Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
wine-mirror
GitHub Repository: wine-mirror/wine
Path: blob/master/libs/mpg123/src/libmpg123/icy2utf8.c
4394 views
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 <[email protected]>
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
/* Includes string and stdlib headers... */
31
#include "../compat/compat.h"
32
33
/* ThOr: too lazy for this type check; also we use char/short all around anyway.
34
Of cource, it would be the proper way to use _these_ kind of types all around. */
35
#define uint8_t unsigned char
36
#define uint16_t unsigned short
37
38
static const uint8_t cp1252_utf8[] = {
39
/* 0x00 @ 0 */ 0x00,
40
/* 0x01 @ 1 */ 0x01,
41
/* 0x02 @ 2 */ 0x02,
42
/* 0x03 @ 3 */ 0x03,
43
/* 0x04 @ 4 */ 0x04,
44
/* 0x05 @ 5 */ 0x05,
45
/* 0x06 @ 6 */ 0x06,
46
/* 0x07 @ 7 */ 0x07,
47
/* 0x08 @ 8 */ 0x08,
48
/* 0x09 @ 9 */ 0x09,
49
/* 0x0A @ 10 */ 0x0A,
50
/* 0x0B @ 11 */ 0x0B,
51
/* 0x0C @ 12 */ 0x0C,
52
/* 0x0D @ 13 */ 0x0D,
53
/* 0x0E @ 14 */ 0x0E,
54
/* 0x0F @ 15 */ 0x0F,
55
/* 0x10 @ 16 */ 0x10,
56
/* 0x11 @ 17 */ 0x11,
57
/* 0x12 @ 18 */ 0x12,
58
/* 0x13 @ 19 */ 0x13,
59
/* 0x14 @ 20 */ 0x14,
60
/* 0x15 @ 21 */ 0x15,
61
/* 0x16 @ 22 */ 0x16,
62
/* 0x17 @ 23 */ 0x17,
63
/* 0x18 @ 24 */ 0x18,
64
/* 0x19 @ 25 */ 0x19,
65
/* 0x1A @ 26 */ 0x1A,
66
/* 0x1B @ 27 */ 0x1B,
67
/* 0x1C @ 28 */ 0x1C,
68
/* 0x1D @ 29 */ 0x1D,
69
/* 0x1E @ 30 */ 0x1E,
70
/* 0x1F @ 31 */ 0x1F,
71
/* 0x20 @ 32 */ 0x20,
72
/* 0x21 @ 33 */ 0x21,
73
/* 0x22 @ 34 */ 0x22,
74
/* 0x23 @ 35 */ 0x23,
75
/* 0x24 @ 36 */ 0x24,
76
/* 0x25 @ 37 */ 0x25,
77
/* 0x26 @ 38 */ 0x26,
78
/* 0x27 @ 39 */ 0x27,
79
/* 0x28 @ 40 */ 0x28,
80
/* 0x29 @ 41 */ 0x29,
81
/* 0x2A @ 42 */ 0x2A,
82
/* 0x2B @ 43 */ 0x2B,
83
/* 0x2C @ 44 */ 0x2C,
84
/* 0x2D @ 45 */ 0x2D,
85
/* 0x2E @ 46 */ 0x2E,
86
/* 0x2F @ 47 */ 0x2F,
87
/* 0x30 @ 48 */ 0x30,
88
/* 0x31 @ 49 */ 0x31,
89
/* 0x32 @ 50 */ 0x32,
90
/* 0x33 @ 51 */ 0x33,
91
/* 0x34 @ 52 */ 0x34,
92
/* 0x35 @ 53 */ 0x35,
93
/* 0x36 @ 54 */ 0x36,
94
/* 0x37 @ 55 */ 0x37,
95
