sdk/lib/3rdparty/libmpg123/icy2utf8.c

   1 /* mpg123 note: This is BSD-licensed code that is no problem for mpg123 usage under LGPL.
   2    It's Free, understood? ;-) */
   3
   4 /* Another note: This code is basically written by Thorsten Glaser,
   5    Thomas Orgis did just some rearrangements and comments. */
   6
   7 /*-
   8  * Copyright (c) 2008
   9  *      Thorsten Glaser <tg@mirbsd.org>
  10  *
  11  * Provided that these terms and disclaimer and all copyright notices
  12  * are retained or reproduced in an accompanying document, permission
  13  * is granted to deal in this work without restriction, including un-
  14  * limited rights to use, publicly perform, distribute, sell, modify,
  15  * merge, give away, or sublicence.
  16  *
  17  * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to
  18  * the utmost extent permitted by applicable law, neither express nor
  19  * implied; without malicious intent or gross negligence. In no event
  20  * may a licensor, author or contributor be held liable for indirect,
  21  * direct, other damage, loss, or other issues arising in any way out
  22  * of dealing in the work, even if advised of the possibility of such
  23  * damage or existence of a defect, except proven that it results out
  24  * of said person's immediate fault when using the work as intended.
  25  *-
  26  * Convert from ICY encoding (windows-1252 codepage) to UTF-8
  27  */
  28
  29 #include "config.h"
  30 #include "intsym.h"
  31 /* Includes string and stdlib headers... */
  32 #include "compat.h"
  33
  34 /* ThOr: too lazy for this type check; also we use char/short all around anyway.
  35    Of cource, it would be the proper way to use _these_ kind of types all around. */
  36 #define uint8_t  unsigned char
  37 #define uint16_t unsigned short
  38
  39 static const uint8_t cp1252_utf8[] = {
  40         /* 0x00 @   0 */        0x00,
  41         /* 0x01 @   1 */        0x01,
  42         /* 0x02 @   2 */        0x02,
  43         /* 0x03 @   3 */        0x03,
  44         /* 0x04 @   4 */        0x04,
  45         /* 0x05 @   5 */        0x05,
  46         /* 0x06 @   6 */        0x06,
  47         /* 0x07 @   7 */        0x07,
  48         /* 0x08 @   8 */        0x08,
  49         /* 0x09 @   9 */        0x09,
  50         /* 0x0A @  10 */        0x0A,
  51         /* 0x0B @  11 */        0x0B,
  52         /* 0x0C @  12 */        0x0C,
  53         /* 0x0D @  13 */        0x0D,
  54         /* 0x0E @  14 */        0x0E,
  55         /* 0x0F @  15 */        0x0F,
  56         /* 0x10 @  16 */        0x10,
  57         /* 0x11 @  17 */        0x11,
  58         /* 0x12 @  18 */        0x12,
  59         /* 0x13 @  19 */        0x13,
  60         /* 0x14 @  20 */        0x14,
  61         /* 0x15 @  21 */        0x15,
  62         /* 0x16 @  22 */        0x16,
  63         /* 0x17 @  23 */        0x17,
  64         /* 0x18 @  24 */        0x18,
  65         /* 0x19 @  25 */        0x19,
  66         /* 0x1A @  26 */        0x1A,
  67         /* 0x1B @  27 */        0x1B,
  68         /* 0x1C @  28 */        0x1C,
  69         /* 0x1D @  29 */        0x1D,
  70         /* 0x1E @  30 */        0x1E,
  71         /* 0x1F @  31 */        0x1F,
