projects / squirrelmail.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
We're living in 2004 now... perl is your friend for these kinds of things :)
[squirrelmail.git] / functions / i18n.php
CommitLineData
59177427 1<?php
1fd97780 2
35586184 3/**
2ba706ef 4 * functions/i18n.php
35586184 5 *
2ba706ef 6 * Copyright (c) 1999-2004 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 * $Id$
d6c32258 16 * @package squirrelmail
35586184 17 */
18
d6c32258 19/** Everything uses global.php... */
961ca3d8 20require_once(SM_PATH . 'functions/global.php');
21
d6c32258 22/**
51468260 23 * Converts string from given charset to charset, that can be displayed by user translation.
24 *
25 * Function by default returns html encoded strings, if translation uses different encoding.
26 * If Japanese translation is used - function returns string converted to euc-jp
27 * If iconv or recode functions are enabled and translation uses utf-8 - function returns utf-8 encoded string.
28 * If $charset is not supported - function returns unconverted string.
d6c32258 29 *
51468260 30 * sanitizing of html tags is also done by this function.
31 *
d6c32258 32 * @param string $charset
33 * @param string $string Text to be decoded
51468260 34 * @return string decoded string
d6c32258 35 */
a2a7852b 36function charset_decode ($charset, $string) {
3ec81e63 37 global $languages, $squirrelmail_language, $default_charset;
edf2c0ba 38 global $use_php_recode, $use_php_iconv, $agresive_decoding;
a2a7852b 39
3714db45 40 if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
41 function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) {
6fbd125b 42 $string = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $string);
43 }
b05c8961 44
3ec81e63 45 $charset = strtolower($charset);
46
47 set_my_charset();
48
49 // Variables that allow to use functions without function_exist() calls
edf2c0ba 50 if (! isset($use_php_recode) || $use_php_recode=="" ) {
51 $use_php_recode=false; }
52 if (! isset($use_php_iconv) || $use_php_iconv=="" ) {
53 $use_php_iconv=false; }
3ec81e63 54
55 // Don't do conversion if charset is the same.
56 if ( $charset == strtolower($default_charset) )
57 return htmlspecialchars($string);
58
59 // catch iso-8859-8-i thing
60 if ( $charset == "iso-8859-8-i" )
61 $charset = "iso-8859-8";
62
63 /*
64 * Recode converts html special characters automatically if you use
65 * 'charset..html' decoding. There is no documented way to put -d option
66 * into php recode function call.
67 */
68 if ( $use_php_recode ) {
69 if ( $default_charset == "utf-8" ) {
70 // other charsets can be converted to utf-8 without loss.
71 // and output string is smaller
72 $string = recode_string($charset . "..utf-8",$string);
73 return htmlspecialchars($string);
74 } else {
75 $string = recode_string($charset . "..html",$string);
76 // recode does not convert single quote, htmlspecialchars does.
77 $string = str_replace("'", '&#039;', $string);
78 return $string;
79 }
80 }
81
82 // iconv functions does not have html target and can be used only with utf-8
83 if ( $use_php_iconv && $default_charset=='utf-8') {
84 $string = iconv($charset,$default_charset,$string);
85 return htmlspecialchars($string);
86 }
87
88 // If we don't use recode and iconv, we'll do it old way.
89
a2a7852b 90 /* All HTML special characters are 7 bit and can be replaced first */
cef054e4 91
098ea084 92 $string = htmlspecialchars ($string);
a2a7852b 93
5dd23dac 94 /* controls cpu and memory intensive decoding cycles */
edf2c0ba 95 if (! isset($agresive_decoding) || $agresive_decoding=="" ) {
96 $agresive_decoding=false; }
5dd23dac 97
b142de74 98 $decode=fixcharset($charset);
99 $decodefile=SM_PATH . 'functions/decode/' . $decode . '.php';
100 if (file_exists($decodefile)) {
101 include_once($decodefile);
102 $ret = call_user_func('charset_decode_'.$decode, $string);
a2a7852b 103 } else {
b142de74 104 $ret = $string;
a2a7852b 105 }
106 return( $ret );
107}
b142de74 108
109/**
110 * Makes charset name suitable for decoding cycles
111 *
112 * @param string $charset Name of charset
113 * @return string $charset Adjusted name of charset
114 */
115function fixcharset($charset) {
116 // minus removed from function names
117 $charset=str_replace('-','_',$charset);
118
119 // windows-125x and cp125x charsets
120 $charset=str_replace('windows_','cp',$charset);
a2a7852b 121
b142de74 122 // ibm > cp
123 $charset=str_replace('ibm','cp',$charset);
124
125 // iso-8859-8-i -> iso-8859-8
126 // use same cycle until I'll find differences
127 $charset=str_replace('iso_8859_8_i','iso_8859_8',$charset);
128
129 return $charset;
130}
a2a7852b 131
51468260 132/**
a2a7852b 133 * Set up the language to be output
134 * if $do_search is true, then scan the browser information
135 * for a possible language that we know
51468260 136 *
137 * Function sets system locale environment (LC_ALL, LANG, LANGUAGE),
138 * gettext translation bindings and html header information.