/* 0x38 @ 56 */ 0x38,
96
/* 0x39 @ 57 */ 0x39,
97
/* 0x3A @ 58 */ 0x3A,
98
/* 0x3B @ 59 */ 0x3B,
99
/* 0x3C @ 60 */ 0x3C,
100
/* 0x3D @ 61 */ 0x3D,
101
/* 0x3E @ 62 */ 0x3E,
102
/* 0x3F @ 63 */ 0x3F,
103
/* 0x40 @ 64 */ 0x40,
104
/* 0x41 @ 65 */ 0x41,
105
/* 0x42 @ 66 */ 0x42,
106
/* 0x43 @ 67 */ 0x43,
107
/* 0x44 @ 68 */ 0x44,
108
/* 0x45 @ 69 */ 0x45,
109
/* 0x46 @ 70 */ 0x46,
110
/* 0x47 @ 71 */ 0x47,
111
/* 0x48 @ 72 */ 0x48,
112
/* 0x49 @ 73 */ 0x49,
113
/* 0x4A @ 74 */ 0x4A,
114
/* 0x4B @ 75 */ 0x4B,
115
/* 0x4C @ 76 */ 0x4C,
116
/* 0x4D @ 77 */ 0x4D,
117
/* 0x4E @ 78 */ 0x4E,
118
/* 0x4F @ 79 */ 0x4F,
119
/* 0x50 @ 80 */ 0x50,
120
/* 0x51 @ 81 */ 0x51,
121
/* 0x52 @ 82 */ 0x52,
122
/* 0x53 @ 83 */ 0x53,
123
/* 0x54 @ 84 */ 0x54,
124
/* 0x55 @ 85 */ 0x55,
125
/* 0x56 @ 86 */ 0x56,
126
/* 0x57 @ 87 */ 0x57,
127
/* 0x58 @ 88 */ 0x58,
128
/* 0x59 @ 89 */ 0x59,
129
/* 0x5A @ 90 */ 0x5A,
130
/* 0x5B @ 91 */ 0x5B,
131
/* 0x5C @ 92 */ 0x5C,
132
/* 0x5D @ 93 */ 0x5D,
133
/* 0x5E @ 94 */ 0x5E,
134
/* 0x5F @ 95 */ 0x5F,
135
/* 0x60 @ 96 */ 0x60,
136
/* 0x61 @ 97 */ 0x61,
137
/* 0x62 @ 98 */ 0x62,
138
/* 0x63 @ 99 */ 0x63,
139
/* 0x64 @ 100 */ 0x64,
140
/* 0x65 @ 101 */ 0x65,
141
/* 0x66 @ 102 */ 0x66,
142
/* 0x67 @ 103 */ 0x67,
143
/* 0x68 @ 104 */ 0x68,
144
/* 0x69 @ 105 */ 0x69,
145
/* 0x6A @ 106 */ 0x6A,
146
/* 0x6B @ 107 */ 0x6B,
147
/* 0x6C @ 108 */ 0x6C,
148
/* 0x6D @ 109 */ 0x6D,
149
/* 0x6E @ 110 */ 0x6E,
150
/* 0x6F @ 111 */ 0x6F,
151
/* 0x70 @ 112 */ 0x70,
152
/* 0x71 @ 113 */ 0x71,
153
/* 0x72 @ 114 */ 0x72,
154
/* 0x73 @ 115 */ 0x73,
155
/* 0x74 @ 116 */ 0x74,
156
/* 0x75 @ 117 */ 0x75,
157
/* 0x76 @ 118 */ 0x76,
158
/* 0x77 @ 119 */ 0x77,
159
/* 0x78 @ 120 */ 0x78,
160
/* 0x79 @ 121 */ 0x79,
161
/* 0x7A @ 122 */ 0x7A,
162
/* 0x7B @ 123 */ 0x7B,
163
/* 0x7C @ 124 */ 0x7C,
164
/* 0x7D @ 125 */ 0x7D,
165
/* 0x7E @ 126 */ 0x7E,
166
/* 0x7F @ 127 */ 0x7F,
167
/* 0x80 @ 128 */ 0xE2, 0x82, 0xAC,
168
/* 0x81 @ 131 */ 0xEF, 0xBF, 0xBD,
169
/* 0x82 @ 134 */ 0xE2, 0x80, 0x9A,
170
/* 0x83 @ 137 */ 0xC6, 0x92,
171
/* 0x84 @ 139 */ 0xE2, 0x80, 0x9E,
