59177427 |
1 | <?php |
1fd97780 |
2 | |
35586184 |
3 | /** |
4 | * i18n.php |
5 | * |
76911253 |
6 | * Copyright (c) 1999-2003 The SquirrelMail Project Team |
35586184 |
7 | * Licensed under the GNU GPL. For full terms see the file COPYING. |
8 | * |
9 | * This file contains variuos functions that are needed to do |
10 | * internationalization of SquirrelMail. |
11 | * |
12 | * Internally the output character set is used. Other characters are |
13 | * encoded using Unicode entities according to HTML 4.0. |
14 | * |
15 | * $Id$ |
16 | */ |
17 | |
961ca3d8 |
18 | require_once(SM_PATH . 'functions/global.php'); |
19 | |
a2a7852b |
20 | /* Decodes a string to the internal encoding from the given charset */ |
21 | function charset_decode ($charset, $string) { |
6fbd125b |
22 | global $languages, $squirrelmail_language; |
a2a7852b |
23 | |
3714db45 |
24 | if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && |
25 | function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) { |
6fbd125b |
26 | $string = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $string); |
27 | } |
b05c8961 |
28 | |
a2a7852b |
29 | /* All HTML special characters are 7 bit and can be replaced first */ |
cef054e4 |
30 | |
098ea084 |
31 | $string = htmlspecialchars ($string); |
a2a7852b |
32 | |
33 | $charset = strtolower($charset); |
34 | |
94965562 |
35 | set_my_charset() ; |
36 | |
a2a7852b |
37 | if (ereg('iso-8859-([[:digit:]]+)', $charset, $res)) { |
38 | if ($res[1] == '1') { |
39 | $ret = charset_decode_iso_8859_1 ($string); |
40 | } else if ($res[1] == '2') { |
41 | $ret = charset_decode_iso_8859_2 ($string); |
9be313d5 |
42 | } else if ($res[1] == '4') { |
43 | $ret = charset_decode_iso_8859_4 ($string); |
94965562 |
44 | } else if ($res[1] == '5') { |
45 | $ret = charset_decode_iso_8859_5 ($string); |
a2a7852b |
46 | } else if ($res[1] == '7') { |
47 | $ret = charset_decode_iso_8859_7 ($string); |
3ab35042 |
48 | } else if ($res[1] == '9') { |
49 | $ret = charset_decode_iso_8859_9 ($string); |
9be313d5 |
50 | } else if ($res[1] == '13') { |
51 | $ret = charset_decode_iso_8859_13 ($string); |
a2a7852b |
52 | } else if ($res[1] == '15') { |
53 | $ret = charset_decode_iso_8859_15 ($string); |
54 | } else { |
55 | $ret = charset_decode_iso_8859_default ($string); |
56 | } |
57 | } else if ($charset == 'ns_4551-1') { |
58 | $ret = charset_decode_ns_4551_1 ($string); |
59 | } else if ($charset == 'koi8-r') { |
60 | $ret = charset_decode_koi8r ($string); |
1c0e847f |
61 | } else if ($charset == 'koi8-u') { |
62 | $ret = charset_decode_koi8u ($string); |
a2a7852b |
63 | } else if ($charset == 'windows-1251') { |
ecd877a8 |
64 | $ret = charset_decode_windows_1251 ($string); |
3ab35042 |
65 | } else if ($charset == 'windows-1253') { |
66 | $ret = charset_decode_windows_1253 ($string); |
67 | } else if ($charset == 'windows-1254') { |
68 | $ret = charset_decode_windows_1254 ($string); |
c48a8ca7 |
69 | } else if ($charset == 'windows-1255') { |
70 | $ret = charset_decode_windows_1255 ($string); |
71 | } else if ($charset == 'windows-1256') { |
72 | $ret = charset_decode_windows_1256 ($string); |
c37a12f8 |
73 | } else if ($charset == 'windows-1257') { |
74 | $ret = charset_decode_windows_1257 ($string); |
3ab35042 |
75 | } else if ($charset == 'utf-8') { |
76 | $ret = charset_decode_utf8 ($string); |
a2a7852b |
77 | } else { |
78 | $ret = $string; |
79 | } |
80 | return( $ret ); |
81 | } |
82 | |
83 | /* |
84 | iso-8859-1 is the same as Latin 1 and is normally used |
85 | in western europe. |
86 | */ |
87 | function charset_decode_iso_8859_1 ($string) { |
88 | global $default_charset; |
13e0c649 |
89 | |
a2a7852b |
90 | if (strtolower($default_charset) <> 'iso-8859-1') { |
91 | /* Only do the slow convert if there are 8-bit characters */ |
92 | if (ereg("[\200-\377]", $string)) { |
066c374f |
93 | $string = str_replace("\201", '', $string); |
94 | $string = str_replace("\202", '‚', $string); |
95 | $string = str_replace("\203", 'ƒ', $string); |
96 | $string = str_replace("\204", '„', $string); |
97 | $string = str_replace("\205", '…', $string); |
98 | $string = str_replace("\206", '†', $string); |
99 | $string = str_replace("\207", '‡', $string); |
100 | $string = str_replace("\210", 'ˆ', $string); |
101 | $string = str_replace("\211", '‰', $string); |
102 | $string = str_replace("\212", 'Š', $string); |
103 | $string = str_replace("\213", '‹', $string); |
104 | $string = str_replace("\214", 'Œ', $string); |
