59177427 |
1 | <?php |
1fd97780 |
2 | |
35586184 |
3 | /** |
4 | * i18n.php |
5 | * |
15e6162e |
6 | * Copyright (c) 1999-2002 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 | |
a2a7852b |
18 | /* Decodes a string to the internal encoding from the given charset */ |
19 | function charset_decode ($charset, $string) { |
6fbd125b |
20 | global $languages, $squirrelmail_language; |
a2a7852b |
21 | |
3714db45 |
22 | if (isset($languages[$squirrelmail_language]['XTRA_CODE']) && |
23 | function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) { |
6fbd125b |
24 | $string = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $string); |
25 | } |
b05c8961 |
26 | |
a2a7852b |
27 | /* All HTML special characters are 7 bit and can be replaced first */ |
28 | $string = htmlspecialchars ($string); |
29 | |
30 | $charset = strtolower($charset); |
31 | |
a2a7852b |
32 | if (ereg('iso-8859-([[:digit:]]+)', $charset, $res)) { |
33 | if ($res[1] == '1') { |
34 | $ret = charset_decode_iso_8859_1 ($string); |
35 | } else if ($res[1] == '2') { |
36 | $ret = charset_decode_iso_8859_2 ($string); |
9be313d5 |
37 | } else if ($res[1] == '4') { |
38 | $ret = charset_decode_iso_8859_4 ($string); |
a2a7852b |
39 | } else if ($res[1] == '7') { |
40 | $ret = charset_decode_iso_8859_7 ($string); |
9be313d5 |
41 | } else if ($res[1] == '13') { |
42 | $ret = charset_decode_iso_8859_13 ($string); |
a2a7852b |
43 | } else if ($res[1] == '15') { |
44 | $ret = charset_decode_iso_8859_15 ($string); |
45 | } else { |
46 | $ret = charset_decode_iso_8859_default ($string); |
47 | } |
48 | } else if ($charset == 'ns_4551-1') { |
49 | $ret = charset_decode_ns_4551_1 ($string); |
50 | } else if ($charset == 'koi8-r') { |
51 | $ret = charset_decode_koi8r ($string); |
52 | } else if ($charset == 'windows-1251') { |
53 | $ret = charset_decode_koi8r ($string); |
c37a12f8 |
54 | } else if ($charset == 'windows-1257') { |
55 | $ret = charset_decode_windows_1257 ($string); |
a2a7852b |
56 | } else { |
57 | $ret = $string; |
58 | } |
59 | return( $ret ); |
60 | } |
61 | |
62 | /* |
63 | iso-8859-1 is the same as Latin 1 and is normally used |
64 | in western europe. |
65 | */ |
66 | function charset_decode_iso_8859_1 ($string) { |
67 | global $default_charset; |
13e0c649 |
68 | |
a2a7852b |
69 | if (strtolower($default_charset) <> 'iso-8859-1') { |
70 | /* Only do the slow convert if there are 8-bit characters */ |
71 | if (ereg("[\200-\377]", $string)) { |
066c374f |
72 | $string = str_replace("\201", '', $string); |
73 | $string = str_replace("\202", '‚', $string); |
74 | $string = str_replace("\203", 'ƒ', $string); |
75 | $string = str_replace("\204", '„', $string); |
76 | $string = str_replace("\205", '…', $string); |
77 | $string = str_replace("\206", '†', $string); |
78 | $string = str_replace("\207", '‡', $string); |
79 | $string = str_replace("\210", 'ˆ', $string); |
80 | $string = str_replace("\211", '‰', $string); |
81 | $string = str_replace("\212", 'Š', $string); |
82 | $string = str_replace("\213", '‹', $string); |
83 | $string = str_replace("\214", 'Œ', $string); |
84 | $string = str_replace("\215", '', $string); |
85 | $string = str_replace("\216", 'Ž', $string); |
86 | $string = str_replace("\217", '', $string); |
87 | $string = str_replace("\220", '', $string); |
88 | $string = str_replace("\221", '‘', $string); |
89 | $string = str_replace("\222", '’', $string); |
90 | $string = str_replace("\223", '“', $string); |
91 | $string = str_replace("\224", '”', $string); |
92 | $string = str_replace("\225", '•', $string); |
93 | $string = str_replace("\226", '–', $string); |
94 | $string = str_replace("\227", '—', $string); |
95 | $string = str_replace("\230", '˜', $string); |
96 | $string = str_replace("\231", '™', $string); |
97 | $string = str_replace("\232", 'š', $string); |
98 | $string = str_replace("\233", '›', $string); |
99 | $string = str_replace("\234", 'œ', $string); |
100 | $string = str_replace("\235", '', $string); |
101 | $string = str_replace("\236", 'ž', $string); |
102 | $string = str_replace("\237", 'Ÿ', $string); |
103 | $string = str_replace("\240", ' ', $string); |
104 | $string = str_replace("\241", '¡', $string); |
105 | $string = str_replace("\242", '¢', $string); |
106 | $string = str_replace("\243", '£', $string); |
107 | $string = str_replace("\244", '¤', $string); |
108 | $string = str_replace("\245", '¥', $string); |
109 | $string = str_replace("\246", '¦', $string); |
110 | $string = str_replace("\247", '§', $string); |
111 | $string = str_replace("\250", '¨', $string); |
112 | $string = str_replace("\251", '©', $string); |