172
/* 0x85 @ 142 */ 0xE2, 0x80, 0xA6,
173
/* 0x86 @ 145 */ 0xE2, 0x80, 0xA0,
174
/* 0x87 @ 148 */ 0xE2, 0x80, 0xA1,
175
/* 0x88 @ 151 */ 0xCB, 0x86,
176
/* 0x89 @ 153 */ 0xE2, 0x80, 0xB0,
177
/* 0x8A @ 156 */ 0xC5, 0xA0,
178
/* 0x8B @ 158 */ 0xE2, 0x80, 0xB9,
179
/* 0x8C @ 161 */ 0xC5, 0x92,
180
/* 0x8D @ 163 */ 0xEF, 0xBF, 0xBD,
181
/* 0x8E @ 166 */ 0xC5, 0xBD,
182
/* 0x8F @ 168 */ 0xEF, 0xBF, 0xBD,
183
/* 0x90 @ 171 */ 0xEF, 0xBF, 0xBD,
184
/* 0x91 @ 174 */ 0xE2, 0x80, 0x98,
185
/* 0x92 @ 177 */ 0xE2, 0x80, 0x99,
186
/* 0x93 @ 180 */ 0xE2, 0x80, 0x9C,
187
/* 0x94 @ 183 */ 0xE2, 0x80, 0x9D,
188
/* 0x95 @ 186 */ 0xE2, 0x80, 0xA2,
189
/* 0x96 @ 189 */ 0xE2, 0x80, 0x93,
190
/* 0x97 @ 192 */ 0xE2, 0x80, 0x94,
191
/* 0x98 @ 195 */ 0xCB, 0x9C,
192
/* 0x99 @ 197 */ 0xE2, 0x84, 0xA2,
193
/* 0x9A @ 200 */ 0xC5, 0xA1,
194
/* 0x9B @ 202 */ 0xE2, 0x80, 0xBA,
195
/* 0x9C @ 205 */ 0xC5, 0x93,
196
/* 0x9D @ 207 */ 0xEF, 0xBF, 0xBD,
197
/* 0x9E @ 210 */ 0xC5, 0xBE,
198
/* 0x9F @ 212 */ 0xC5, 0xB8,
199
/* 0xA0 @ 214 */ 0xC2, 0xA0,
200
/* 0xA1 @ 216 */ 0xC2, 0xA1,
201
/* 0xA2 @ 218 */ 0xC2, 0xA2,
202
/* 0xA3 @ 220 */ 0xC2, 0xA3,
203
/* 0xA4 @ 222 */ 0xC2, 0xA4,
204
/* 0xA5 @ 224 */ 0xC2, 0xA5,
205
/* 0xA6 @ 226 */ 0xC2, 0xA6,
206
/* 0xA7 @ 228 */ 0xC2, 0xA7,
207
/* 0xA8 @ 230 */ 0xC2, 0xA8,
208
/* 0xA9 @ 232 */ 0xC2, 0xA9,
209
/* 0xAA @ 234 */ 0xC2, 0xAA,
210
/* 0xAB @ 236 */ 0xC2, 0xAB,
211
/* 0xAC @ 238 */ 0xC2, 0xAC,
212
/* 0xAD @ 240 */ 0xC2, 0xAD,
213
/* 0xAE @ 242 */ 0xC2, 0xAE,
214
/* 0xAF @ 244 */ 0xC2, 0xAF,
215
/* 0xB0 @ 246 */ 0xC2, 0xB0,
216
/* 0xB1 @ 248 */ 0xC2, 0xB1,
217
/* 0xB2 @ 250 */ 0xC2, 0xB2,
218
/* 0xB3 @ 252 */ 0xC2, 0xB3,
219
/* 0xB4 @ 254 */ 0xC2, 0xB4,
220
/* 0xB5 @ 256 */ 0xC2, 0xB5,
221
/* 0xB6 @ 258 */ 0xC2, 0xB6,
222
/* 0xB7 @ 260 */ 0xC2, 0xB7,
223
/* 0xB8 @ 262 */ 0xC2, 0xB8,
224
/* 0xB9 @ 264 */ 0xC2, 0xB9,
225
/* 0xBA @ 266 */ 0xC2, 0xBA,
226
/* 0xBB @ 268 */ 0xC2, 0xBB,
227
/* 0xBC @ 270 */ 0xC2, 0xBC,
228
/* 0xBD @ 272 */ 0xC2, 0xBD,
229
/* 0xBE @ 274 */ 0xC2, 0xBE,
230
/* 0xBF @ 276 */ 0xC2, 0xBF,
