1] = (Uint8) (ch >> 8);
slouken@1895
   667
                    p[0] = (Uint8) ch;
slouken@1895
   668
                    dst += 2;
slouken@1895
   669
                    dstlen -= 2;
slouken@1895
   670
                } else {
slouken@1895
   671
                    Uint16 W1, W2;
slouken@1895
   672
                    if (dstlen < 4) {
slouken@1895
   673
                        return SDL_ICONV_E2BIG;
slouken@1895
   674
                    }
slouken@1895
   675
                    ch = ch - 0x10000;
slouken@1895
   676
                    W1 = 0xD800 | (Uint16) ((ch >> 10) & 0x3FF);
slouken@1895
   677
                    W2 = 0xDC00 | (Uint16) (ch & 0x3FF);
slouken@1895
   678
                    p[1] = (Uint8) (W1 >> 8);
slouken@1895
   679
                    p[0] = (Uint8) W1;
slouken@1895
   680
                    p[3] = (Uint8) (W2 >> 8);
slouken@1895
   681
                    p[2] = (Uint8) W2;
slouken@1895
   682
                    dst += 4;
slouken@1895
   683
                    dstlen -= 4;
slouken@1895
   684
                }
slouken@1895
   685
            }
slouken@1895
   686
            break;
slouken@1895
   687
        case ENCODING_UTF32BE:
slouken@1895
   688
            {
slouken@1895
   689
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   690
                if (ch > 0x10FFFF) {
slouken@1895
   691
                    ch = UNKNOWN_UNICODE;
slouken@1895
   692
                }
slouken@1895
   693
                if (dstlen < 4) {
slouken@1895
   694
                    return SDL_ICONV_E2BIG;
slouken@1895
   695
                }
slouken@1895
   696
                p[0] = (Uint8) (ch >> 24);
slouken@1895
   697
                p[1] = (Uint8) (ch >> 16);
slouken@1895
   698
                p[2] = (Uint8) (ch >> 8);
slouken@1895
   699
                p[3] = (Uint8) ch;
slouken@1895
   700
                dst += 4;
slouken@1895
   701
                dstlen -= 4;
slouken@1895
   702
            }
slouken@1895
   703
            break;
slouken@1895
   704
        case ENCODING_UTF32LE:
slouken@1895
   705
            {
slouken@1895
   706
                Uint8 *p = (Uint8 *) dst;
slouken@1895
   707
                if (ch > 0x10FFFF) {
slouken@1895
   708
                    ch = UNKNOWN_UNICODE;
slouken@1895
   709
                }
slouken@1895
   710
                if (dstlen < 4) {
slouken@1895
   711
                    return SDL_ICONV_E2BIG;
slouken@1895
   712
                }
slouken@1895
   713
                p[3] = (Uint8) (ch >> 24);
slouken@1895
   714
                p[2] = (Uint8) (ch >> 16);
slouken@1895
   715
                p[1] = (Uint8) (ch >> 8);
slouken@1895
   716
                p[0] = (Uint8) ch;
slouken@1895
   717
                dst += 4;
slouken@1895
   718
                dstlen -= 4;
slouken@1895
   719
            }
slouken@1895
   720
            break;
slouken@1895
   721
        case ENCODING_UCS2:
slouken@1895
   722
            {
slouken@1895
   723
                Uint16 *p = (Uint16 *) dst;
slouken@1895
   724
                if (ch > 0xFFFF) {
slouken@1895
   725
                    ch = UNKNOWN_UNICODE;
slouken@1895
   726
                }
slouken@1895
   727
                if (dstlen < 2) {
slouken@1895
   728
                    return SDL_ICONV_E2BIG;
slouken@1895
   729
                }
slouken@1895
   730
                *p = (Uint16) ch;
slouken@1895
   731
                dst += 2;
slouken@1895
   732
                dstlen -= 2;
slouken@1895
   733
            }
slouken@1895
   734
            break;
slouken@1895
   735
        case ENCODING_UCS4:
slouken@1895
   736
            {
slouken@1895
   737
                Uint32 *p = (Uint32 *) dst;
slouken@1895
   738
                if (ch > 0x7FFFFFFF) {
slouken@1895
   739
                    ch = UNKNOWN_UNICODE;
slouken@1895
   740
                }
slouken@1895
   741
                if (dstlen < 4) {
slouken@1895
   742