  72         /* 0x20 @  32 */        0x20,
  73         /* 0x21 @  33 */        0x21,
  74         /* 0x22 @  34 */        0x22,
  75         /* 0x23 @  35 */        0x23,
  76         /* 0x24 @  36 */        0x24,
  77         /* 0x25 @  37 */        0x25,
  78         /* 0x26 @  38 */        0x26,
  79         /* 0x27 @  39 */        0x27,
  80         /* 0x28 @  40 */        0x28,
  81         /* 0x29 @  41 */        0x29,
  82         /* 0x2A @  42 */        0x2A,
  83         /* 0x2B @  43 */        0x2B,
  84         /* 0x2C @  44 */        0x2C,
  85         /* 0x2D @  45 */        0x2D,
  86         /* 0x2E @  46 */        0x2E,
  87         /* 0x2F @  47 */        0x2F,
  88         /* 0x30 @  48 */        0x30,
  89         /* 0x31 @  49 */        0x31,
  90         /* 0x32 @  50 */        0x32,
  91         /* 0x33 @  51 */        0x33,
  92         /* 0x34 @  52 */        0x34,
  93         /* 0x35 @  53 */        0x35,
  94         /* 0x36 @  54 */        0x36,
  95         /* 0x37 @  55 */        0x37,
  96         /* 0x38 @  56 */        0x38,
  97         /* 0x39 @  57 */        0x39,
  98         /* 0x3A @  58 */        0x3A,
  99         /* 0x3B @  59 */        0x3B,
 100         /* 0x3C @  60 */        0x3C,
 101         /* 0x3D @  61 */        0x3D,
 102         /* 0x3E @  62 */        0x3E,
 103         /* 0x3F @  63 */        0x3F,
 104         /* 0x40 @  64 */        0x40,
 105         /* 0x41 @  65 */        0x41,
 106         /* 0x42 @  66 */        0x42,
 107         /* 0x43 @  67 */        0x43,
 108         /* 0x44 @  68 */        0x44,
 109         /* 0x45 @  69 */        0x45,
 110         /* 0x46 @  70 */        0x46,
 111         /* 0x47 @  71 */        0x47,
 112         /* 0x48 @  72 */        0x48,
 113         /* 0x49 @  73 */        0x49,
 114         /* 0x4A @  74 */        0x4A,
 115         /* 0x4B @  75 */        0x4B,
 116         /* 0x4C @  76 */        0x4C,
 117         /* 0x4D @  77 */        0x4D,
 118         /* 0x4E @  78 */        0x4E,
 119         /* 0x4F @  79 */        0x4F,
 120         /* 0x50 @  80 */        0x50,
 121         /* 0x51 @  81 */        0x51,
 122         /* 0x52 @  82 */        0x52,
 123         /* 0x53 @  83 */        0x53,
 124         /* 0x54 @  84 */        0x54,
 125         /* 0x55 @  85 */        0x55,
 126         /* 0x56 @  86 */        0x56,
 127         /* 0x57 @  87 */        0x57,
 128         /* 0x58 @  88 */        0x58,
 129         /* 0x59 @  89 */        0x59,
 130         /* 0x5A @  90 */        0x5A,
 131         /* 0x5B @  91 */        0x5B,
 132         /* 0x5C @  92 */        0x5C,
 133         /* 0x5D @  93 */        0x5D,
 134         /* 0x5E @  94 */        0x5E,
 135         /* 0x5F @  95 */        0x5F,
 136         /* 0x60 @  96 */        0x60,
 137         /* 0x61 @  97 */        0x61,
 138         /* 0x62 @  98 */        0x62,
 139         /* 0x63 @  99 */        0x63,
 140         /* 0x64 @ 100 */        0x64,
 141         /* 0x65 @ 101 */        0x65,
 142         /* 0x66 @ 102 */        0x66,
 143         /* 0x67 @ 103 */        0x67,
 144         /* 0x68 @ 104 */        0x68,
 145         /* 0x69 @ 105 */        0x69,
 146         /* 0x6A @ 106 */        0x6A,
 147         /* 0x6B @ 107 */        0x6B,