105 | $string = str_replace("\215", '', $string); |
106 | $string = str_replace("\216", 'Ž', $string); |
107 | $string = str_replace("\217", '', $string); |
108 | $string = str_replace("\220", '', $string); |
109 | $string = str_replace("\221", '‘', $string); |
110 | $string = str_replace("\222", '’', $string); |
111 | $string = str_replace("\223", '“', $string); |
112 | $string = str_replace("\224", '”', $string); |
113 | $string = str_replace("\225", '•', $string); |
114 | $string = str_replace("\226", '–', $string); |
115 | $string = str_replace("\227", '—', $string); |
116 | $string = str_replace("\230", '˜', $string); |
117 | $string = str_replace("\231", '™', $string); |
118 | $string = str_replace("\232", 'š', $string); |
119 | $string = str_replace("\233", '›', $string); |
120 | $string = str_replace("\234", 'œ', $string); |
121 | $string = str_replace("\235", '', $string); |
122 | $string = str_replace("\236", 'ž', $string); |
123 | $string = str_replace("\237", 'Ÿ', $string); |
124 | $string = str_replace("\240", ' ', $string); |
125 | $string = str_replace("\241", '¡', $string); |
126 | $string = str_replace("\242", '¢', $string); |
127 | $string = str_replace("\243", '£', $string); |
128 | $string = str_replace("\244", '¤', $string); |
129 | $string = str_replace("\245", '¥', $string); |
130 | $string = str_replace("\246", '¦', $string); |
131 | $string = str_replace("\247", '§', $string); |
132 | $string = str_replace("\250", '¨', $string); |
133 | $string = str_replace("\251", '©', $string); |
134 | $string = str_replace("\252", 'ª', $string); |
135 | $string = str_replace("\253", '«', $string); |
136 | $string = str_replace("\254", '¬', $string); |
137 | $string = str_replace("\255", '­', $string); |
138 | $string = str_replace("\256", '®', $string); |
139 | $string = str_replace("\257", '¯', $string); |
140 | $string = str_replace("\260", '°', $string); |
141 | $string = str_replace("\261", '±', $string); |
142 | $string = str_replace("\262", '²', $string); |
143 | $string = str_replace("\263", '³', $string); |
144 | $string = str_replace("\264", '´', $string); |
145 | $string = str_replace("\265", 'µ', $string); |
146 | $string = str_replace("\266", '¶', $string); |
147 | $string = str_replace("\267", '·', $string); |
148 | $string = str_replace("\270", '¸', $string); |
149 | $string = str_replace("\271", '¹', $string); |
150 | $string = str_replace("\272", 'º', $string); |
151 | $string = str_replace("\273", '»', $string); |
152 | $string = str_replace("\274", '¼', $string); |
153 | $string = str_replace("\275", '½', $string); |
154 | $string = str_replace("\276", '¾', $string); |
155 | $string = str_replace("\277", '¿', $string); |
156 | $string = str_replace("\300", 'À', $string); |
157 | $string = str_replace("\301", 'Á', $string); |
158 | $string = str_replace("\302", 'Â', $string); |
159 | $string = str_replace("\303", 'Ã', $string); |
160 | $string = str_replace("\304", 'Ä', $string); |
161 | $string = str_replace("\305", 'Å', $string); |
162 | $string = str_replace("\306", 'Æ', $string); |
163 | $string = str_replace("\307", 'Ç', $string); |
164 | $string = str_replace("\310", 'È', $string); |
165 | $string = str_replace("\311", 'É', $string); |
166 | $string = str_replace("\312", 'Ê', $string); |
167 | $string = str_replace("\313", 'Ë', $string); |
168 | $string = str_replace("\314", 'Ì', $string); |
169 | $string = str_replace("\315", 'Í', $string); |
170 | $string = str_replace("\316", 'Î', $string); |
171 | $string = str_replace("\317", 'Ï', $string); |
172 | $string = str_replace("\320", 'Ð', $string); |
173 | $string = str_replace("\321", 'Ñ', $string); |
174 | $string = str_replace("\322", 'Ò', $string); |
175 | $string = str_replace("\323", 'Ó', $string); |
176 | $string = str_replace("\324", 'Ô', $string); |
177 | $string = str_replace("\325", 'Õ', $string); |
178 | $string = str_replace("\326", 'Ö', $string); |
179 | $string = str_replace("\327", '×', $string); |
180 | $string = str_replace("\330", 'Ø', $string); |
181 | $string = str_replace("\331", 'Ù', $string); |
182 | $string = str_replace("\332", 'Ú', $string); |
183 | $string = str_replace("\333", 'Û', $string); |
184 | $string = str_replace("\334", 'Ü', $string); |
185 | $string = str_replace("\335", 'Ý', $string); |
186 | $string = str_replace("\336", 'Þ', $string); |
187 | $string = str_replace("\337", 'ß', $string); |
188 | $string = str_replace("\340", 'à', $string); |
189 | $string = str_replace("\341", 'á', $string); |