231
/* 0xC0 @ 278 */ 0xC3, 0x80,
232
/* 0xC1 @ 280 */ 0xC3, 0x81,
233
/* 0xC2 @ 282 */ 0xC3, 0x82,
234
/* 0xC3 @ 284 */ 0xC3, 0x83,
235
/* 0xC4 @ 286 */ 0xC3, 0x84,
236
/* 0xC5 @ 288 */ 0xC3, 0x85,
237
/* 0xC6 @ 290 */ 0xC3, 0x86,
238
/* 0xC7 @ 292 */ 0xC3, 0x87,
239
/* 0xC8 @ 294 */ 0xC3, 0x88,
240
/* 0xC9 @ 296 */ 0xC3, 0x89,
241
/* 0xCA @ 298 */ 0xC3, 0x8A,
242
/* 0xCB @ 300 */ 0xC3, 0x8B,
243
/* 0xCC @ 302 */ 0xC3, 0x8C,
244
/* 0xCD @ 304 */ 0xC3, 0x8D,
245
/* 0xCE @ 306 */ 0xC3, 0x8E,
246
/* 0xCF @ 308 */ 0xC3, 0x8F,
247
/* 0xD0 @ 310 */ 0xC3, 0x90,
248
/* 0xD1 @ 312 */ 0xC3, 0x91,
249
/* 0xD2 @ 314 */ 0xC3, 0x92,
250
/* 0xD3 @ 316 */ 0xC3, 0x93,
251
/* 0xD4 @ 318 */ 0xC3, 0x94,
252
/* 0xD5 @ 320 */ 0xC3, 0x95,
253
/* 0xD6 @ 322 */ 0xC3, 0x96,
254
/* 0xD7 @ 324 */ 0xC3, 0x97,
255
/* 0xD8 @ 326 */ 0xC3, 0x98,
256
/* 0xD9 @ 328 */ 0xC3, 0x99,
257
/* 0xDA @ 330 */ 0xC3, 0x9A,
258
/* 0xDB @ 332 */ 0xC3, 0x9B,
259
/* 0xDC @ 334 */ 0xC3, 0x9C,
260
/* 0xDD @ 336 */ 0xC3, 0x9D,
261
/* 0xDE @ 338 */ 0xC3, 0x9E,
262
/* 0xDF @ 340 */ 0xC3, 0x9F,
263
/* 0xE0 @ 342 */ 0xC3, 0xA0,
264
/* 0xE1 @ 344 */ 0xC3, 0xA1,
265
/* 0xE2 @ 346 */ 0xC3, 0xA2,
266
/* 0xE3 @ 348 */ 0xC3, 0xA3,
267
/* 0xE4 @ 350 */ 0xC3, 0xA4,
268
/* 0xE5 @ 352 */ 0xC3, 0xA5,
269
/* 0xE6 @ 354 */ 0xC3, 0xA6,
270
/* 0xE7 @ 356 */ 0xC3, 0xA7,
271
/* 0xE8 @ 358 */ 0xC3, 0xA8,
272
/* 0xE9 @ 360 */ 0xC3, 0xA9,
273
/* 0xEA @ 362 */ 0xC3, 0xAA,
274
/* 0xEB @ 364 */ 0xC3, 0xAB,
275
/* 0xEC @ 366 */ 0xC3, 0xAC,
276
/* 0xED @ 368 */ 0xC3, 0xAD,
277
/* 0xEE @ 370 */ 0xC3, 0xAE,
278
/* 0xEF @ 372 */ 0xC3, 0xAF,
279
/* 0xF0 @ 374 */ 0xC3, 0xB0,
280
/* 0xF1 @ 376 */ 0xC3, 0xB1,
281
/* 0xF2 @ 378 */ 0xC3, 0xB2,
282
/* 0xF3 @ 380 */ 0xC3, 0xB3,
283
/* 0xF4 @ 382 */ 0xC3, 0xB4,
284
/* 0xF5 @ 384 */ 0xC3, 0xB5,
285
/* 0xF6 @ 386 */ 0xC3, 0xB6,
286
/* 0xF7 @ 388 */ 0xC3, 0xB7,
287
/* 0xF8 @ 390 */ 0xC3, 0xB8,
288
/* 0xF9 @ 392 */ 0xC3, 0xB9,
289
/* 0xFA @ 394 */ 0xC3, 0xBA,
290
/* 0xFB @ 396 */ 0xC3, 0xBB,
291
/* 0xFC @ 398 */ 0xC3, 0xBC,
292
/* 0xFD @ 400 */ 0xC3, 0xBD,
293
