1 | <?php |
---|
2 | // +----------------------------------------------------------------------+ |
---|
3 | // | PHP version 5 | |
---|
4 | // +----------------------------------------------------------------------+ |
---|
5 | // | Copyright (c) 2002-2006 James Heinrich, Allan Hansen | |
---|
6 | // +----------------------------------------------------------------------+ |
---|
7 | // | This source file is subject to version 2 of the GPL license, | |
---|
8 | // | that is bundled with this package in the file license.txt and is | |
---|
9 | // | available through the world-wide-web at the following url: | |
---|
10 | // | http://www.gnu.org/copyleft/gpl.html | |
---|
11 | // +----------------------------------------------------------------------+ |
---|
12 | // | getID3() - http://getid3.sourceforge.net or http://www.getid3.org | |
---|
13 | // +----------------------------------------------------------------------+ |
---|
14 | // | Authors: James Heinrich <infoØgetid3*org> | |
---|
15 | // | Allan Hansen <ahØartemis*dk> | |
---|
16 | // +----------------------------------------------------------------------+ |
---|
17 | // | module.lib.iconv_replacement.php | |
---|
18 | // | getID3() library file. | |
---|
19 | // | dependencies: NONE, required by getid3.php if no iconv() present. | |
---|
20 | // +----------------------------------------------------------------------+ |
---|
21 | // |
---|
22 | // $Id: module.lib.iconv_replacement.php 3318 2009-05-20 21:54:10Z vdigital $ |
---|
23 | |
---|
24 | |
---|
25 | class getid3_iconv_replacement |
---|
26 | { |
---|
27 | |
---|
28 | public static function iconv($in_charset, $out_charset, $string) { |
---|
29 | |
---|
30 | if ($in_charset == $out_charset) { |
---|
31 | return $string; |
---|
32 | } |
---|
33 | |
---|
34 | static $supported_charsets = array ( |
---|
35 | 'ISO-8859-1' => 'iso88591', |
---|
36 | 'UTF-8' => 'utf8', |
---|
37 | 'UTF-16BE' => 'utf16be', |
---|
38 | 'UTF-16LE' => 'utf16le', |
---|
39 | 'UTF-16' => 'utf16' |
---|
40 | ); |
---|
41 | |
---|
42 | // Convert |
---|
43 | $function_name = 'iconv_' . @$supported_charsets[$in_charset] . '_' . @$supported_charsets[$out_charset]; |
---|
44 | |
---|
45 | if (is_callable(array('getid3_iconv_replacement', $function_name))) { |
---|
46 | return getid3_iconv_replacement::$function_name($string); |
---|
47 | } |
---|
48 | |
---|
49 | // Invalid charset used |
---|
50 | if (!@$supported_charsets[$in_charset]) { |
---|
51 | throw new getid3_exception('PHP does not have iconv() support - cannot use ' . $in_charset . ' charset.'); |
---|
52 | } |
---|
53 | |
---|
54 | if (!@$supported_charsets[$out_charset]) { |
---|
55 | throw new getid3_exception('PHP does not have iconv() support - cannot use ' . $out_charset . ' charset.'); |
---|
56 | } |
---|
57 | } |
---|
58 | |
---|
59 | |
---|
60 | |
---|
61 | public static function iconv_int_utf8($charval) { |
---|
62 | if ($charval < 128) { |
---|
63 | // 0bbbbbbb |
---|
64 | $newcharstring = chr($charval); |
---|
65 | } elseif ($charval < 2048) { |
---|
66 | // 110bbbbb 10bbbbbb |
---|
67 | $newcharstring = chr(($charval >> 6) | 0xC0); |
---|
68 | $newcharstring .= chr(($charval & 0x3F) | 0x80); |
---|
69 | } elseif ($charval < 65536) { |
---|
70 | // 1110bbbb 10bbbbbb 10bbbbbb |
---|
71 | $newcharstring = chr(($charval >> 12) | 0xE0); |
---|