190 | $string = str_replace("\342", 'â', $string); |
191 | $string = str_replace("\343", 'ã', $string); |
192 | $string = str_replace("\344", 'ä', $string); |
193 | $string = str_replace("\345", 'å', $string); |
194 | $string = str_replace("\346", 'æ', $string); |
195 | $string = str_replace("\347", 'ç', $string); |
196 | $string = str_replace("\350", 'è', $string); |
197 | $string = str_replace("\351", 'é', $string); |
198 | $string = str_replace("\352", 'ê', $string); |
199 | $string = str_replace("\353", 'ë', $string); |
200 | $string = str_replace("\354", 'ì', $string); |
201 | $string = str_replace("\355", 'í', $string); |
202 | $string = str_replace("\356", 'î', $string); |
203 | $string = str_replace("\357", 'ï', $string); |
204 | $string = str_replace("\360", 'ð', $string); |
205 | $string = str_replace("\361", 'ñ', $string); |
206 | $string = str_replace("\362", 'ò', $string); |
207 | $string = str_replace("\363", 'ó', $string); |
208 | $string = str_replace("\364", 'ô', $string); |
209 | $string = str_replace("\365", 'õ', $string); |
210 | $string = str_replace("\366", 'ö', $string); |
211 | $string = str_replace("\367", '÷', $string); |
212 | $string = str_replace("\370", 'ø', $string); |
213 | $string = str_replace("\371", 'ù', $string); |
214 | $string = str_replace("\372", 'ú', $string); |
215 | $string = str_replace("\373", 'û', $string); |
216 | $string = str_replace("\374", 'ü', $string); |
217 | $string = str_replace("\375", 'ý', $string); |
218 | $string = str_replace("\376", 'þ', $string); |
219 | $string = str_replace("\377", 'ÿ', $string); |
a2a7852b |
220 | } |
221 | } |
222 | |
223 | return ($string); |
224 | } |
225 | |
226 | /* iso-8859-2 is used for some eastern European languages */ |
227 | function charset_decode_iso_8859_2 ($string) { |
228 | global $default_charset; |
229 | |
230 | if (strtolower($default_charset) == 'iso-8859-2') |
231 | return $string; |
232 | |
233 | /* Only do the slow convert if there are 8-bit characters */ |
234 | if (! ereg("[\200-\377]", $string)) |
235 | return $string; |
236 | |
237 | /* NO-BREAK SPACE */ |
238 | $string = str_replace("\240", ' ', $string); |
239 | /* LATIN CAPITAL LETTER A WITH OGONEK */ |
240 | $string = str_replace("\241", 'Ą', $string); |
241 | /* BREVE */ |
242 | $string = str_replace("\242", '˘', $string); |
243 | // LATIN CAPITAL LETTER L WITH STROKE |
244 | $string = str_replace("\243", 'Ł', $string); |
245 | // CURRENCY SIGN |
246 | $string = str_replace("\244", '¤', $string); |
247 | // LATIN CAPITAL LETTER L WITH CARON |
248 | $string = str_replace("\245", 'Ľ', $string); |
249 | // LATIN CAPITAL LETTER S WITH ACUTE |
250 | $string = str_replace("\246", 'Ś', $string); |
251 | // SECTION SIGN |
252 | $string = str_replace("\247", '§', $string); |
253 | // DIAERESIS |
254 | $string = str_replace("\250", '¨', $string); |
255 | // LATIN CAPITAL LETTER S WITH CARON |
256 | $string = str_replace("\251", 'Š', $string); |
257 | // LATIN CAPITAL LETTER S WITH CEDILLA |
258 | $string = str_replace("\252", 'Ş', $string); |
259 | // LATIN CAPITAL LETTER T WITH CARON |
260 | $string = str_replace("\253", 'Ť', $string); |
261 | // LATIN CAPITAL LETTER Z WITH ACUTE |
262 | $string = str_replace("\254", 'Ź', $string); |
263 | // SOFT HYPHEN |
264 | $string = str_replace("\255", '­', $string); |
265 | // LATIN CAPITAL LETTER Z WITH CARON |
266 | $string = str_replace("\256", 'Ž', $string); |
267 | // LATIN CAPITAL LETTER Z WITH DOT ABOVE |
268 | $string = str_replace("\257", 'Ż', $string); |
269 | // DEGREE SIGN |
270 | $string = str_replace("\260", '°', $string); |
271 | // LATIN SMALL LETTER A WITH OGONEK |
272 | $string = str_replace("\261", 'ą', $string); |
273 | // OGONEK |
274 | $string = str_replace("\262", '˛', $string); |
275 | // LATIN SMALL LETTER L WITH STROKE |
276 | $string = str_replace("\263", 'ł', $string); |
277 | // ACUTE ACCENT |
278 | $string = str_replace("\264", '´', $string); |
279 | // LATIN SMALL LETTER L WITH CARON |
280 | $string = str_replace("\265", 'ľ', $string); |
281 | // LATIN SMALL LETTER S WITH ACUTE |
282 | $string = str_replace("\266", 'ś', $string); |
283 | // CARON |
284 | $string = str_replace("\267", 'ˇ', $string); |
285 | // CEDILLA |
286 | $string = str_replace("\270", '¸', $string); |
287 | // LATIN SMALL LETTER S WITH CARON |
288 | $string = str_replace("\271", 'š', $string); |
289 | // LATIN SMALL LETTER S WITH CEDILLA |
290 | $string = str_replace("\272", 'ş', $string); |
291 | // LATIN SMALL LETTER T WITH CARON |