139 *
5679405c 140 * Function returns error codes, if there is some fatal error.
51468260 141 * 0 = no error,
142 * 1 = mbstring support is not present,
143 * 2 = mbstring support is not present, user's translation reverted to en_US.
144 *
145 * @param string $sm_language translation used by user's interface
146 * @param bool $do_search use browser's preferred language detection functions. Defaults to false.
147 * @param bool $default set $sm_language to $squirrelmail_default_language if language detection fails or language is not set. Defaults to false.
148 * @return int function execution error codes.
a2a7852b 149 */
67a8c90a 150function set_up_language($sm_language, $do_search = false, $default = false) {
a2a7852b 151
152 static $SetupAlready = 0;
9eb0fbd4 153 global $use_gettext, $languages,
a2a7852b 154 $squirrelmail_language, $squirrelmail_default_language,
51468260 155 $sm_notAlias, $username, $data_dir;
a2a7852b 156
157 if ($SetupAlready) {
158 return;
159 }
a65846a7 160
5c920668 161 $SetupAlready = TRUE;
961ca3d8 162 sqgetGlobalVar('HTTP_ACCEPT_LANGUAGE', $accept_lang, SQ_SERVER);
a2a7852b 163
961ca3d8 164 if ($do_search && ! $sm_language && isset($accept_lang)) {
165 $sm_language = substr($accept_lang, 0, 2);
a2a7852b 166 }
66d7950f 167
67a8c90a 168 if ((!$sm_language||$default) && isset($squirrelmail_default_language)) {
a2a7852b 169 $squirrelmail_language = $squirrelmail_default_language;
66d7950f 170 $sm_language = $squirrelmail_default_language;
a2a7852b 171 }
172 $sm_notAlias = $sm_language;
3ec81e63 173
174 // Catching removed translation
175 // System reverts to English translation if user prefs contain translation
2ba706ef 176 // that is not available in $languages array
3ec81e63 177 if (!isset($languages[$sm_notAlias])) {
178 $sm_notAlias="en_US";
179 }
180
a2a7852b 181 while (isset($languages[$sm_notAlias]['ALIAS'])) {
182 $sm_notAlias = $languages[$sm_notAlias]['ALIAS'];
183 }
184
88cb1b4d 185 if ( isset($sm_language) &&
5c920668 186 $use_gettext &&
187 $sm_language != '' &&
188 isset($languages[$sm_notAlias]['CHARSET']) ) {
a65846a7 189 bindtextdomain( 'squirrelmail', SM_PATH . 'locale/' );
88cb1b4d 190 textdomain( 'squirrelmail' );
7928848f 191 if (function_exists('bind_textdomain_codeset')) {
a5970d71 192 if ($sm_notAlias == 'ja_JP') {
193 bind_textdomain_codeset ("squirrelmail", 'EUC-JP');
194 } else {
195 bind_textdomain_codeset ("squirrelmail", $languages[$sm_notAlias]['CHARSET'] );
196 }
7928848f 197 }
f2374580 198 if (isset($languages[$sm_notAlias]['LOCALE'])){
199 $longlocale=$languages[$sm_notAlias]['LOCALE'];
200 } else {
201 $longlocale=$sm_notAlias;
202 }
88cb1b4d 203 if ( !ini_get('safe_mode') &&
f2374580 204 getenv( 'LC_ALL' ) != $longlocale ) {
205 putenv( "LC_ALL=$longlocale" );
206 putenv( "LANG=$longlocale" );
207 putenv( "LANGUAGE=$longlocale" );
a2a7852b 208 }
f2374580 209 setlocale(LC_ALL, $longlocale);
d9d28bdc 210
211 // Set text direction/alignment variables
212 if (isset($languages[$sm_notAlias]['DIR']) &&
213 $languages[$sm_notAlias]['DIR'] == 'rtl') {
214 $text_direction='rtl';
215 $left_align='right';
216 $right_align='left';
217 } else {
218 $text_direction='ltr';
219 $left_align='left';
220 $right_align='right';
221 }
222
f2374580 223 $squirrelmail_language = $sm_notAlias;
a5970d71 224 if ($squirrelmail_language == 'ja_JP') {
b05c8961 225 header ('Content-Type: text/html; charset=EUC-JP');
226 if (!function_exists('mb_internal_encoding')) {
51468260 227 // Error messages can't be displayed here
228 $error = 1;
229 // Revert to English if possible.