72 | $newcharstring .= chr(($charval >> 6) | 0xC0); |
---|
73 | $newcharstring .= chr(($charval & 0x3F) | 0x80); |
---|
74 | } else { |
---|
75 | // 11110bbb 10bbbbbb 10bbbbbb 10bbbbbb |
---|
76 | $newcharstring = chr(($charval >> 18) | 0xF0); |
---|
77 | $newcharstring .= chr(($charval >> 12) | 0xC0); |
---|
78 | $newcharstring .= chr(($charval >> 6) | 0xC0); |
---|
79 | $newcharstring .= chr(($charval & 0x3F) | 0x80); |
---|
80 | } |
---|
81 | return $newcharstring; |
---|
82 | } |
---|
83 | |
---|
84 | |
---|
85 | |
---|
86 | // ISO-8859-1 => UTF-8 |
---|
87 | public static function iconv_iso88591_utf8($string, $bom=false) { |
---|
88 | if (function_exists('utf8_encode')) { |
---|
89 | return utf8_encode($string); |
---|
90 | } |
---|
91 | // utf8_encode() unavailable, use getID3()'s iconv() conversions (possibly PHP is compiled without XML support) |
---|
92 | $newcharstring = ''; |
---|
93 | if ($bom) { |
---|
94 | $newcharstring .= "\xEF\xBB\xBF"; |
---|
95 | } |
---|
96 | for ($i = 0; $i < strlen($string); $i++) { |
---|
97 | $charval = ord($string{$i}); |
---|
98 | $newcharstring .= getid3_iconv_replacement::iconv_int_utf8($charval); |
---|
99 | } |
---|
100 | return $newcharstring; |
---|
101 | } |
---|
102 | |
---|
103 | |
---|
104 | |
---|
105 | // ISO-8859-1 => UTF-16BE |
---|
106 | public static function iconv_iso88591_utf16be($string, $bom=false) { |
---|
107 | $newcharstring = ''; |
---|
108 | if ($bom) { |
---|
109 | $newcharstring .= "\xFE\xFF"; |
---|
110 | } |
---|
111 | for ($i = 0; $i < strlen($string); $i++) { |
---|
112 | $newcharstring .= "\x00".$string{$i}; |
---|
113 | } |
---|
114 | return $newcharstring; |
---|
115 | } |
---|
116 | |
---|
117 | |
---|
118 | |
---|
119 | // ISO-8859-1 => UTF-16LE |
---|
120 | public static function iconv_iso88591_utf16le($string, $bom=false) { |
---|
121 | $newcharstring = ''; |
---|
122 | if ($bom) { |
---|
123 | $newcharstring .= "\xFF\xFE"; |
---|
124 | } |
---|
125 | for ($i = 0; $i < strlen($string); $i++) { |
---|
126 | $newcharstring .= $string{$i}."\x00"; |
---|
127 | } |
---|
128 | return $newcharstring; |
---|
129 | } |
---|
130 | |
---|
131 | |
---|
132 | |
---|
133 | // ISO-8859-1 => UTF-16 |
---|
134 | public static function iconv_iso88591_utf16($string) { |
---|
135 | return getid3_lib::iconv_iso88591_utf16le($string, true); |
---|
136 | } |
---|
137 | |
---|
138 | |
---|
139 | |
---|
140 | // UTF-8 => ISO-8859-1 |
---|
141 | public static function iconv_utf8_iso88591($string) { |
---|
142 | if (function_exists('utf8_decode')) { |
---|
143 | return utf8_decode($string); |
---|
144 | } |
---|
145 | // utf8_decode() unavailable, use getID3()'s iconv() conversions (possibly PHP is compiled without XML support) |
---|
146 | $newcharstring = ''; |
---|
147 | $offset = 0; |
---|
148 | $stringlength = strlen($string); |
---|
149 | while ($offset < $stringlength) { |
---|
150 | if ((ord($string{$offset}) | 0x07) == 0xF7) { |
---|
151 | // 11110bbb 10bbbbbb 10bbbbbb 10bbbbbb |
---|
152 | $charval = ((ord($string{($offset + 0)}) & 0x07) << 18) & |
---|
153 | ((ord($string{($offset + 1)}) & 0x3F) << 12) & |
---|
154 | ((ord($string{($offset + 2)}) & 0x3F) << 6) & |
---|
155 | (ord($string{($offset + 3)}) & 0x3F); |
---|
156 | $offset += 4; |
---|
157 | } elseif ((ord($string{$offset}) | 0x0F) == 0xEF) { |
---|