292 | $string = str_replace("\273", 'ť', $string); |
293 | // LATIN SMALL LETTER Z WITH ACUTE |
294 | $string = str_replace("\274", 'ź', $string); |
295 | // DOUBLE ACUTE ACCENT |
296 | $string = str_replace("\275", '˝', $string); |
297 | // LATIN SMALL LETTER Z WITH CARON |
298 | $string = str_replace("\276", 'ž', $string); |
299 | // LATIN SMALL LETTER Z WITH DOT ABOVE |
300 | $string = str_replace("\277", 'ż', $string); |
301 | // LATIN CAPITAL LETTER R WITH ACUTE |
302 | $string = str_replace("\300", 'Ŕ', $string); |
303 | // LATIN CAPITAL LETTER A WITH ACUTE |
304 | $string = str_replace("\301", 'Á', $string); |
305 | // LATIN CAPITAL LETTER A WITH CIRCUMFLEX |
306 | $string = str_replace("\302", 'Â', $string); |
307 | // LATIN CAPITAL LETTER A WITH BREVE |
308 | $string = str_replace("\303", 'Ă', $string); |
309 | // LATIN CAPITAL LETTER A WITH DIAERESIS |
310 | $string = str_replace("\304", 'Ä', $string); |
311 | // LATIN CAPITAL LETTER L WITH ACUTE |
312 | $string = str_replace("\305", 'Ĺ', $string); |
313 | // LATIN CAPITAL LETTER C WITH ACUTE |
314 | $string = str_replace("\306", 'Ć', $string); |
315 | // LATIN CAPITAL LETTER C WITH CEDILLA |
316 | $string = str_replace("\307", 'Ç', $string); |
317 | // LATIN CAPITAL LETTER C WITH CARON |
318 | $string = str_replace("\310", 'Č', $string); |
319 | // LATIN CAPITAL LETTER E WITH ACUTE |
320 | $string = str_replace("\311", 'É', $string); |
321 | // LATIN CAPITAL LETTER E WITH OGONEK |
322 | $string = str_replace("\312", 'Ę', $string); |
323 | // LATIN CAPITAL LETTER E WITH DIAERESIS |
324 | $string = str_replace("\313", 'Ë', $string); |
325 | // LATIN CAPITAL LETTER E WITH CARON |
326 | $string = str_replace("\314", 'Ě', $string); |
327 | // LATIN CAPITAL LETTER I WITH ACUTE |
328 | $string = str_replace("\315", 'Í', $string); |
329 | // LATIN CAPITAL LETTER I WITH CIRCUMFLEX |
330 | $string = str_replace("\316", 'Î', $string); |
331 | // LATIN CAPITAL LETTER D WITH CARON |
332 | $string = str_replace("\317", 'Ď', $string); |
333 | // LATIN CAPITAL LETTER D WITH STROKE |
334 | $string = str_replace("\320", 'Đ', $string); |
335 | // LATIN CAPITAL LETTER N WITH ACUTE |
336 | $string = str_replace("\321", 'Ń', $string); |
337 | // LATIN CAPITAL LETTER N WITH CARON |
338 | $string = str_replace("\322", 'Ň', $string); |
339 | // LATIN CAPITAL LETTER O WITH ACUTE |
340 | $string = str_replace("\323", 'Ó', $string); |
341 | // LATIN CAPITAL LETTER O WITH CIRCUMFLEX |
342 | $string = str_replace("\324", 'Ô', $string); |
343 | // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE |
344 | $string = str_replace("\325", 'Ő', $string); |
345 | // LATIN CAPITAL LETTER O WITH DIAERESIS |
346 | $string = str_replace("\326", 'Ö', $string); |
347 | // MULTIPLICATION SIGN |
348 | $string = str_replace("\327", '×', $string); |
349 | // LATIN CAPITAL LETTER R WITH CARON |
350 | $string = str_replace("\330", 'Ř', $string); |
351 | // LATIN CAPITAL LETTER U WITH RING ABOVE |
352 | $string = str_replace("\331", 'Ů', $string); |
353 | // LATIN CAPITAL LETTER U WITH ACUTE |
354 | $string = str_replace("\332", 'Ú', $string); |
355 | // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE |
356 | $string = str_replace("\333", 'Ű', $string); |
357 | // LATIN CAPITAL LETTER U WITH DIAERESIS |
358 | $string = str_replace("\334", 'Ü', $string); |
359 | // LATIN CAPITAL LETTER Y WITH ACUTE |
360 | $string = str_replace("\335", 'Ý', $string); |
361 | // LATIN CAPITAL LETTER T WITH CEDILLA |
362 | $string = str_replace("\336", 'Ţ', $string); |
363 | // LATIN SMALL LETTER SHARP S |
364 | $string = str_replace("\337", 'ß', $string); |
365 | // LATIN SMALL LETTER R WITH ACUTE |
366 | $string = str_replace("\340", 'ŕ', $string); |
367 | // LATIN SMALL LETTER A WITH ACUTE |
368 | $string = str_replace("\341", 'á', $string); |
369 | // LATIN SMALL LETTER A WITH CIRCUMFLEX |
370 | $string = str_replace("\342", 'â', $string); |
371 | // LATIN SMALL LETTER A WITH BREVE |
372 | $string = str_replace("\343", 'ă', $string); |
373 | // LATIN SMALL LETTER A WITH DIAERESIS |
374 | $string = str_replace("\344", 'ä', $string); |
375 | // LATIN SMALL LETTER L WITH ACUTE |
376 | $string = str_replace("\345", 'ĺ', $string); |
377 | // LATIN SMALL LETTER C WITH ACUTE |
378 | $string = str_replace("\346", 'ć', $string); |
379 | // LATIN SMALL LETTER C WITH CEDILLA |
380 | $string = str_replace("\347", 'ç', $string); |
381 | // LATIN SMALL LETTER C WITH CARON |
382 | $string = str_replace("\350", 'č', $string); |