113 | $string = str_replace("\252", 'ª', $string); |
114 | $string = str_replace("\253", '«', $string); |
115 | $string = str_replace("\254", '¬', $string); |
116 | $string = str_replace("\255", '­', $string); |
117 | $string = str_replace("\256", '®', $string); |
118 | $string = str_replace("\257", '¯', $string); |
119 | $string = str_replace("\260", '°', $string); |
120 | $string = str_replace("\261", '±', $string); |
121 | $string = str_replace("\262", '²', $string); |
122 | $string = str_replace("\263", '³', $string); |
123 | $string = str_replace("\264", '´', $string); |
124 | $string = str_replace("\265", 'µ', $string); |
125 | $string = str_replace("\266", '¶', $string); |
126 | $string = str_replace("\267", '·', $string); |
127 | $string = str_replace("\270", '¸', $string); |
128 | $string = str_replace("\271", '¹', $string); |
129 | $string = str_replace("\272", 'º', $string); |
130 | $string = str_replace("\273", '»', $string); |
131 | $string = str_replace("\274", '¼', $string); |
132 | $string = str_replace("\275", '½', $string); |
133 | $string = str_replace("\276", '¾', $string); |
134 | $string = str_replace("\277", '¿', $string); |
135 | $string = str_replace("\300", 'À', $string); |
136 | $string = str_replace("\301", 'Á', $string); |
137 | $string = str_replace("\302", 'Â', $string); |
138 | $string = str_replace("\303", 'Ã', $string); |
139 | $string = str_replace("\304", 'Ä', $string); |
140 | $string = str_replace("\305", 'Å', $string); |
141 | $string = str_replace("\306", 'Æ', $string); |
142 | $string = str_replace("\307", 'Ç', $string); |
143 | $string = str_replace("\310", 'È', $string); |
144 | $string = str_replace("\311", 'É', $string); |
145 | $string = str_replace("\312", 'Ê', $string); |
146 | $string = str_replace("\313", 'Ë', $string); |
147 | $string = str_replace("\314", 'Ì', $string); |
148 | $string = str_replace("\315", 'Í', $string); |
149 | $string = str_replace("\316", 'Î', $string); |
150 | $string = str_replace("\317", 'Ï', $string); |
151 | $string = str_replace("\320", 'Ð', $string); |
152 | $string = str_replace("\321", 'Ñ', $string); |
153 | $string = str_replace("\322", 'Ò', $string); |
154 | $string = str_replace("\323", 'Ó', $string); |
155 | $string = str_replace("\324", 'Ô', $string); |
156 | $string = str_replace("\325", 'Õ', $string); |
157 | $string = str_replace("\326", 'Ö', $string); |
158 | $string = str_replace("\327", '×', $string); |
159 | $string = str_replace("\330", 'Ø', $string); |
160 | $string = str_replace("\331", 'Ù', $string); |
161 | $string = str_replace("\332", 'Ú', $string); |
162 | $string = str_replace("\333", 'Û', $string); |
163 | $string = str_replace("\334", 'Ü', $string); |
164 | $string = str_replace("\335", 'Ý', $string); |
165 | $string = str_replace("\336", 'Þ', $string); |
166 | $string = str_replace("\337", 'ß', $string); |
167 | $string = str_replace("\340", 'à', $string); |
168 | $string = str_replace("\341", 'á', $string); |
169 | $string = str_replace("\342", 'â', $string); |
170 | $string = str_replace("\343", 'ã', $string); |
171 | $string = str_replace("\344", 'ä', $string); |
172 | $string = str_replace("\345", 'å', $string); |
173 | $string = str_replace("\346", 'æ', $string); |
174 | $string = str_replace("\347", 'ç', $string); |
175 | $string = str_replace("\350", 'è', $string); |
176 | $string = str_replace("\351", 'é', $string); |
177 | $string = str_replace("\352", 'ê', $string); |
178 | $string = str_replace("\353", 'ë', $string); |
179 | $string = str_replace("\354", 'ì', $string); |
180 | $string = str_replace("\355", 'í', $string); |
181 | $string = str_replace("\356", 'î', $string); |
182 | $string = str_replace("\357", 'ï', $string); |
183 | $string = str_replace("\360", 'ð', $string); |
184 | $string = str_replace("\361", 'ñ', $string); |
185 | $string = str_replace("\362", 'ò', $string); |
186 | $string = str_replace("\363", 'ó', $string); |
187 | $string = str_replace("\364", 'ô', $string); |
188 | $string = str_replace("\365", 'õ', $string); |
189 | $string = str_replace("\366", 'ö', $string); |
190 | $string = str_replace("\367", '÷', $string); |
191 | $string = str_replace("\370", 'ø', $string); |
192 | $string = str_replace("\371", 'ù', $string); |
193 | $string = str_replace("\372", 'ú', $string); |
194 | $string = str_replace("\373", 'û', $string); |
195 | $string = str_replace("\374", 'ü', $string); |
196 | $string = str_replace("\375", 'ý', $string); |
197 | $string = str_replace("\376", 'þ', $string); |
198 | $string = str_replace("\377", 'ÿ', $string); |
a2a7852b |
199 | } |
200 | } |
201 | |
202 | return ($string); |
203 | } |
204 | |