/* 0xFE @ 402 */ 0xC3, 0xBE,
294
/* 0xFF @ 404 */ 0xC3, 0xBF,
295
};
296
297
static const uint16_t tblofs[257] = {
298
/* 0x00 */ 0, 1, 2, 3, 4, 5, 6, 7,
299
/* 0x08 */ 8, 9, 10, 11, 12, 13, 14, 15,
300
/* 0x10 */ 16, 17, 18, 19, 20, 21, 22, 23,
301
/* 0x18 */ 24, 25, 26, 27, 28, 29, 30, 31,
302
/* 0x20 */ 32, 33, 34, 35, 36, 37, 38, 39,
303
/* 0x28 */ 40, 41, 42, 43, 44, 45, 46, 47,
304
/* 0x30 */ 48, 49, 50, 51, 52, 53, 54, 55,
305
/* 0x38 */ 56, 57, 58, 59, 60, 61, 62, 63,
306
/* 0x40 */ 64, 65, 66, 67, 68, 69, 70, 71,
307
/* 0x48 */ 72, 73, 74, 75, 76, 77, 78, 79,
308
/* 0x50 */ 80, 81, 82, 83, 84, 85, 86, 87,
309
/* 0x58 */ 88, 89, 90, 91, 92, 93, 94, 95,
310
/* 0x60 */ 96, 97, 98, 99, 100, 101, 102, 103,
311
/* 0x68 */ 104, 105, 106, 107, 108, 109, 110, 111,
312
/* 0x70 */ 112, 113, 114, 115, 116, 117, 118, 119,
313
/* 0x78 */ 120, 121, 122, 123, 124, 125, 126, 127,
314
/* 0x80 */ 128, 131, 134, 137, 139, 142, 145, 148,
315
/* 0x88 */ 151, 153, 156, 158, 161, 163, 166, 168,
316
/* 0x90 */ 171, 174, 177, 180, 183, 186, 189, 192,
317
/* 0x98 */ 195, 197, 200, 202, 205, 207, 210, 212,
318
/* 0xA0 */ 214, 216, 218, 220, 222, 224, 226, 228,
319
/* 0xA8 */ 230, 232, 234, 236, 238, 240, 242, 244,
320
/* 0xB0 */ 246, 248, 250, 252, 254, 256, 258, 260,
321
/* 0xB8 */ 262, 264, 266, 268, 270, 272, 274, 276,
322
/* 0xC0 */ 278, 280, 282, 284, 286, 288, 290, 292,
323
/* 0xC8 */ 294, 296, 298, 300, 302, 304, 306, 308,
324
/* 0xD0 */ 310, 312, 314, 316, 318, 320, 322, 324,
325
/* 0xD8 */ 326, 328, 330, 332, 334, 336, 338, 340,
326
/* 0xE0 */ 342, 344, 346, 348, 350, 352, 354, 356,
327
/* 0xE8 */ 358, 360, 362, 364, 366, 368, 370, 372,
328
/* 0xF0 */ 374, 376, 378, 380, 382, 384, 386, 388,
329
/* 0xF8 */ 390, 392, 394, 396, 398, 400, 402, 404,
330
/* sizeof (cp1252_utf8) */ 406
331
};
332
333
/* Check if a string qualifies as UTF-8. */
334
static int
335
is_utf8(const char* src)
336
{
337
uint8_t ch;
338
size_t i;
339
const uint8_t* s = (const uint8_t*) src;
340
341
/* We make a loop over every character, until we find a null one.