383 | // LATIN SMALL LETTER E WITH ACUTE |
384 | $string = str_replace("\351", 'é', $string); |
385 | // LATIN SMALL LETTER E WITH OGONEK |
386 | $string = str_replace("\352", 'ę', $string); |
387 | // LATIN SMALL LETTER E WITH DIAERESIS |
388 | $string = str_replace("\353", 'ë', $string); |
389 | // LATIN SMALL LETTER E WITH CARON |
390 | $string = str_replace("\354", 'ě', $string); |
391 | // LATIN SMALL LETTER I WITH ACUTE |
392 | $string = str_replace("\355", 'í', $string); |
393 | // LATIN SMALL LETTER I WITH CIRCUMFLEX |
394 | $string = str_replace("\356", 'î', $string); |
395 | // LATIN SMALL LETTER D WITH CARON |
396 | $string = str_replace("\357", 'ď', $string); |
397 | // LATIN SMALL LETTER D WITH STROKE |
398 | $string = str_replace("\360", 'đ', $string); |
399 | // LATIN SMALL LETTER N WITH ACUTE |
400 | $string = str_replace("\361", 'ń', $string); |
401 | // LATIN SMALL LETTER N WITH CARON |
402 | $string = str_replace("\362", 'ň', $string); |
403 | // LATIN SMALL LETTER O WITH ACUTE |
404 | $string = str_replace("\363", 'ó', $string); |
405 | // LATIN SMALL LETTER O WITH CIRCUMFLEX |
406 | $string = str_replace("\364", 'ô', $string); |
407 | // LATIN SMALL LETTER O WITH DOUBLE ACUTE |
408 | $string = str_replace("\365", 'ő', $string); |
409 | // LATIN SMALL LETTER O WITH DIAERESIS |
410 | $string = str_replace("\366", 'ö', $string); |
411 | // DIVISION SIGN |
412 | $string = str_replace("\367", '÷', $string); |
413 | // LATIN SMALL LETTER R WITH CARON |
414 | $string = str_replace("\370", 'ř', $string); |
415 | // LATIN SMALL LETTER U WITH RING ABOVE |
416 | $string = str_replace("\371", 'ů', $string); |
417 | // LATIN SMALL LETTER U WITH ACUTE |
418 | $string = str_replace("\372", 'ú', $string); |
419 | // LATIN SMALL LETTER U WITH DOUBLE ACUTE |
420 | $string = str_replace("\373", 'ű', $string); |
421 | // LATIN SMALL LETTER U WITH DIAERESIS |
422 | $string = str_replace("\374", 'ü', $string); |
423 | // LATIN SMALL LETTER Y WITH ACUTE |
424 | $string = str_replace("\375", 'ý', $string); |
425 | // LATIN SMALL LETTER T WITH CEDILLA |
426 | $string = str_replace("\376", 'ţ', $string); |
427 | // DOT ABOVE |
428 | $string = str_replace("\377", '˙', $string); |
429 | |
430 | return $string; |
431 | } |
432 | |
9be313d5 |
433 | /* |
c37a12f8 |
434 | ISO/IEC 8859-4:1998 Latin Alphabet No. 4 |
9be313d5 |
435 | */ |
436 | |
437 | function charset_decode_iso_8859_4 ($string) { |
94965562 |
438 | global $default_charset; |
9be313d5 |
439 | |
c37a12f8 |
440 | if (strtolower($default_charset) == 'iso-8859-4') |
441 | return $string; |
c37a12f8 |
442 | |
443 | /* Only do the slow convert if there are 8-bit characters */ |
444 | if (! ereg("[\200-\377]", $string)) |
445 | return $string; |
446 | |
c37a12f8 |
447 | $string = str_replace ("\241", 'Ą', $string); |
c37a12f8 |
448 | $string = str_replace ("\242", 'ĸ', $string); |
c37a12f8 |
449 | $string = str_replace ("\243", 'Ŗ', $string); |
c37a12f8 |
450 | $string = str_replace ("\245", 'Ĩ', $string); |
c37a12f8 |
451 | $string = str_replace ("\246", 'Ļ', $string); |
c37a12f8 |
452 | $string = str_replace ("\251", 'Š', $string); |
c37a12f8 |
453 | $string = str_replace ("\252", 'Ē', $string); |
c37a12f8 |
454 | $string = str_replace ("\253", 'Ģ', $string); |
c37a12f8 |
455 | $string = str_replace ("\254", 'Ŧ', $string); |
c37a12f8 |
456 | $string = str_replace ("\256", 'Ž', $string); |
c37a12f8 |
457 | $string = str_replace ("\261", 'ą', $string); |
c37a12f8 |
458 | $string = str_replace ("\262", '˛', $string); |
c37a12f8 |
459 | $string = str_replace ("\263", 'ŗ', $string); |
c37a12f8 |
460 | $string = str_replace ("\265", 'ĩ', $string); |
c37a12f8 |
461 | $string = str_replace ("\266", 'ļ', $string); |
c37a12f8 |
462 | $string = str_replace ("\267", 'ˇ', $string); |
c37a12f8 |
463 | $string = str_replace ("\271", 'š', $string); |
c37a12f8 |
464 | $string = str_replace ("\272", 'ē', $string); |
c37a12f8 |
465 | $string = str_replace ("\273", 'ģ', $string); |
c37a12f8 |
466 | $string = str_replace ("\274", 'ŧ', $string); |
c37a12f8 |
467 | $string = str_replace ("\275", 'Ŋ', $string); |
c37a12f8 |
468 | $string = str_replace ("\276", 'ž', $string); |
c37a12f8 |
469 | $string = str_replace ("\277", 'ŋ', $string); |
c37a12f8 |
470 | $string = str_replace ("\300", 'Ā', $string); |
c37a12f8 |
471 | $string = str_replace ("\307", 'Į', $string); |
c37a12f8 |
472 | $string = str_replace ("\310", 'Č', $string); |
c37a12f8 |