205 | /* iso-8859-2 is used for some eastern European languages */ |
206 | function charset_decode_iso_8859_2 ($string) { |
207 | global $default_charset; |
208 | |
209 | if (strtolower($default_charset) == 'iso-8859-2') |
210 | return $string; |
211 | |
212 | /* Only do the slow convert if there are 8-bit characters */ |
213 | if (! ereg("[\200-\377]", $string)) |
214 | return $string; |
215 | |
216 | /* NO-BREAK SPACE */ |
217 | $string = str_replace("\240", ' ', $string); |
218 | /* LATIN CAPITAL LETTER A WITH OGONEK */ |
219 | $string = str_replace("\241", 'Ą', $string); |
220 | /* BREVE */ |
221 | $string = str_replace("\242", '˘', $string); |
222 | // LATIN CAPITAL LETTER L WITH STROKE |
223 | $string = str_replace("\243", 'Ł', $string); |
224 | // CURRENCY SIGN |
225 | $string = str_replace("\244", '¤', $string); |
226 | // LATIN CAPITAL LETTER L WITH CARON |
227 | $string = str_replace("\245", 'Ľ', $string); |
228 | // LATIN CAPITAL LETTER S WITH ACUTE |
229 | $string = str_replace("\246", 'Ś', $string); |
230 | // SECTION SIGN |
231 | $string = str_replace("\247", '§', $string); |
232 | // DIAERESIS |
233 | $string = str_replace("\250", '¨', $string); |
234 | // LATIN CAPITAL LETTER S WITH CARON |
235 | $string = str_replace("\251", 'Š', $string); |
236 | // LATIN CAPITAL LETTER S WITH CEDILLA |
237 | $string = str_replace("\252", 'Ş', $string); |
238 | // LATIN CAPITAL LETTER T WITH CARON |
239 | $string = str_replace("\253", 'Ť', $string); |
240 | // LATIN CAPITAL LETTER Z WITH ACUTE |
241 | $string = str_replace("\254", 'Ź', $string); |
242 | // SOFT HYPHEN |
243 | $string = str_replace("\255", '­', $string); |
244 | // LATIN CAPITAL LETTER Z WITH CARON |
245 | $string = str_replace("\256", 'Ž', $string); |
246 | // LATIN CAPITAL LETTER Z WITH DOT ABOVE |
247 | $string = str_replace("\257", 'Ż', $string); |
248 | // DEGREE SIGN |
249 | $string = str_replace("\260", '°', $string); |
250 | // LATIN SMALL LETTER A WITH OGONEK |
251 | $string = str_replace("\261", 'ą', $string); |
252 | // OGONEK |
253 | $string = str_replace("\262", '˛', $string); |
254 | // LATIN SMALL LETTER L WITH STROKE |
255 | $string = str_replace("\263", 'ł', $string); |
256 | // ACUTE ACCENT |
257 | $string = str_replace("\264", '´', $string); |
258 | // LATIN SMALL LETTER L WITH CARON |
259 | $string = str_replace("\265", 'ľ', $string); |
260 | // LATIN SMALL LETTER S WITH ACUTE |
261 | $string = str_replace("\266", 'ś', $string); |
262 | // CARON |
263 | $string = str_replace("\267", 'ˇ', $string); |
264 | // CEDILLA |
265 | $string = str_replace("\270", '¸', $string); |
266 | // LATIN SMALL LETTER S WITH CARON |
267 | $string = str_replace("\271", 'š', $string); |
268 | // LATIN SMALL LETTER S WITH CEDILLA |
269 | $string = str_replace("\272", 'ş', $string); |
270 | // LATIN SMALL LETTER T WITH CARON |
271 | $string = str_replace("\273", 'ť', $string); |
272 | // LATIN SMALL LETTER Z WITH ACUTE |
273 | $string = str_replace("\274", 'ź', $string); |
274 | // DOUBLE ACUTE ACCENT |
275 | $string = str_replace("\275", '˝', $string); |
276 | // LATIN SMALL LETTER Z WITH CARON |
277 | $string = str_replace("\276", 'ž', $string); |
278 | // LATIN SMALL LETTER Z WITH DOT ABOVE |
279 | $string = str_replace("\277", 'ż', $string); |
280 | // LATIN CAPITAL LETTER R WITH ACUTE |
281 | $string = str_replace("\300", 'Ŕ', $string); |
282 | // LATIN CAPITAL LETTER A WITH ACUTE |
283 | $string = str_replace("\301", 'Á', $string); |
284 | // LATIN CAPITAL LETTER A WITH CIRCUMFLEX |
285 | $string = str_replace("\302", 'Â', $string); |
286 | // LATIN CAPITAL LETTER A WITH BREVE |
287 | $string = str_replace("\303", 'Ă', $string); |
288 | // LATIN CAPITAL LETTER A WITH DIAERESIS |
289 | $string = str_replace("\304", 'Ä', $string); |
290 | // LATIN CAPITAL LETTER L WITH ACUTE |
291 | $string = str_replace("\305", 'Ĺ', $string); |
292 | // LATIN CAPITAL LETTER C WITH ACUTE |
293 | $string = str_replace("\306", 'Ć', $string); |
294 | // LATIN CAPITAL LETTER C WITH CEDILLA |
295 | $string = str_replace("\307", 'Ç', $string); |
296 | // LATIN CAPITAL LETTER C WITH CARON |
297 | $string = str_replace("\310", 'Č', $string); |
298 | // LATIN CAPITAL LETTER E WITH ACUTE |
299 | $string = str_replace("\311", 'É', $string); |
300 | // LATIN CAPITAL LETTER E WITH OGONEK |
301 | $string = str_replace("\312", 'Ę', $string); |
302 | // LATIN CAPITAL LETTER E WITH DIAERESIS |
303 | $string = str_replace("\313", 'Ë', $string); |
304 | // LATIN CAPITAL LETTER E WITH CARON |
305 | $string = str_replace("\314", 'Ě', $string); |
306 | // LATIN CAPITAL LETTER I WITH ACUTE |
307 | $string = str_replace("\315", 'Í', $string); |