                    return SDL_ICONV_E2BIG;
slouken@1895
   743
                }
slouken@1895
   744
                *p = ch;
slouken@1895
   745
                dst += 4;
slouken@1895
   746
                dstlen -= 4;
slouken@1895
   747
            }
slouken@1895
   748
            break;
slouken@1895
   749
        }
slouken@1501
   750
slouken@1895
   751
        /* Update state */
slouken@1895
   752
        *inbuf = src;
slouken@1895
   753
        *inbytesleft = srclen;
slouken@1895
   754
        *outbuf = dst;
slouken@1895
   755
        *outbytesleft = dstlen;
slouken@1895
   756
        ++total;
slouken@1895
   757
    }
slouken@1895
   758
    return total;
slouken@1501
   759
}
slouken@1501
   760
slouken@1895
   761
int
slouken@1895
   762
SDL_iconv_close(SDL_iconv_t cd)
slouken@1501
   763
{
slouken@1895
   764
    if (cd && cd != (SDL_iconv_t) - 1) {
slouken@1895
   765
        SDL_free(cd);
slouken@1895
   766
    }
slouken@1895
   767
    return 0;
slouken@1501
   768
}
slouken@1501
   769
slouken@1501
   770
#endif /* !HAVE_ICONV */
slouken@1501
   771
slouken@1895
   772
char *
slouken@2135
   773
SDL_iconv_string(const char *tocode, const char *fromcode, const char *inbuf,
slouken@1895
   774
                 size_t inbytesleft)
slouken@1501
   775
{
slouken@1895
   776
    SDL_iconv_t cd;
slouken@1895
   777
    char *string;
slouken@1895
   778
    size_t stringsize;
slouken@1895
   779
    char *outbuf;
slouken@1895
   780
    size_t outbytesleft;
slouken@1895
   781
    size_t retCode = 0;
slouken@1501
   782
slouken@1895
   783
    cd = SDL_iconv_open(tocode, fromcode);
slouken@1895
   784
    if (cd == (SDL_iconv_t) - 1) {
slouken@1895
   785
        return NULL;
slouken@1895
   786
    }
slouken@1501
   787
slouken@1895
   788
    stringsize = inbytesleft > 4 ? inbytesleft : 4;
slouken@1895
   789
    string = SDL_malloc(stringsize);
slouken@1895
   790
    if (!string) {
slouken@1895
   791
        SDL_iconv_close(cd);
slouken@1895
   792
        return NULL;
slouken@1895
   793
    }
slouken@1895
   794
    outbuf = string;
slouken@1895
   795
    outbytesleft = stringsize;
slouken@1895
   796
    SDL_memset(outbuf, 0, 4);
slouken@1501
   797
slouken@1895
   798
    while (inbytesleft > 0) {
slouken@1895
   799
        retCode = SDL_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
slouken@1895
   800
        switch (retCode) {
slouken@1895
   801
        case SDL_ICONV_E2BIG:
slouken@1895
   802
            {
slouken@1895
   803
                char *oldstring = string;
slouken@1895
   804
                stringsize *= 2;
slouken@1895
   805
                string = SDL_realloc(string, stringsize);
slouken@1895
   806
                if (!string) {
slouken@1895
   807
                    SDL_iconv_close(cd);
slouken@1895
   808
                    return NULL;
slouken@1895
   809
                }
slouken@1895
   810
                outbuf = string + (outbuf - oldstring);
slouken@1895
   811
                outbytesleft = stringsize - (outbuf - string);
slouken@1895
   812
                SDL_memset(outbuf, 0, 4);
slouken@1895
   813
            }
slouken@1895
   814
            break;
slouken@1895
   815
        case SDL_ICONV_EILSEQ:
slouken@1895
   816
            /* Try skipping some input data - not perfect, but... */
slouken@1895
   817
            ++inbuf;
slouken@1895
   818
            --inbytesleft;
slouken@1895
   819
            break;
slouken@1895
   820
        case SDL_ICONV_EINVAL:
slouken@1895
   821
        case SDL_ICONV_ERROR:
slouken@1895
   822
            /* We can't continue... */
slouken@1895
   823
            inbytesleft = 0;
slouken@1895
   824
            break;
slouken@1895
   825
        }
slouken@1895
   826
    }
slouken@1895
   827
    SDL_iconv_close(cd);
slouken@1501
   828
slouken@1895
   829
    return string;
slouken@1501
   830
}
slouken@1895
   831
slouken@1895
   832
/* vi: set ts=4 sw=4 expandtab: */