158 | // 1110bbbb 10bbbbbb 10bbbbbb |
---|
159 | $charval = ((ord($string{($offset + 0)}) & 0x0F) << 12) & |
---|
160 | ((ord($string{($offset + 1)}) & 0x3F) << 6) & |
---|
161 | (ord($string{($offset + 2)}) & 0x3F); |
---|
162 | $offset += 3; |
---|
163 | } elseif ((ord($string{$offset}) | 0x1F) == 0xDF) { |
---|
164 | // 110bbbbb 10bbbbbb |
---|
165 | $charval = ((ord($string{($offset + 0)}) & 0x1F) << 6) & |
---|
166 | (ord($string{($offset + 1)}) & 0x3F); |
---|
167 | $offset += 2; |
---|
168 | } elseif ((ord($string{$offset}) | 0x7F) == 0x7F) { |
---|
169 | // 0bbbbbbb |
---|
170 | $charval = ord($string{$offset}); |
---|
171 | $offset += 1; |
---|
172 | } else { |
---|
173 | // error? throw some kind of warning here? |
---|
174 | $charval = false; |
---|
175 | $offset += 1; |
---|
176 | } |
---|
177 | if ($charval !== false) { |
---|
178 | $newcharstring .= (($charval < 256) ? chr($charval) : '?'); |
---|
179 | } |
---|
180 | } |
---|
181 | return $newcharstring; |
---|
182 | } |
---|
183 | |
---|
184 | |
---|
185 | |
---|
186 | // UTF-8 => UTF-16BE |
---|
187 | public static function iconv_utf8_utf16be($string, $bom=false) { |
---|
188 | $newcharstring = ''; |
---|
189 | if ($bom) { |
---|
190 | $newcharstring .= "\xFE\xFF"; |
---|
191 | } |
---|
192 | $offset = 0; |
---|
193 | $stringlength = strlen($string); |
---|
194 | while ($offset < $stringlength) { |
---|
195 | if ((ord($string{$offset}) | 0x07) == 0xF7) { |
---|
196 | // 11110bbb 10bbbbbb 10bbbbbb 10bbbbbb |
---|
197 | $charval = ((ord($string{($offset + 0)}) & 0x07) << 18) & |
---|
198 | ((ord($string{($offset + 1)}) & 0x3F) << 12) & |
---|
199 | ((ord($string{($offset + 2)}) & 0x3F) << 6) & |
---|
200 | (ord($string{($offset + 3)}) & 0x3F); |
---|
201 | $offset += 4; |
---|
202 | } elseif ((ord($string{$offset}) | 0x0F) == 0xEF) { |
---|
203 | // 1110bbbb 10bbbbbb 10bbbbbb |
---|
204 | $charval = ((ord($string{($offset + 0)}) & 0x0F) << 12) & |
---|
205 | ((ord($string{($offset + 1)}) & 0x3F) << 6) & |
---|
206 | (ord($string{($offset + 2)}) & 0x3F); |
---|
207 | $offset += 3; |
---|
208 | } elseif ((ord($string{$offset}) | 0x1F) == 0xDF) { |
---|
209 | // 110bbbbb 10bbbbbb |
---|
210 | $charval = ((ord($string{($offset + 0)}) & 0x1F) << 6) & |
---|
211 | (ord($string{($offset + 1)}) & 0x3F); |
---|
212 | $offset += 2; |
---|
213 | } elseif ((ord($string{$offset}) | 0x7F) == 0x7F) { |
---|
214 | // 0bbbbbbb |
---|
215 | $charval = ord($string{$offset}); |
---|
216 | $offset += 1; |
---|
217 | } else { |
---|
218 | // error? throw some kind of warning here? |
---|
219 | $charval = false; |
---|
220 | $offset += 1; |
---|
221 | } |
---|
222 | if ($charval !== false) { |
---|
223 | $newcharstring .= (($charval < 65536) ? getid3_lib::BigEndian2String($charval, 2) : "\x00".'?'); |
---|
224 | } |
---|
225 | } |
---|
226 | return $newcharstring; |
---|
227 | } |
---|
228 | |
---|
229 | |
---|
230 | |
---|
231 | // UTF-8 => UTF-16LE |
---|
232 | public static function iconv_utf8_utf16le($string, $bom=false) { |
---|
233 | $newcharstring = ''; |
---|
234 | if ($bom) { |
---|
235 | $newcharstring .= "\xFF\xFE"; |
---|
236 | } |
---|
237 | $offset = 0; |
---|
238 | $stringlength = strlen($string); |
---|
239 | while ($offset < $stringlength) { |
---|
240 | if ((ord($string{$offset}) | 0x07) == 0xF7) { |
---|
241 | // 11110bbb 10bbbbbb 10bbbbbb 10bbbbbb |