308 | // LATIN CAPITAL LETTER I WITH CIRCUMFLEX |
309 | $string = str_replace("\316", 'Î', $string); |
310 | // LATIN CAPITAL LETTER D WITH CARON |
311 | $string = str_replace("\317", 'Ď', $string); |
312 | // LATIN CAPITAL LETTER D WITH STROKE |
313 | $string = str_replace("\320", 'Đ', $string); |
314 | // LATIN CAPITAL LETTER N WITH ACUTE |
315 | $string = str_replace("\321", 'Ń', $string); |
316 | // LATIN CAPITAL LETTER N WITH CARON |
317 | $string = str_replace("\322", 'Ň', $string); |
318 | // LATIN CAPITAL LETTER O WITH ACUTE |
319 | $string = str_replace("\323", 'Ó', $string); |
320 | // LATIN CAPITAL LETTER O WITH CIRCUMFLEX |
321 | $string = str_replace("\324", 'Ô', $string); |
322 | // LATIN CAPITAL LETTER O WITH DOUBLE ACUTE |
323 | $string = str_replace("\325", 'Ő', $string); |
324 | // LATIN CAPITAL LETTER O WITH DIAERESIS |
325 | $string = str_replace("\326", 'Ö', $string); |
326 | // MULTIPLICATION SIGN |
327 | $string = str_replace("\327", '×', $string); |
328 | // LATIN CAPITAL LETTER R WITH CARON |
329 | $string = str_replace("\330", 'Ř', $string); |
330 | // LATIN CAPITAL LETTER U WITH RING ABOVE |
331 | $string = str_replace("\331", 'Ů', $string); |
332 | // LATIN CAPITAL LETTER U WITH ACUTE |
333 | $string = str_replace("\332", 'Ú', $string); |
334 | // LATIN CAPITAL LETTER U WITH DOUBLE ACUTE |
335 | $string = str_replace("\333", 'Ű', $string); |
336 | // LATIN CAPITAL LETTER U WITH DIAERESIS |
337 | $string = str_replace("\334", 'Ü', $string); |
338 | // LATIN CAPITAL LETTER Y WITH ACUTE |
339 | $string = str_replace("\335", 'Ý', $string); |
340 | // LATIN CAPITAL LETTER T WITH CEDILLA |
341 | $string = str_replace("\336", 'Ţ', $string); |
342 | // LATIN SMALL LETTER SHARP S |
343 | $string = str_replace("\337", 'ß', $string); |
344 | // LATIN SMALL LETTER R WITH ACUTE |
345 | $string = str_replace("\340", 'ŕ', $string); |
346 | // LATIN SMALL LETTER A WITH ACUTE |
347 | $string = str_replace("\341", 'á', $string); |
348 | // LATIN SMALL LETTER A WITH CIRCUMFLEX |
349 | $string = str_replace("\342", 'â', $string); |
350 | // LATIN SMALL LETTER A WITH BREVE |
351 | $string = str_replace("\343", 'ă', $string); |
352 | // LATIN SMALL LETTER A WITH DIAERESIS |
353 | $string = str_replace("\344", 'ä', $string); |
354 | // LATIN SMALL LETTER L WITH ACUTE |
355 | $string = str_replace("\345", 'ĺ', $string); |
356 | // LATIN SMALL LETTER C WITH ACUTE |
357 | $string = str_replace("\346", 'ć', $string); |
358 | // LATIN SMALL LETTER C WITH CEDILLA |
359 | $string = str_replace("\347", 'ç', $string); |
360 | // LATIN SMALL LETTER C WITH CARON |
361 | $string = str_replace("\350", 'č', $string); |
362 | // LATIN SMALL LETTER E WITH ACUTE |
363 | $string = str_replace("\351", 'é', $string); |
364 | // LATIN SMALL LETTER E WITH OGONEK |
365 | $string = str_replace("\352", 'ę', $string); |
366 | // LATIN SMALL LETTER E WITH DIAERESIS |
367 | $string = str_replace("\353", 'ë', $string); |
368 | // LATIN SMALL LETTER E WITH CARON |
369 | $string = str_replace("\354", 'ě', $string); |
370 | // LATIN SMALL LETTER I WITH ACUTE |
371 | $string = str_replace("\355", 'í', $string); |
372 | // LATIN SMALL LETTER I WITH CIRCUMFLEX |
373 | $string = str_replace("\356", 'î', $string); |
374 | // LATIN SMALL LETTER D WITH CARON |
375 | $string = str_replace("\357", 'ď', $string); |
376 | // LATIN SMALL LETTER D WITH STROKE |
377 | $string = str_replace("\360", 'đ', $string); |
378 | // LATIN SMALL LETTER N WITH ACUTE |
379 | $string = str_replace("\361", 'ń', $string); |
380 | // LATIN SMALL LETTER N WITH CARON |
381 | $string = str_replace("\362", 'ň', $string); |
382 | // LATIN SMALL LETTER O WITH ACUTE |
383 | $string = str_replace("\363", 'ó', $string); |
384 | // LATIN SMALL LETTER O WITH CIRCUMFLEX |
385 | $string = str_replace("\364", 'ô', $string); |
386 | // LATIN SMALL LETTER O WITH DOUBLE ACUTE |
387 | $string = str_replace("\365", 'ő', $string); |
388 | // LATIN SMALL LETTER O WITH DIAERESIS |
389 | $string = str_replace("\366", 'ö', $string); |
390 | // DIVISION SIGN |
391 | $string = str_replace("\367", '÷', $string); |
392 | // LATIN SMALL LETTER R WITH CARON |
393 | $string = str_replace("\370", 'ř', $string); |
394 | // LATIN SMALL LETTER U WITH RING ABOVE |
395 | $string = str_replace("\371", 'ů', $string); |
396 | // LATIN SMALL LETTER U WITH ACUTE |
397 | $string = str_replace("\372", 'ú', $string); |
398 | // LATIN SMALL LETTER U WITH DOUBLE ACUTE |
399 | $string = str_replace("\373", 'ű', $string); |
400 | // LATIN SMALL LETTER U WITH DIAERESIS |
401 | $string = str_replace("\374", 'ü', $string); |