 148         /* 0x6C @ 108 */        0x6C,
 149         /* 0x6D @ 109 */        0x6D,
 150         /* 0x6E @ 110 */        0x6E,
 151         /* 0x6F @ 111 */        0x6F,
 152         /* 0x70 @ 112 */        0x70,
 153         /* 0x71 @ 113 */        0x71,
 154         /* 0x72 @ 114 */        0x72,
 155         /* 0x73 @ 115 */        0x73,
 156         /* 0x74 @ 116 */        0x74,
 157         /* 0x75 @ 117 */        0x75,
 158         /* 0x76 @ 118 */        0x76,
 159         /* 0x77 @ 119 */        0x77,
 160         /* 0x78 @ 120 */        0x78,
 161         /* 0x79 @ 121 */        0x79,
 162         /* 0x7A @ 122 */        0x7A,
 163         /* 0x7B @ 123 */        0x7B,
 164         /* 0x7C @ 124 */        0x7C,
 165         /* 0x7D @ 125 */        0x7D,
 166         /* 0x7E @ 126 */        0x7E,
 167         /* 0x7F @ 127 */        0x7F,
 168         /* 0x80 @ 128 */        0xE2, 0x82, 0xAC,
 169         /* 0x81 @ 131 */        0xEF, 0xBF, 0xBD,
 170         /* 0x82 @ 134 */        0xE2, 0x80, 0x9A,
 171         /* 0x83 @ 137 */        0xC6, 0x92,
 172         /* 0x84 @ 139 */        0xE2, 0x80, 0x9E,
 173         /* 0x85 @ 142 */        0xE2, 0x80, 0xA6,
 174         /* 0x86 @ 145 */        0xE2, 0x80, 0xA0,
 175         /* 0x87 @ 148 */        0xE2, 0x80, 0xA1,
 176         /* 0x88 @ 151 */        0xCB, 0x86,
 177         /* 0x89 @ 153 */        0xE2, 0x80, 0xB0,
 178         /* 0x8A @ 156 */        0xC5, 0xA0,
 179         /* 0x8B @ 158 */        0xE2, 0x80, 0xB9,
 180         /* 0x8C @ 161 */        0xC5, 0x92,
 181         /* 0x8D @ 163 */        0xEF, 0xBF, 0xBD,
 182         /* 0x8E @ 166 */        0xC5, 0xBD,
 183         /* 0x8F @ 168 */        0xEF, 0xBF, 0xBD,
 184         /* 0x90 @ 171 */        0xEF, 0xBF, 0xBD,
 185         /* 0x91 @ 174 */        0xE2, 0x80, 0x98,
 186         /* 0x92 @ 177 */        0xE2, 0x80, 0x99,
 187         /* 0x93 @ 180 */        0xE2, 0x80, 0x9C,
 188         /* 0x94 @ 183 */        0xE2, 0x80, 0x9D,
 189         /* 0x95 @ 186 */        0xE2, 0x80, 0xA2,
 190         /* 0x96 @ 189 */        0xE2, 0x80, 0x93,
 191         /* 0x97 @ 192 */        0xE2, 0x80, 0x94,
 192         /* 0x98 @ 195 */        0xCB, 0x9C,
 193         /* 0x99 @ 197 */        0xE2, 0x84, 0xA2,
 194         /* 0x9A @ 200 */        0xC5, 0xA1,
 195         /* 0x9B @ 202 */        0xE2, 0x80, 0xBA,
 196         /* 0x9C @ 205 */        0xC5, 0x93,
 197         /* 0x9D @ 207 */        0xEF, 0xBF, 0xBD,
 198         /* 0x9E @ 210 */        0xC5, 0xBE,
 199         /* 0x9F @ 212 */        0xC5, 0xB8,
 200         /* 0xA0 @ 214 */        0xC2, 0xA0,
 201         /* 0xA1 @ 216 */        0xC2, 0xA1,
 202         /* 0xA2 @ 218 */        0xC2, 0xA2,
 203         /* 0xA3 @ 220 */        0xC2, 0xA3,
 204         /* 0xA4 @ 222 */        0xC2, 0xA4,
 205         /* 0xA5 @ 224 */        0xC2, 0xA5,
 206         /* 0xA6 @ 226 */        0xC2, 0xA6,
 207         /* 0xA7 @ 228 */        0xC2, 0xA7,
 208         /* 0xA8 @ 230 */        0xC2, 0xA8,
 209         /* 0xA9 @ 232 */        0xC2, 0xA9,
 210         /* 0xAA @ 234 */        0xC2, 0xAA,
 211         /* 0xAB @ 236 */        0xC2, 0xAB,
 212         /* 0xAC @ 238 */        0xC2, 0xAC,
 213         /* 0xAD @ 240 */        0xC2, 0xAD,