---|
242 | $charval = ((ord($string{($offset + 0)}) & 0x07) << 18) & |
---|
243 | ((ord($string{($offset + 1)}) & 0x3F) << 12) & |
---|
244 | ((ord($string{($offset + 2)}) & 0x3F) << 6) & |
---|
245 | (ord($string{($offset + 3)}) & 0x3F); |
---|
246 | $offset += 4; |
---|
247 | } elseif ((ord($string{$offset}) | 0x0F) == 0xEF) { |
---|
248 | // 1110bbbb 10bbbbbb 10bbbbbb |
---|
249 | $charval = ((ord($string{($offset + 0)}) & 0x0F) << 12) & |
---|
250 | ((ord($string{($offset + 1)}) & 0x3F) << 6) & |
---|
251 | (ord($string{($offset + 2)}) & 0x3F); |
---|
252 | $offset += 3; |
---|
253 | } elseif ((ord($string{$offset}) | 0x1F) == 0xDF) { |
---|
254 | // 110bbbbb 10bbbbbb |
---|
255 | $charval = ((ord($string{($offset + 0)}) & 0x1F) << 6) & |
---|
256 | (ord($string{($offset + 1)}) & 0x3F); |
---|
257 | $offset += 2; |
---|
258 | } elseif ((ord($string{$offset}) | 0x7F) == 0x7F) { |
---|
259 | // 0bbbbbbb |
---|
260 | $charval = ord($string{$offset}); |
---|
261 | $offset += 1; |
---|
262 | } else { |
---|
263 | // error? maybe throw some warning here? |
---|
264 | $charval = false; |
---|
265 | $offset += 1; |
---|
266 | } |
---|
267 | if ($charval !== false) { |
---|
268 | $newcharstring .= (($charval < 65536) ? getid3_lib::LittleEndian2String($charval, 2) : '?'."\x00"); |
---|
269 | } |
---|
270 | } |
---|
271 | return $newcharstring; |
---|
272 | } |
---|
273 | |
---|
274 | |
---|
275 | |
---|
276 | // UTF-8 => UTF-16 |
---|
277 | public static function iconv_utf8_utf16($string) { |
---|
278 | return getid3_lib::iconv_utf8_utf16le($string, true); |
---|
279 | } |
---|
280 | |
---|
281 | |
---|
282 | |
---|
283 | // UTF-16BE => ISO-8859-1 |
---|
284 | public static function iconv_utf16be_iso88591($string) { |
---|
285 | if (substr($string, 0, 2) == "\xFE\xFF") { |
---|
286 | // strip BOM |
---|
287 | $string = substr($string, 2); |
---|
288 | } |
---|
289 | $newcharstring = ''; |
---|
290 | for ($i = 0; $i < strlen($string); $i += 2) { |
---|
291 | $charval = getid3_lib::BigEndian2Int(substr($string, $i, 2)); |
---|
292 | $newcharstring .= (($charval < 256) ? chr($charval) : '?'); |
---|
293 | } |
---|
294 | return $newcharstring; |
---|
295 | } |
---|
296 | |
---|
297 | |
---|
298 | |
---|
299 | // UTF-16BE => UTF-8 |
---|
300 | public static function iconv_utf16be_utf8($string) { |
---|
301 | if (substr($string, 0, 2) == "\xFE\xFF") { |
---|
302 | // strip BOM |
---|
303 | $string = substr($string, 2); |
---|
304 | } |
---|
305 | $newcharstring = ''; |
---|
306 | for ($i = 0; $i < strlen($string); $i += 2) { |
---|
307 | $charval = getid3_lib::BigEndian2Int(substr($string, $i, 2)); |
---|
308 | $newcharstring .= getid3_iconv_replacement::iconv_int_utf8($charval); |
---|
309 | } |
---|
310 | return $newcharstring; |
---|
311 | } |
---|
312 | |
---|
313 | |
---|
314 | |
---|
315 | // UTF-16BE => UTF-16LE |
---|
316 | public static function iconv_utf16be_utf16le($string) { |
---|
317 | return getid3_iconv_replacement::iconv_utf8_utf16le(getid3_iconv_replacement::iconv_utf16be_utf8($string)); |
---|
318 | } |
---|
319 | |
---|
320 | |
---|
321 | |
---|
322 | // UTF-16BE => UTF-16 |
---|
323 | public static function iconv_utf16be_utf16($string) { |
---|
324 | return getid3_iconv_replacement::iconv_utf8_utf16(getid3_iconv_replacement::iconv_utf16be_utf8($string)); |
---|