402 | // LATIN SMALL LETTER Y WITH ACUTE |
403 | $string = str_replace("\375", 'ý', $string); |
404 | // LATIN SMALL LETTER T WITH CEDILLA |
405 | $string = str_replace("\376", 'ţ', $string); |
406 | // DOT ABOVE |
407 | $string = str_replace("\377", '˙', $string); |
408 | |
409 | return $string; |
410 | } |
411 | |
9be313d5 |
412 | /* |
c37a12f8 |
413 | ISO/IEC 8859-4:1998 Latin Alphabet No. 4 |
9be313d5 |
414 | */ |
415 | |
416 | function charset_decode_iso_8859_4 ($string) { |
c37a12f8 |
417 | global $default_charset, $languages, $sm_notAlias; |
9be313d5 |
418 | |
c37a12f8 |
419 | if (strtolower($default_charset) == 'iso-8859-4') |
420 | return $string; |
421 | if (strtolower($languages[$sm_notAlias]['CHARSET']) == 'iso-8859-4') |
422 | return $string; |
423 | |
424 | /* Only do the slow convert if there are 8-bit characters */ |
425 | if (! ereg("[\200-\377]", $string)) |
426 | return $string; |
427 | |
428 | // latin capital letter a with ogonek 161 -> 260 |
429 | $string = str_replace ("\241", 'Ą', $string); |
430 | // latin small letter kra 162 -> 312 |
431 | $string = str_replace ("\242", 'ĸ', $string); |
432 | // latin capital letter r with cedilla 163 -> 342 |
433 | $string = str_replace ("\243", 'Ŗ', $string); |
434 | // latin capital letter i with tilde 165 -> 296 |
435 | $string = str_replace ("\245", 'Ĩ', $string); |
436 | // latin capital letter l with cedilla 166 -> 315 |
437 | $string = str_replace ("\246", 'Ļ', $string); |
438 | // latin capital letter s with caron 169 -> 352 |
439 | $string = str_replace ("\251", 'Š', $string); |
440 | // latin capital letter e with macron 170 -> 274 |
441 | $string = str_replace ("\252", 'Ē', $string); |
442 | // latin capital letter g with cedilla 171 -> 290 |
443 | $string = str_replace ("\253", 'Ģ', $string); |
444 | // latin capital letter t with stroke 172 -> 358 |
445 | $string = str_replace ("\254", 'Ŧ', $string); |
446 | // latin capital letter z with caron 174 -> 381 |
447 | $string = str_replace ("\256", 'Ž', $string); |
448 | // latin small letter a with ogonek 177 -> 261 |
449 | $string = str_replace ("\261", 'ą', $string); |
450 | // ogonek 178 -> 731 |
451 | $string = str_replace ("\262", '˛', $string); |
452 | // latin small letter r with cedilla 179 -> 343 |
453 | $string = str_replace ("\263", 'ŗ', $string); |
454 | // latin small letter i with tilde 181 -> 297 |
455 | $string = str_replace ("\265", 'ĩ', $string); |
456 | // latin small letter l with cedilla 182 -> 316 |
457 | $string = str_replace ("\266", 'ļ', $string); |
458 | // caron 183 -> 711 |
459 | $string = str_replace ("\267", 'ˇ', $string); |
460 | // latin small letter s with caron 185 -> 353 |
461 | $string = str_replace ("\271", 'š', $string); |
462 | // latin small letter e with macron 186 -> 275 |
463 | $string = str_replace ("\272", 'ē', $string); |
464 | // latin small letter g with cedilla 187 -> 291 |
465 | $string = str_replace ("\273", 'ģ', $string); |
466 | // latin small letter t with stroke 188 -> 359 |
467 | $string = str_replace ("\274", 'ŧ', $string); |
468 | // latin capital letter eng 189 -> 330 |
469 | $string = str_replace ("\275", 'Ŋ', $string); |
470 | // latin small letter z with caron 190 -> 382 |
471 | $string = str_replace ("\276", 'ž', $string); |
472 | // latin small letter eng 191 -> 331 |
473 | $string = str_replace ("\277", 'ŋ', $string); |
474 | // latin capital letter a with macron 192 -> 256 |
475 | $string = str_replace ("\300", 'Ā', $string); |
476 | // latin capital letter i with ogonek 199 -> 302 |
477 | $string = str_replace ("\307", 'Į', $string); |
478 | // latin capital letter c with caron 200 -> 268 |
479 | $string = str_replace ("\310", 'Č', $string); |
480 | // latin capital letter e with ogonek 202 -> 280 |
481 | $string = str_replace ("\312", 'Ę', $string); |
482 | // latin capital letter e with dot above 204 -> 278 |
483 | $string = str_replace ("\314", 'Ė', $string); |
484 | // latin capital letter i with macron 207 -> 298 |
485 | $string = str_replace ("\317", 'Ī', $string); |
486 | // latin capital letter d with stroke 208 -> 272 |
487 | $string = str_replace ("\320", 'Đ', $string); |
488 | // latin capital letter n with cedilla 209 -> 325 |
489 | $string = str_replace ("\321", 'Ņ', $string); |
490 | // latin capital letter o with macron 210 -> 332 |
491 | $string = str_replace ("\322", 'Ō', $string); |
492 | // latin capital letter k with cedilla 211 -> 310 |
493 | $string = str_replace ("\323", 'Ķ', $string); |
494 | // latin capital letter u with ogonek 217 -> 370 |
495 | $string = str_replace ("\331", 'Ų', $string); |
496 | // latin capital letter u with tilde 221 -> 360 |