 214         /* 0xAE @ 242 */        0xC2, 0xAE,
 215         /* 0xAF @ 244 */        0xC2, 0xAF,
 216         /* 0xB0 @ 246 */        0xC2, 0xB0,
 217         /* 0xB1 @ 248 */        0xC2, 0xB1,
 218         /* 0xB2 @ 250 */        0xC2, 0xB2,
 219         /* 0xB3 @ 252 */        0xC2, 0xB3,
 220         /* 0xB4 @ 254 */        0xC2, 0xB4,
 221         /* 0xB5 @ 256 */        0xC2, 0xB5,
 222         /* 0xB6 @ 258 */        0xC2, 0xB6,
 223         /* 0xB7 @ 260 */        0xC2, 0xB7,
 224         /* 0xB8 @ 262 */        0xC2, 0xB8,
 225         /* 0xB9 @ 264 */        0xC2, 0xB9,
 226         /* 0xBA @ 266 */        0xC2, 0xBA,
 227         /* 0xBB @ 268 */        0xC2, 0xBB,
 228         /* 0xBC @ 270 */        0xC2, 0xBC,
 229         /* 0xBD @ 272 */        0xC2, 0xBD,
 230         /* 0xBE @ 274 */        0xC2, 0xBE,
 231         /* 0xBF @ 276 */        0xC2, 0xBF,
 232         /* 0xC0 @ 278 */        0xC3, 0x80,
 233         /* 0xC1 @ 280 */        0xC3, 0x81,
 234         /* 0xC2 @ 282 */        0xC3, 0x82,
 235         /* 0xC3 @ 284 */        0xC3, 0x83,
 236         /* 0xC4 @ 286 */        0xC3, 0x84,
 237         /* 0xC5 @ 288 */        0xC3, 0x85,
 238         /* 0xC6 @ 290 */        0xC3, 0x86,
 239         /* 0xC7 @ 292 */        0xC3, 0x87,
 240         /* 0xC8 @ 294 */        0xC3, 0x88,
 241         /* 0xC9 @ 296 */        0xC3, 0x89,
 242         /* 0xCA @ 298 */        0xC3, 0x8A,
 243         /* 0xCB @ 300 */        0xC3, 0x8B,
 244         /* 0xCC @ 302 */        0xC3, 0x8C,
 245         /* 0xCD @ 304 */        0xC3, 0x8D,
 246         /* 0xCE @ 306 */        0xC3, 0x8E,
 247         /* 0xCF @ 308 */        0xC3, 0x8F,
 248         /* 0xD0 @ 310 */        0xC3, 0x90,
 249         /* 0xD1 @ 312 */        0xC3, 0x91,
 250         /* 0xD2 @ 314 */        0xC3, 0x92,
 251         /* 0xD3 @ 316 */        0xC3, 0x93,
 252         /* 0xD4 @ 318 */        0xC3, 0x94,
 253         /* 0xD5 @ 320 */        0xC3, 0x95,
 254         /* 0xD6 @ 322 */        0xC3, 0x96,
 255         /* 0xD7 @ 324 */        0xC3, 0x97,
 256         /* 0xD8 @ 326 */        0xC3, 0x98,
 257         /* 0xD9 @ 328 */        0xC3, 0x99,
 258         /* 0xDA @ 330 */        0xC3, 0x9A,
 259         /* 0xDB @ 332 */        0xC3, 0x9B,
 260         /* 0xDC @ 334 */        0xC3, 0x9C,
 261         /* 0xDD @ 336 */        0xC3, 0x9D,
 262         /* 0xDE @ 338 */        0xC3, 0x9E,
 263         /* 0xDF @ 340 */        0xC3, 0x9F,
 264         /* 0xE0 @ 342 */        0xC3, 0xA0,
 265         /* 0xE1 @ 344 */        0xC3, 0xA1,
 266         /* 0xE2 @ 346 */        0xC3, 0xA2,
 267         /* 0xE3 @ 348 */        0xC3, 0xA3,
 268         /* 0xE4 @ 350 */        0xC3, 0xA4,
 269         /* 0xE5 @ 352 */        0xC3, 0xA5,
 270         /* 0xE6 @ 354 */        0xC3, 0xA6,
 271         /* 0xE7 @ 356 */        0xC3, 0xA7,
 272         /* 0xE8 @ 358 */        0xC3, 0xA8,
 273         /* 0xE9 @ 360 */        0xC3, 0xA9,
 274         /* 0xEA @ 362 */        0xC3, 0xAA,
 275         /* 0xEB @ 364 */        0xC3, 0xAB,
 276         /* 0xEC @ 366 */        0xC3, 0xAC,
 277         /* 0xED @ 368 */        0xC3, 0xAD,
 278         /* 0xEE @ 370 */        0xC3, 0xAE,
 279         /* 0xEF @ 372 */        0xC3, 0xAF,