325 | } |
---|
326 | |
---|
327 | |
---|
328 | |
---|
329 | // UTF-16LE => ISO-8859-1 |
---|
330 | public static function iconv_utf16le_iso88591($string) { |
---|
331 | if (substr($string, 0, 2) == "\xFF\xFE") { |
---|
332 | // strip BOM |
---|
333 | $string = substr($string, 2); |
---|
334 | } |
---|
335 | $newcharstring = ''; |
---|
336 | for ($i = 0; $i < strlen($string); $i += 2) { |
---|
337 | $charval = getid3_lib::LittleEndian2Int(substr($string, $i, 2)); |
---|
338 | $newcharstring .= (($charval < 256) ? chr($charval) : '?'); |
---|
339 | } |
---|
340 | return $newcharstring; |
---|
341 | } |
---|
342 | |
---|
343 | |
---|
344 | |
---|
345 | // UTF-16LE => UTF-8 |
---|
346 | public static function iconv_utf16le_utf8($string) { |
---|
347 | if (substr($string, 0, 2) == "\xFF\xFE") { |
---|
348 | // strip BOM |
---|
349 | $string = substr($string, 2); |
---|
350 | } |
---|
351 | $newcharstring = ''; |
---|
352 | for ($i = 0; $i < strlen($string); $i += 2) { |
---|
353 | $charval = getid3_lib::LittleEndian2Int(substr($string, $i, 2)); |
---|
354 | $newcharstring .= getid3_iconv_replacement::iconv_int_utf8($charval); |
---|
355 | } |
---|
356 | return $newcharstring; |
---|
357 | } |
---|
358 | |
---|
359 | |
---|
360 | |
---|
361 | // UTF-16LE => UTF-16BE |
---|
362 | public static function iconv_utf16le_utf16be($string) { |
---|
363 | return getid3_iconv_replacement::iconv_utf8_utf16be(getid3_iconv_replacement::iconv_utf16le_utf8($string)); |
---|
364 | } |
---|
365 | |
---|
366 | |
---|
367 | |
---|
368 | // UTF-16LE => UTF-16 |
---|
369 | public static function iconv_utf16le_utf16($string) { |
---|
370 | return getid3_iconv_replacement::iconv_utf8_utf16(getid3_iconv_replacement::iconv_utf16le_utf8($string)); |
---|
371 | } |
---|
372 | |
---|
373 | |
---|
374 | |
---|
375 | // UTF-16 => ISO-8859-1 |
---|
376 | public static function iconv_utf16_iso88591($string) { |
---|
377 | $bom = substr($string, 0, 2); |
---|
378 | if ($bom == "\xFE\xFF") { |
---|
379 | return getid3_lib::iconv_utf16be_iso88591(substr($string, 2)); |
---|
380 | } elseif ($bom == "\xFF\xFE") { |
---|
381 | return getid3_lib::iconv_utf16le_iso88591(substr($string, 2)); |
---|
382 | } |
---|
383 | return $string; |
---|
384 | } |
---|
385 | |
---|
386 | |
---|
387 | |
---|
388 | // UTF-16 => UTF-8 |
---|
389 | public static function iconv_utf16_utf8($string) { |
---|
390 | $bom = substr($string, 0, 2); |
---|
391 | if ($bom == "\xFE\xFF") { |
---|
392 | return getid3_iconv_replacement::iconv_utf16be_utf8(substr($string, 2)); |
---|
393 | } elseif ($bom == "\xFF\xFE") { |
---|
394 | return getid3_iconv_replacement::iconv_utf16le_utf8(substr($string, 2)); |
---|
395 | } |
---|
396 | return $string; |
---|
397 | } |
---|
398 | |
---|
399 | |
---|
400 | |
---|
401 | // UTF-16 => UTF-16BE |
---|
402 | public static function iconv_utf16_utf16be($string) { |
---|
403 | return getid3_iconv_replacement::iconv_utf8_utf16be(getid3_iconv_replacement::iconv_utf16_utf8($string)); |
---|
404 | } |
---|
405 | |
---|
406 | |
---|
407 | |
---|
408 | // UTF-16 => UTF-16LE |
---|
409 | public static function iconv_utf16_utf16le($string) { |
---|
410 | return getid3_iconv_replacement::iconv_utf8_utf16le(getid3_iconv_replacement::iconv_utf16_utf8($string)); |
---|
411 | } |
---|
412 | |
---|
413 | } |
---|
414 | |
---|
415 | ?> |
---|