497 | $string = str_replace ("\335", 'Ũ', $string); |
498 | // latin capital letter u with macron 222 -> 362 |
499 | $string = str_replace ("\336", 'Ū', $string); |
500 | // latin small letter a with macron 224 -> 257 |
501 | $string = str_replace ("\340", 'ā', $string); |
502 | // latin small letter i with ogonek 231 -> 303 |
503 | $string = str_replace ("\347", 'į', $string); |
504 | // latin small letter c with caron 232 -> 269 |
505 | $string = str_replace ("\350", 'č', $string); |
506 | // latin small letter e with ogonek 234 -> 281 |
507 | $string = str_replace ("\352", 'ę', $string); |
508 | // latin small letter e with dot above 236 -> 279 |
509 | $string = str_replace ("\354", 'ė', $string); |
510 | // latin small letter i with macron 239 -> 299 |
511 | $string = str_replace ("\357", 'ī', $string); |
512 | // latin small letter d with stroke 240 -> 273 |
513 | $string = str_replace ("\360", 'đ', $string); |
514 | // latin small letter n with cedilla 241 -> 326 |
515 | $string = str_replace ("\361", 'ņ', $string); |
516 | // latin small letter o with macron 242 -> 333 |
517 | $string = str_replace ("\362", 'ō', $string); |
518 | // latin small letter k with cedilla 243 -> 311 |
519 | $string = str_replace ("\363", 'ķ', $string); |
520 | // latin small letter u with ogonek 249 -> 371 |
521 | $string = str_replace ("\371", 'ų', $string); |
522 | // latin small letter u with tilde 253 -> 361 |
523 | $string = str_replace ("\375", 'ũ', $string); |
524 | // latin small letter u with macron 254 -> 363 |
525 | $string = str_replace ("\376", 'ū', $string); |
526 | // dot above 255 -> 729 |
527 | $string = str_replace ("\377", '˙', $string); |
528 | |
529 | // rest of charset is the same as ISO-8859-1 |
9be313d5 |
530 | return (charset_decode_iso_8859_1($string)); |
531 | } |
532 | |
a2a7852b |
533 | /* iso-8859-7 is Greek. */ |
534 | function charset_decode_iso_8859_7 ($string) { |
535 | global $default_charset; |
536 | |
537 | if (strtolower($default_charset) == 'iso-8859-7') { |
538 | return $string; |
539 | } |
540 | |
541 | /* Only do the slow convert if there are 8-bit characters */ |
542 | if (!ereg("[\200-\377]", $string)) { |
543 | return $string; |
544 | } |
545 | |
546 | /* Some diverse characters in the beginning */ |
547 | $string = str_replace("\240", ' ', $string); |
548 | $string = str_replace("\241", '‘', $string); |
549 | $string = str_replace("\242", '’', $string); |
550 | $string = str_replace("\243", '£', $string); |
551 | $string = str_replace("\246", '¦', $string); |
552 | $string = str_replace("\247", '§', $string); |
553 | $string = str_replace("\250", '¨', $string); |
554 | $string = str_replace("\251", '©', $string); |
555 | $string = str_replace("\253", '«', $string); |
556 | $string = str_replace("\254", '¬', $string); |
557 | $string = str_replace("\255", '­', $string); |
558 | $string = str_replace("\257", '―', $string); |
559 | $string = str_replace("\260", '°', $string); |
560 | $string = str_replace("\261", '±', $string); |
561 | $string = str_replace("\262", '²', $string); |
562 | $string = str_replace("\263", '³', $string); |
563 | |
564 | /* Horizontal bar (parentheki pavla) */ |
565 | $string = str_replace ("\257", '―', $string); |
566 | |
567 | /* |
568 | * ISO-8859-7 characters from 11/04 (0xB4) to 11/06 (0xB6) |
569 | * These are Unicode 900-902 |
570 | */ |
b85a4575 |
571 | $string = preg_replace("/([\264-\266])/e","'&#' . (ord('\\1')+720);",$string); |
49c17806 |
572 | |
a2a7852b |
573 | /* 11/07 (0xB7) Middle dot is the same in iso-8859-1 */ |
574 | $string = str_replace("\267", '·', $string); |
575 | |
576 | /* |
577 | * ISO-8859-7 characters from 11/08 (0xB8) to 11/10 (0xBA) |
578 | * These are Unicode 900-902 |
579 | */ |
b85a4575 |
580 | $string = preg_replace("/([\270-\272])/e","'&#' . (ord('\\1')+720);",$string); |
a2a7852b |
581 | |
582 | /* |
583 | * 11/11 (0xBB) Right angle quotation mark is the same as in |
584 | * iso-8859-1 |
585 | */ |
586 | $string = str_replace("\273", '»', $string); |
587 | |
588 | /* And now the rest of the charset */ |
b85a4575 |
589 | $string = preg_replace("/([\274-\376])/e","'&#'.(ord('\\1')+720);",$string); |
a2a7852b |
590 | |
591 | return $string; |
592 | } |
593 | |
9be313d5 |
594 | /* |
c37a12f8 |
595 | ISO/IEC 8859-13:1998 Latin Alphabet No. 7 (Baltic Rim) |
9be313d5 |
596 | */ |
9be313d5 |
597 | function charset_decode_iso_8859_13 ($string) { |
c37a12f8 |
598 | global $default_charset, $languages, $sm_notAlias; |
599 | |
600 | if (strtolower($default_charset) == 'iso-8859-13') |
601 | return $string; |