473 | $string = str_replace ("\312", 'Ę', $string); |
c37a12f8 |
474 | $string = str_replace ("\314", 'Ė', $string); |
c37a12f8 |
475 | $string = str_replace ("\317", 'Ī', $string); |
c37a12f8 |
476 | $string = str_replace ("\320", 'Đ', $string); |
c37a12f8 |
477 | $string = str_replace ("\321", 'Ņ', $string); |
c37a12f8 |
478 | $string = str_replace ("\322", 'Ō', $string); |
c37a12f8 |
479 | $string = str_replace ("\323", 'Ķ', $string); |
c37a12f8 |
480 | $string = str_replace ("\331", 'Ų', $string); |
c37a12f8 |
481 | $string = str_replace ("\335", 'Ũ', $string); |
c37a12f8 |
482 | $string = str_replace ("\336", 'Ū', $string); |
c37a12f8 |
483 | $string = str_replace ("\340", 'ā', $string); |
c37a12f8 |
484 | $string = str_replace ("\347", 'į', $string); |
c37a12f8 |
485 | $string = str_replace ("\350", 'č', $string); |
c37a12f8 |
486 | $string = str_replace ("\352", 'ę', $string); |
c37a12f8 |
487 | $string = str_replace ("\354", 'ė', $string); |
c37a12f8 |
488 | $string = str_replace ("\357", 'ī', $string); |
c37a12f8 |
489 | $string = str_replace ("\360", 'đ', $string); |
c37a12f8 |
490 | $string = str_replace ("\361", 'ņ', $string); |
c37a12f8 |
491 | $string = str_replace ("\362", 'ō', $string); |
c37a12f8 |
492 | $string = str_replace ("\363", 'ķ', $string); |
c37a12f8 |
493 | $string = str_replace ("\371", 'ų', $string); |
c37a12f8 |
494 | $string = str_replace ("\375", 'ũ', $string); |
c37a12f8 |
495 | $string = str_replace ("\376", 'ū', $string); |
c37a12f8 |
496 | $string = str_replace ("\377", '˙', $string); |
497 | |
498 | // rest of charset is the same as ISO-8859-1 |
9be313d5 |
499 | return (charset_decode_iso_8859_1($string)); |
500 | } |
501 | |
1c0e847f |
502 | /* ISO-8859-5 is Cyrillic */ |
503 | function charset_decode_iso_8859_5 ($string) { |
504 | global $default_charset; |
505 | |
506 | if (strtolower($default_charset) == 'iso-8859-5') { |
507 | return $string; |
508 | } |
509 | |
510 | /* Only do the slow convert if there are 8-bit characters */ |
511 | if (! ereg("[\200-\377]", $string)) |
512 | return $string; |
513 | |
514 | // NO-BREAK SPACE |
515 | $string = str_replace("\240", ' ', $string); |
516 | // 161-172 -> 1025-1036 (+864) |
517 | $string = preg_replace("/([\241-\254])/e","'&#' . (ord('\\1')+864) . ';'",$string); |
518 | // SOFT HYPHEN |
519 | $string = str_replace("\255", '­', $string); |
520 | // 174-239 -> 1038-1103 (+864) |
521 | $string = preg_replace("/([\256-\357])/e","'&#' . (ord('\\1')+864) . ';'",$string); |
522 | // NUMERO SIGN |
523 | $string = str_replace("\360", '№', $string); |
524 | // 241-252 -> 1105-1116 (+864) |
525 | $string = preg_replace("/([\361-\374])/e","'&#' . (ord('\\1')+864) . ';'",$string); |
526 | // SECTION SIGN |
527 | $string = str_replace("\375", '§', $string); |
528 | // CYRILLIC SMALL LETTER SHORT U (Byelorussian) |
529 | $string = str_replace("\376", 'ў', $string); |
530 | // CYRILLIC SMALL LETTER DZHE |
531 | $string = str_replace("\377", 'џ', $string); |
532 | |
533 | return $string; |
534 | } |
535 | |
a2a7852b |
536 | /* iso-8859-7 is Greek. */ |
537 | function charset_decode_iso_8859_7 ($string) { |
538 | global $default_charset; |
539 | |
540 | if (strtolower($default_charset) == 'iso-8859-7') { |
541 | return $string; |
542 | } |
543 | |
544 | /* Only do the slow convert if there are 8-bit characters */ |
545 | if (!ereg("[\200-\377]", $string)) { |
546 | return $string; |
547 | } |
548 | |
549 | /* Some diverse characters in the beginning */ |
550 | $string = str_replace("\240", ' ', $string); |
551 | $string = str_replace("\241", '‘', $string); |
552 | $string = str_replace("\242", '’', $string); |
553 | $string = str_replace("\243", '£', $string); |
554 | $string = str_replace("\246", '¦', $string); |
555 | $string = str_replace("\247", '§', $string); |
556 | $string = str_replace("\250", '¨', $string); |
557 | $string = str_replace("\251", '©', $string); |
558 | $string = str_replace("\253", '«', $string); |
559 | $string = str_replace("\254", '¬', $string); |
560 | $string = str_replace("\255", '­', $string); |
561 | $string = str_replace("\257", '―', $string); |
562 | $string = str_replace("\260", '°', $string); |
563 | $string = str_replace("\261", '±', $string); |
564 | $string = str_replace("\262", '²', $string); |
565 | $string = str_replace("\263", '³', $string); |
566 | |
567 | /* Horizontal bar (parentheki pavla) */ |
568 | $string = str_replace ("\257", '―', $string); |
569 | |
570 | /* |
571 | * ISO-8859-7 characters from 11/04 (0xB4) to 11/06 (0xB6) |