 280         /* 0xF0 @ 374 */        0xC3, 0xB0,
 281         /* 0xF1 @ 376 */        0xC3, 0xB1,
 282         /* 0xF2 @ 378 */        0xC3, 0xB2,
 283         /* 0xF3 @ 380 */        0xC3, 0xB3,
 284         /* 0xF4 @ 382 */        0xC3, 0xB4,
 285         /* 0xF5 @ 384 */        0xC3, 0xB5,
 286         /* 0xF6 @ 386 */        0xC3, 0xB6,
 287         /* 0xF7 @ 388 */        0xC3, 0xB7,
 288         /* 0xF8 @ 390 */        0xC3, 0xB8,
 289         /* 0xF9 @ 392 */        0xC3, 0xB9,
 290         /* 0xFA @ 394 */        0xC3, 0xBA,
 291         /* 0xFB @ 396 */        0xC3, 0xBB,
 292         /* 0xFC @ 398 */        0xC3, 0xBC,
 293         /* 0xFD @ 400 */        0xC3, 0xBD,
 294         /* 0xFE @ 402 */        0xC3, 0xBE,
 295         /* 0xFF @ 404 */        0xC3, 0xBF,
 296 };
 297
 298 static const uint16_t tblofs[257] = {
 299         /* 0x00 */   0,   1,   2,   3,   4,   5,   6,   7,
 300         /* 0x08 */   8,   9,  10,  11,  12,  13,  14,  15,
 301         /* 0x10 */  16,  17,  18,  19,  20,  21,  22,  23,
 302         /* 0x18 */  24,  25,  26,  27,  28,  29,  30,  31,
 303         /* 0x20 */  32,  33,  34,  35,  36,  37,  38,  39,
 304         /* 0x28 */  40,  41,  42,  43,  44,  45,  46,  47,
 305         /* 0x30 */  48,  49,  50,  51,  52,  53,  54,  55,
 306         /* 0x38 */  56,  57,  58,  59,  60,  61,  62,  63,
 307         /* 0x40 */  64,  65,  66,  67,  68,  69,  70,  71,
 308         /* 0x48 */  72,  73,  74,  75,  76,  77,  78,  79,
 309         /* 0x50 */  80,  81,  82,  83,  84,  85,  86,  87,
 310         /* 0x58 */  88,  89,  90,  91,  92,  93,  94,  95,
 311         /* 0x60 */  96,  97,  98,  99, 100, 101, 102, 103,
 312         /* 0x68 */ 104, 105, 106, 107, 108, 109, 110, 111,
 313         /* 0x70 */ 112, 113, 114, 115, 116, 117, 118, 119,
 314         /* 0x78 */ 120, 121, 122, 123, 124, 125, 126, 127,
 315         /* 0x80 */ 128, 131, 134, 137, 139, 142, 145, 148,
 316         /* 0x88 */ 151, 153, 156, 158, 161, 163, 166, 168,
 317         /* 0x90 */ 171, 174, 177, 180, 183, 186, 189, 192,
 318         /* 0x98 */ 195, 197, 200, 202, 205, 207, 210, 212,
 319         /* 0xA0 */ 214, 216, 218, 220, 222, 224, 226, 228,
 320         /* 0xA8 */ 230, 232, 234, 236, 238, 240, 242, 244,
 321         /* 0xB0 */ 246, 248, 250, 252, 254, 256, 258, 260,
 322         /* 0xB8 */ 262, 264, 266, 268, 270, 272, 274, 276,
 323         /* 0xC0 */ 278, 280, 282, 284, 286, 288, 290, 292,
 324         /* 0xC8 */ 294, 296, 298, 300, 302, 304, 306, 308,
 325         /* 0xD0 */ 310, 312, 314, 316, 318, 320, 322, 324,
 326         /* 0xD8 */ 326, 328, 330, 332, 334, 336, 338, 340,
 327         /* 0xE0 */ 342, 344, 346, 348, 350, 352, 354, 356,
 328         /* 0xE8 */ 358, 360, 362, 364, 366, 368, 370, 372,
 329         /* 0xF0 */ 374, 376, 378, 380, 382, 384, 386, 388,
 330         /* 0xF8 */ 390, 392, 394, 396, 398, 400, 402, 404,
 331         /* sizeof (cp1252_utf8) */ 406
 332 };
 333
 334 /* Check if a string qualifies as UTF-8. */
 335 static int
 336 is_utf8(const char* src)
 337 {
 338         uint8_t ch;
 339         size_t i;
 340         const uint8_t* s = (const uint8_t*) src;
 341
 342         /* We make a loop over every character, until we find a null one.