602 | if (strtolower($languages[$sm_notAlias]['CHARSET']) == 'iso-8859-13') |
603 | return $string; |
604 | |
605 | /* Only do the slow convert if there are 8-bit characters */ |
606 | if (! ereg("[\200-\377]", $string)) |
607 | return $string; |
9be313d5 |
608 | |
c37a12f8 |
609 | // right double quotation mark 161 -> 8221 |
610 | $string = str_replace ("\241", '”', $string); |
611 | // double low-9 quotation mark 165 -> 8222 |
612 | $string = str_replace ("\245", '„', $string); |
613 | // latin capital letter o with stroke 168 -> 216 |
614 | $string = str_replace ("\250", 'Ø', $string); |
615 | // latin capital letter r with cedilla 170 -> 342 |
616 | $string = str_replace ("\252", 'Ŗ', $string); |
617 | // latin capital letter ae 175 -> 198 |
618 | $string = str_replace ("\257", 'Æ', $string); |
619 | // left double quotation mark 180 -> 8220 |
620 | $string = str_replace ("\264", '“', $string); |
621 | // latin small letter o with stroke 184 -> 248 |
622 | $string = str_replace ("\270", 'ø', $string); |
623 | // latin small letter r with cedilla 186 -> 343 |
624 | $string = str_replace ("\272", 'ŗ', $string); |
625 | // latin small letter ae 191 -> 230 |
626 | $string = str_replace ("\277", 'æ', $string); |
627 | // latin capital letter a with ogonek 192 -> 260 |
628 | $string = str_replace ("\300", 'Ą', $string); |
629 | // latin capital letter i with ogonek 193 -> 302 |
630 | $string = str_replace ("\301", 'Į', $string); |
631 | // latin capital letter a with macron 194 -> 256 |
632 | $string = str_replace ("\302", 'Ā', $string); |
633 | // latin capital letter c with acute 195 -> 262 |
634 | $string = str_replace ("\303", 'Ć', $string); |
635 | // latin capital letter e with ogonek 198 -> 280 |
636 | $string = str_replace ("\306", 'Ę', $string); |
637 | // latin capital letter e with macron 199 -> 274 |
638 | $string = str_replace ("\307", 'Ē', $string); |
639 | // latin capital letter c with caron 200 -> 268 |
640 | $string = str_replace ("\310", 'Č', $string); |
641 | // latin capital letter z with acute 202 -> 377 |
642 | $string = str_replace ("\312", 'Ź', $string); |
643 | // latin capital letter e with dot above 203 -> 278 |
644 | $string = str_replace ("\313", 'Ė', $string); |
645 | // latin capital letter g with cedilla 204 -> 290 |
646 | $string = str_replace ("\314", 'Ģ', $string); |
647 | // latin capital letter k with cedilla 205 -> 310 |
648 | $string = str_replace ("\315", 'Ķ', $string); |
649 | // latin capital letter i with macron 206 -> 298 |
650 | $string = str_replace ("\316", 'Ī', $string); |
651 | // latin capital letter l with cedilla 207 -> 315 |
652 | $string = str_replace ("\317", 'Ļ', $string); |
653 | // latin capital letter s with caron 208 -> 352 |
654 | $string = str_replace ("\320", 'Š', $string); |
655 | // latin capital letter n with acute 209 -> 323 |
656 | $string = str_replace ("\321", 'Ń', $string); |
657 | // latin capital letter n with cedilla 210 -> 325 |
658 | $string = str_replace ("\322", 'Ņ', $string); |
659 | // latin capital letter o with macron 212 -> 332 |
660 | $string = str_replace ("\324", 'Ō', $string); |
661 | // latin capital letter u with ogonek 216 -> 370 |
662 | $string = str_replace ("\330", 'Ų', $string); |
663 | // latin capital letter l with stroke 217 -> 321 |
664 | $string = str_replace ("\331", 'Ł', $string); |
665 | // latin capital letter s with acute 218 -> 346 |
666 | $string = str_replace ("\332", 'Ś', $string); |
667 | // latin capital letter u with macron 219 -> 362 |
668 | $string = str_replace ("\333", 'Ū', $string); |
669 | // latin capital letter z with dot above 221 -> 379 |
670 | $string = str_replace ("\335", 'Ż', $string); |
671 | // latin capital letter z with caron 222 -> 381 |
672 | $string = str_replace ("\336", 'Ž', $string); |
673 | // latin small letter a with ogonek 224 -> 261 |
674 | $string = str_replace ("\340", 'ą', $string); |
675 | // latin small letter i with ogonek 225 -> 303 |
676 | $string = str_replace ("\341", 'į', $string); |
677 | // latin small letter a with macron 226 -> 257 |
678 | $string = str_replace ("\342", 'ā', $string); |
679 | // latin small letter c with acute 227 -> 263 |
680 | $string = str_replace ("\343", 'ć', $string); |
681 | // latin small letter e with ogonek 230 -> 281 |
682 | $string = str_replace ("\346", 'ę', $string); |
683 | // latin small letter e with macron 231 -> 275 |
684 | $string = str_replace ("\347", 'ē', $string); |
685 | // latin small letter c with caron 232 -> 269 |
686 | $string = str_replace ("\350", 'č', $string); |
687 | // latin small letter z with acute 234 -> 378 |
688 | $string = str_replace ("\352", 'ź', $string); |