342
Remember: The string is supposed to end with a NUL, so ahead checks are safe. */
343
while ((ch = *s++)) {
344
/* Ye olde 7bit ASCII chars 'rr fine for anything */
345
if(ch < 0x80) continue;
346
347
/* Now, we watch out for non-UTF conform sequences. */
348
else if ((ch < 0xC2) || (ch > 0xFD))
349
return 0;
350
/* check for some misformed sequences */
351
if (((ch == 0xC2) && (s[0] < 0xA0)) ||
352
((ch == 0xEF) && (s[0] == 0xBF) && (s[1] > 0xBD)))
353
/* XXX add more for outside the BMP */
354
return 0;
355
356
/* Check the continuation bytes. */
357
if (ch < 0xE0) i = 1;
358
else if (ch < 0xF0) i = 2;
359
else if (ch < 0xF8) i = 3;
360
else if (ch < 0xFC) i = 4;
361
else
362
i = 5;
363
364
while (i--)
365
if ((*s++ & 0xC0) != 0x80)
366
return 0;
367
}
368
369
/* If no check failed, the string indeed looks like valid UTF-8. */
370
return 1;
371
}
372
373
/* The main conversion routine.
374
ICY in CP-1252 (or UTF-8 alreay) to UTF-8 encoded string.
375
If force is applied, it will always encode to UTF-8, without checking. */
376
char *
377
INT123_icy2utf8(const char *src, int force)
378
{
379
const uint8_t *s = (const uint8_t *)src;
380
size_t srclen, dstlen, i, k;
381
uint8_t ch, *d;
382
char *dst;
383
384
/* Some funny streams from Apple/iTunes give ICY info in UTF-8 already.
385
So, be prepared and don't try to re-encode such. Unless forced. */
386
if(!force && is_utf8(src)) return (INT123_compat_strdup(src));
387
388
srclen = strlen(src) + 1;
389
/* allocate conservatively */
390
if ((d = malloc(srclen * 3)) == NULL)
391
return (NULL);
392
393
i = 0;
394
dstlen = 0;
395
while (i < srclen) {
396
ch = s[i++];
397
k = tblofs[ch];
398
while (k < tblofs[ch + 1])
399
d[dstlen++] = cp1252_utf8[k++];
400
}
401
402
/* dstlen includes trailing NUL since srclen also does */
403
if ((dst = realloc(d, dstlen)) == NULL) {
404
free(d);
405
return (NULL);
406
}
407
return (dst);
408
}
409
410
/* This stuff is for testing only. */
411
#ifdef TEST
412
static const char intext[] = "\225 Gr\374\337e kosten 0,55 \200\205";
413
414
#include <stdio.h>
415
416
int
417
main(void)
418
{
419
char *t, *t2;
420
421
if ((t = INT123_icy2utf8(intext, 0)) == NULL) {
422
fprintf(stderr, "out of memory\n");
423
return (1);
424
}
425
426
/* make sure it won't be converted twice */
427
if ((t2 = INT123_icy2utf8(t), 0) == NULL) {
428
fprintf(stderr, "out of memory\n");
429
return (1);
430
}
431
432
printf("Result is:\t\343\200\214%s\343\200\215\n"
433
"\t\t\343\200\214%s\343\200\215\n", t, t2);
434
435
free(t);
436
free(t2);
437
return (0);
438
}
439
#endif
440
441