572 | * These are Unicode 900-902 |
573 | */ |
b85a4575 |
574 | $string = preg_replace("/([\264-\266])/e","'&#' . (ord('\\1')+720);",$string); |
49c17806 |
575 | |
a2a7852b |
576 | /* 11/07 (0xB7) Middle dot is the same in iso-8859-1 */ |
577 | $string = str_replace("\267", '·', $string); |
578 | |
579 | /* |
580 | * ISO-8859-7 characters from 11/08 (0xB8) to 11/10 (0xBA) |
581 | * These are Unicode 900-902 |
582 | */ |
b85a4575 |
583 | $string = preg_replace("/([\270-\272])/e","'&#' . (ord('\\1')+720);",$string); |
a2a7852b |
584 | |
585 | /* |
586 | * 11/11 (0xBB) Right angle quotation mark is the same as in |
587 | * iso-8859-1 |
588 | */ |
589 | $string = str_replace("\273", '»', $string); |
590 | |
591 | /* And now the rest of the charset */ |
b85a4575 |
592 | $string = preg_replace("/([\274-\376])/e","'&#'.(ord('\\1')+720);",$string); |
a2a7852b |
593 | |
594 | return $string; |
595 | } |
596 | |
3ab35042 |
597 | /* |
cd21d1aa |
598 | ISOIEC 8859-9:1999 Latin Alphabet No. 5 |
961ca3d8 |
599 | |
3ab35042 |
600 | */ |
601 | function charset_decode_iso_8859_9 ($string) { |
94965562 |
602 | global $default_charset; |
3ab35042 |
603 | |
604 | if (strtolower($default_charset) == 'iso-8859-9') |
605 | return $string; |
3ab35042 |
606 | |
607 | /* Only do the slow convert if there are 8-bit characters */ |
608 | if (! ereg("[\200-\377]", $string)) |
609 | return $string; |
610 | |
611 | // latin capital letter g with breve 208->286 |
612 | $string = str_replace("\320", 'Ğ', $string); |
613 | // latin capital letter i with dot above 221->304 |
614 | $string = str_replace("\335", 'İ', $string); |
615 | // latin capital letter s with cedilla 222->350 |
616 | $string = str_replace("\336", 'Ş', $string); |
617 | // latin small letter g with breve 240->287 |
618 | $string = str_replace("\360", 'ğ', $string); |
619 | // latin small letter dotless i 253->305 |
620 | $string = str_replace("\375", 'ı', $string); |
621 | // latin small letter s with cedilla 254->351 |
622 | $string = str_replace("\376", 'ş', $string); |
623 | |
624 | // rest of charset is the same as ISO-8859-1 |
625 | return (charset_decode_iso_8859_1($string)); |
626 | } |
627 | |
628 | |
9be313d5 |
629 | /* |
c37a12f8 |
630 | ISO/IEC 8859-13:1998 Latin Alphabet No. 7 (Baltic Rim) |
9be313d5 |
631 | */ |
9be313d5 |
632 | function charset_decode_iso_8859_13 ($string) { |
94965562 |
633 | global $default_charset; |
c37a12f8 |
634 | |
635 | if (strtolower($default_charset) == 'iso-8859-13') |
636 | return $string; |
c37a12f8 |
637 | |
638 | /* Only do the slow convert if there are 8-bit characters */ |
639 | if (! ereg("[\200-\377]", $string)) |
640 | return $string; |
9be313d5 |
641 | |
c37a12f8 |
642 | $string = str_replace ("\241", '”', $string); |
c37a12f8 |
643 | $string = str_replace ("\245", '„', $string); |
c37a12f8 |
644 | $string = str_replace ("\250", 'Ø', $string); |
c37a12f8 |
645 | $string = str_replace ("\252", 'Ŗ', $string); |
c37a12f8 |
646 | $string = str_replace ("\257", 'Æ', $string); |
c37a12f8 |
647 | $string = str_replace ("\264", '“', $string); |
c37a12f8 |
648 | $string = str_replace ("\270", 'ø', $string); |
c37a12f8 |
649 | $string = str_replace ("\272", 'ŗ', $string); |
c37a12f8 |
650 | $string = str_replace ("\277", 'æ', $string); |
c37a12f8 |
651 | $string = str_replace ("\300", 'Ą', $string); |
c37a12f8 |
652 | $string = str_replace ("\301", 'Į', $string); |
c37a12f8 |
653 | $string = str_replace ("\302", 'Ā', $string); |
c37a12f8 |
654 | $string = str_replace ("\303", 'Ć', $string); |
c37a12f8 |
655 | $string = str_replace ("\306", 'Ę', $string); |
c37a12f8 |
656 | $string = str_replace ("\307", 'Ē', $string); |
c37a12f8 |
657 | $string = str_replace ("\310", 'Č', $string); |
c37a12f8 |
658 | $string = str_replace ("\312", 'Ź', $string); |
c37a12f8 |
659 | $string = str_replace ("\313", 'Ė', $string); |
c37a12f8 |
660 | $string = str_replace ("\314", 'Ģ', $string); |
c37a12f8 |
661 | $string = str_replace ("\315", 'Ķ', $string); |
c37a12f8 |
662 | $string = str_replace ("\316", 'Ī', $string); |
c37a12f8 |
663 | $string = str_replace ("\317", 'Ļ', $string); |
c37a12f8 |
664 | $string = str_replace ("\320", 'Š', $string); |
c37a12f8 |
665 | $string = str_replace ("\321", 'Ń', $string); |
c37a12f8 |
666 | $string = str_replace ("\322", 'Ņ', $string); |
c37a12f8 |
667 | $string = str_replace ("\324", 'Ō', $string); |
c37a12f8 |
668 | $string = str_replace ("\330", 'Ų', $string); |
c37a12f8 |