689 | // latin small letter e with dot above 235 -> 279 |
690 | $string = str_replace ("\353", 'ė', $string); |
691 | // latin small letter g with cedilla 236 -> 291 |
692 | $string = str_replace ("\354", 'ģ', $string); |
693 | // latin small letter k with cedilla 237 -> 311 |
694 | $string = str_replace ("\355", 'ķ', $string); |
695 | // latin small letter i with macron 238 -> 299 |
696 | $string = str_replace ("\356", 'ī', $string); |
697 | // latin small letter l with cedilla 239 -> 316 |
698 | $string = str_replace ("\357", 'ļ', $string); |
699 | // latin small letter s with caron 240 -> 253 |
700 | $string = str_replace ("\360", 'š', $string); |
701 | // latin small letter n with acute 241 -> 324 |
702 | $string = str_replace ("\361", 'ń', $string); |
703 | // latin small letter n with cedilla 242 -> 326 |
704 | $string = str_replace ("\362", 'ņ', $string); |
705 | // latin small letter o with macron 244 -> 333 |
706 | $string = str_replace ("\364", 'ō', $string); |
707 | // latin small letter u with ogonek 248 -> 371 |
708 | $string = str_replace ("\370", 'ų', $string); |
709 | // latin small letter l with stroke 249 -> 322 |
710 | $string = str_replace ("\371", 'ł', $string); |
711 | // latin small letter s with acute 250 -> 347 |
712 | $string = str_replace ("\372", 'ś', $string); |
713 | // latin small letter u with macron 251 -> 363 |
714 | $string = str_replace ("\373", 'ū', $string); |
715 | // latin small letter z with dot above 253 -> 380 |
716 | $string = str_replace ("\375", 'ż', $string); |
717 | // latin small letter z with caron 254 -> 382 |
718 | $string = str_replace ("\376", 'ž', $string); |
719 | // right single quotation mark 255 -> 8217 |
720 | $string = str_replace ("\377", '’', $string); |
721 | |
722 | // rest of charset is the same as ISO-8859-1 |
9be313d5 |
723 | return (charset_decode_iso_8859_1($string)); |
724 | } |
725 | |
a2a7852b |
726 | /* |
727 | * iso-8859-15 is Latin 9 and has very much the same use as Latin 1 |
728 | * but has the Euro symbol and some characters needed for French. |
729 | */ |
730 | function charset_decode_iso_8859_15 ($string) { |
731 | // Euro sign |
732 | $string = str_replace ("\244", '€', $string); |
733 | // Latin capital letter S with caron |
84556805 |
734 | $string = str_replace ("\246", 'Š', $string); |
a2a7852b |
735 | // Latin small letter s with caron |
736 | $string = str_replace ("\250", 'š', $string); |
737 | // Latin capital letter Z with caron |
738 | $string = str_replace ("\264", 'Ž', $string); |
739 | // Latin small letter z with caron |
740 | $string = str_replace ("\270", 'ž', $string); |
741 | // Latin capital ligature OE |
742 | $string = str_replace ("\274", 'Œ', $string); |
743 | // Latin small ligature oe |
744 | $string = str_replace ("\275", 'œ', $string); |
745 | // Latin capital letter Y with diaeresis |
746 | $string = str_replace ("\276", 'Ÿ', $string); |
747 | |
748 | return (charset_decode_iso_8859_1($string)); |
749 | } |
750 | |
84556805 |
751 | /* ISO-8859-5 is Cyrillic */ |
a2a7852b |
752 | function charset_decode_iso_8859_5 ($string) { |
753 | // Convert to KOI8-R, then return this decoded. |
754 | $string = convert_cyr_string($string, 'i', 'k'); |
755 | return charset_decode_koi8r($string); |
756 | } |
757 | |
758 | /* Remove all 8 bit characters from all other ISO-8859 character sets */ |
759 | function charset_decode_iso_8859_default ($string) { |
760 | return (strtr($string, "\240\241\242\243\244\245\246\247". |
1fd97780 |
761 | "\250\251\252\253\254\255\256\257". |
762 | "\260\261\262\263\264\265\266\267". |
763 | "\270\271\272\273\274\275\276\277". |
764 | "\300\301\302\303\304\305\306\307". |
765 | "\310\311\312\313\314\315\316\317". |
766 | "\320\321\322\323\324\325\326\327". |
767 | "\330\331\332\333\334\335\336\337". |
768 | "\340\341\342\343\344\345\346\347". |
769 | "\350\351\352\353\354\355\356\357". |
770 | "\360\361\362\363\364\365\366\367". |
a2a7852b |
771 | "\370\371\372\373\374\375\376\377", |
1fd97780 |
772 | "????????????????????????????????????????". |
773 | "????????????????????????????????????????". |
774 | "????????????????????????????????????????". |
775 | "????????")); |
a2a7852b |
776 | |
777 | } |
778 | |
779 | /* |
780 | * This is the same as ISO-646-NO and is used by some |
781 | * Microsoft programs when sending Norwegian characters |
782 | */ |
783 | function charset_decode_ns_4551_1 ($string) { |
784 | /* |
785 | * These characters are: |
786 | * Latin capital letter AE |
787 | * Latin capital letter O with stroke |
788 | * Latin capital letter A with ring above |
789 | * and the same as small letters |
790 | */ |
791 |