 343            Remember: The string is supposed to end with a NUL, so ahead checks are safe. */
 344         while ((ch = *s++))     {
 345                 /* Ye olde 7bit ASCII chars 'rr fine for anything */
 346                 if(ch < 0x80) continue;
 347
 348                 /* Now, we watch out for non-UTF conform sequences. */
 349                 else if ((ch < 0xC2) || (ch > 0xFD))
 350                         return 0;
 351                 /* check for some misformed sequences */
 352                 if (((ch == 0xC2) && (s[0] < 0xA0)) ||
 353                     ((ch == 0xEF) && (s[0] == 0xBF) && (s[1] > 0xBD)))
 354                         /* XXX add more for outside the BMP */
 355                         return 0;
 356
 357                 /* Check the continuation bytes. */
 358                 if      (ch < 0xE0) i = 1;
 359                 else if (ch < 0xF0) i = 2;
 360                 else if (ch < 0xF8)     i = 3;
 361                 else if (ch < 0xFC)     i = 4;
 362                 else
 363                         i = 5;
 364
 365                 while (i--)
 366                         if ((*s++ & 0xC0) != 0x80)
 367                                 return 0;
 368         }
 369
 370         /* If no check failed, the string indeed looks like valid UTF-8. */
 371         return 1;
 372 }
 373
 374 /* The main conversion routine.
 375    ICY in CP-1252 (or UTF-8 alreay) to UTF-8 encoded string.
 376    If force is applied, it will always encode to UTF-8, without checking. */
 377 char *
 378 icy2utf8(const char *src, int force)
 379 {
 380         const uint8_t *s = (const uint8_t *)src;
 381         size_t srclen, dstlen, i, k;
 382         uint8_t ch, *d;
 383         char *dst;
 384
 385         /* Some funny streams from Apple/iTunes give ICY info in UTF-8 already.
 386            So, be prepared and don't try to re-encode such. Unless forced. */
 387         if(!force && is_utf8(src)) return (compat_strdup(src));
 388
 389         srclen = strlen(src) + 1;
 390         /* allocate conservatively */
 391         if ((d = malloc(srclen * 3)) == NULL)
 392                 return (NULL);
 393
 394         i = 0;
 395         dstlen = 0;
 396         while (i < srclen) {
 397                 ch = s[i++];
 398                 k = tblofs[ch];
 399                 while (k < tblofs[ch + 1])
 400                         d[dstlen++] = cp1252_utf8[k++];
 401         }
 402
 403         /* dstlen includes trailing NUL since srclen also does */
 404         if ((dst = realloc(d, dstlen)) == NULL) {
 405                 free(d);
 406                 return (NULL);
 407         }
 408         return (dst);
 409 }
 410
 411 /* This stuff is for testing only. */
 412 #ifdef TEST
 413 static const char intext[] = "\225 Gr\374\337e kosten 0,55 \200\205";
 414
 415 #include <stdio.h>
 416
 417 int
 418 main(void)
 419 {
 420         char *t, *t2;
 421
 422         if ((t = icy2utf8(intext, 0)) == NULL) {
 423                 fprintf(stderr, "out of memory\n");
 424                 return (1);
 425         }
 426
 427         /* make sure it won't be converted twice */
 428         if ((t2 = icy2utf8(t), 0) == NULL) {
 429                 fprintf(stderr, "out of memory\n");
 430                 return (1);
 431         }
 432
 433         printf("Result is:\t\343\200\214%s\343\200\215\n"
 434                 "\t\t\343\200\214%s\343\200\215\n", t, t2);
 435
 436         free(t);
 437         free(t2);
 438         return (0);
 439 }
 440 #endif