669 | $string = str_replace ("\331", 'Ł', $string); |
c37a12f8 |
670 | $string = str_replace ("\332", 'Ś', $string); |
c37a12f8 |
671 | $string = str_replace ("\333", 'Ū', $string); |
c37a12f8 |
672 | $string = str_replace ("\335", 'Ż', $string); |
c37a12f8 |
673 | $string = str_replace ("\336", 'Ž', $string); |
c37a12f8 |
674 | $string = str_replace ("\340", 'ą', $string); |
c37a12f8 |
675 | $string = str_replace ("\341", 'į', $string); |
c37a12f8 |
676 | $string = str_replace ("\342", 'ā', $string); |
c37a12f8 |
677 | $string = str_replace ("\343", 'ć', $string); |
c37a12f8 |
678 | $string = str_replace ("\346", 'ę', $string); |
c37a12f8 |
679 | $string = str_replace ("\347", 'ē', $string); |
c37a12f8 |
680 | $string = str_replace ("\350", 'č', $string); |
c37a12f8 |
681 | $string = str_replace ("\352", 'ź', $string); |
c37a12f8 |
682 | $string = str_replace ("\353", 'ė', $string); |
c37a12f8 |
683 | $string = str_replace ("\354", 'ģ', $string); |
c37a12f8 |
684 | $string = str_replace ("\355", 'ķ', $string); |
c37a12f8 |
685 | $string = str_replace ("\356", 'ī', $string); |
c37a12f8 |
686 | $string = str_replace ("\357", 'ļ', $string); |
c37a12f8 |
687 | $string = str_replace ("\360", 'š', $string); |
c37a12f8 |
688 | $string = str_replace ("\361", 'ń', $string); |
c37a12f8 |
689 | $string = str_replace ("\362", 'ņ', $string); |
c37a12f8 |
690 | $string = str_replace ("\364", 'ō', $string); |
c37a12f8 |
691 | $string = str_replace ("\370", 'ų', $string); |
c37a12f8 |
692 | $string = str_replace ("\371", 'ł', $string); |
c37a12f8 |
693 | $string = str_replace ("\372", 'ś', $string); |
c37a12f8 |
694 | $string = str_replace ("\373", 'ū', $string); |
c37a12f8 |
695 | $string = str_replace ("\375", 'ż', $string); |
c37a12f8 |
696 | $string = str_replace ("\376", 'ž', $string); |
c37a12f8 |
697 | $string = str_replace ("\377", '’', $string); |
698 | |
699 | // rest of charset is the same as ISO-8859-1 |
9be313d5 |
700 | return (charset_decode_iso_8859_1($string)); |
701 | } |
702 | |
a2a7852b |
703 | /* |
704 | * iso-8859-15 is Latin 9 and has very much the same use as Latin 1 |
705 | * but has the Euro symbol and some characters needed for French. |
706 | */ |
707 | function charset_decode_iso_8859_15 ($string) { |
708 | // Euro sign |
709 | $string = str_replace ("\244", '€', $string); |
710 | // Latin capital letter S with caron |
84556805 |
711 | $string = str_replace ("\246", 'Š', $string); |
a2a7852b |
712 | // Latin small letter s with caron |
713 | $string = str_replace ("\250", 'š', $string); |
714 | // Latin capital letter Z with caron |
715 | $string = str_replace ("\264", 'Ž', $string); |
716 | // Latin small letter z with caron |
717 | $string = str_replace ("\270", 'ž', $string); |
718 | // Latin capital ligature OE |
719 | $string = str_replace ("\274", 'Œ', $string); |
720 | // Latin small ligature oe |
721 | $string = str_replace ("\275", 'œ', $string); |
722 | // Latin capital letter Y with diaeresis |
723 | $string = str_replace ("\276", 'Ÿ', $string); |
724 | |
725 | return (charset_decode_iso_8859_1($string)); |
726 | } |
727 | |
a2a7852b |
728 | |
729 | /* Remove all 8 bit characters from all other ISO-8859 character sets */ |
730 | function charset_decode_iso_8859_default ($string) { |
731 | return (strtr($string, "\240\241\242\243\244\245\246\247". |
1fd97780 |
732 | "\250\251\252\253\254\255\256\257". |
733 | "\260\261\262\263\264\265\266\267". |
734 | "\270\271\272\273\274\275\276\277". |
735 | "\300\301\302\303\304\305\306\307". |
736 | "\310\311\312\313\314\315\316\317". |
737 | "\320\321\322\323\324\325\326\327". |
738 | "\330\331\332\333\334\335\336\337". |
739 | "\340\341\342\343\344\345\346\347". |
740 | "\350\351\352\353\354\355\356\357". |
741 | "\360\361\362\363\364\365\366\367". |
a2a7852b |
742 | "\370\371\372\373\374\375\376\377", |
1fd97780 |
743 | "????????????????????????????????????????". |
744 | "????????????????????????????????????????". |
745 | "????????????????????????????????????????". |
746 | "????????")); |
a2a7852b |
747 | |
748 | } |
749 | |
750 | /* |
751 | * This is the same as ISO-646-NO and is used by some |
752 | * Microsoft programs when sending Norwegian characters |
753 | */ |
754 | function charset_decode_ns_4551_1 ($string) { |
755 | /* |
756 | * These characters are: |
757 | * Latin capital letter AE |
758 | * Latin capital letter O with stroke |
759 | * Latin capital letter A with ring above |
760 | * and the same as small letters |
761 | */ |
762 |