230 if (function_exists('setPref') && $username!='' && $data_dir!="") {
231 setPref($data_dir, $username, 'language', "en_US");
232 $error = 2;
233 }
a5970d71 234 // stop further execution in order not to get php errors on mb_internal_encoding().
51468260 235 return $error;
e842b215 236 }
237 if (function_exists('mb_language')) {
238 mb_language('Japanese');
b05c8961 239 }
240 mb_internal_encoding('EUC-JP');
241 mb_http_output('pass');
242 } else {
5c920668 243 header( 'Content-Type: text/html; charset=' . $languages[$sm_notAlias]['CHARSET'] );
a2a7852b 244 }
245}
51468260 246 return 0;
b05c8961 247}
a2a7852b 248
51468260 249/**
250 * Sets default_charset variable according to the one that is used by user's translations.
251 *
252 * Function changes global $default_charset variable in order to be sure, that it
253 * contains charset used by user's translation. Sanity of $squirrelmail_default_language
254 * and $default_charset combination provided in SquirrelMail config is also tested.
255 *
256 * There can be a $default_charset setting in the
257 * config.php file, but the user may have a different language
258 * selected for a user interface. This function checks the
259 * language selected by the user and tags the outgoing messages
260 * with the appropriate charset corresponding to the language
261 * selection. This is "more right" (tm), than just stamping the
262 * message blindly with the system-wide $default_charset.
263 */
a2a7852b 264function set_my_charset(){
94965562 265 global $data_dir, $username, $default_charset, $languages, $squirrelmail_default_language;
88cb1b4d 266
a2a7852b 267 $my_language = getPref($data_dir, $username, 'language');
5c920668 268 if (!$my_language) {
94965562 269 $my_language = $squirrelmail_default_language ;
5c920668 270 }
3ec81e63 271 // Catch removed translation
272 if (!isset($languages[$my_language])) {
273 $my_language="en_US";
274 }
a2a7852b 275 while (isset($languages[$my_language]['ALIAS'])) {
f7e8861e 276 $my_language = $languages[$my_language]['ALIAS'];
a2a7852b 277 }
5c920668 278 $my_charset = $languages[$my_language]['CHARSET'];
a2a7852b 279 if ($my_charset) {
280 $default_charset = $my_charset;
281 }
282}
283
a2a7852b 284/* ------------------------------ main --------------------------- */
285
5c920668 286global $squirrelmail_language, $languages, $use_gettext;
287
a2a7852b 288if (! isset($squirrelmail_language)) {
289 $squirrelmail_language = '';
290}
291
51468260 292/**
293 * Array specifies the available translations.
294 *
295 * Structure of array:
296 * $languages['language']['variable'] = 'value'
297 *
298 * Possible 'variable' names:
299 * NAME - Translation name in English
300 * CHARSET - Encoding used by translation
301 * ALIAS - used when 'language' is only short name and 'value' should provide long language name
302 * ALTNAME - Native translation name. Any 8bit symbols must be html encoded.
303 * LOCALE - Full locale name (in xx_XX.charset format)
304 * DIR - Text direction. Used to define Right-to-Left languages. Possible values 'rtl' or 'ltr'. If undefined - defaults to 'ltr'
305 * XTRA_CODE - translation uses special functions. 'value' provides name of that extra function
306 *
307 * Each 'language' definition requires NAME+CHARSET or ALIAS variables.
308 *
309 * @name $languages
310 * @global $languages
311 */
a8fa8e33 312$languages['bg_BG']['NAME'] = 'Bulgarian';
313$languages['bg_BG']['ALTNAME'] = '&#1041;&#1098;&#1083;&#1075;&#1072;&#1088;&#1089;&#1082;&#1080;';
314$languages['bg_BG']['CHARSET'] = 'windows-1251';
315$languages['bg']['ALIAS'] = 'bg_BG';
316
317$languages['ca_ES']['NAME'] = 'Catalan';
318$languages['ca_ES']['CHARSET'] = 'iso-8859-1';
319$languages['ca']['ALIAS'] = 'ca_ES';
320
321$languages['cs_CZ']['NAME'] = 'Czech';
322$languages['cs_CZ']['ALTNAME'] = '&#268;e&scaron;tina';
323$languages['cs_CZ']['CHARSET'] = 'iso-8859-2';
324$languages['cs']['ALIAS'] = 'cs_CZ';
325
326$languages['cy_GB']['NAME'] = 'Welsh';
327$languages['cy_GB']['ALTNAME'] = 'Cymraeg';
328$languages['cy_GB']['CHARSET'] = 'iso-8859-1';
329$languages['cy']['ALIAS'] = 'cy_GB';
330
331// Danish locale is da_DK.
332$languages['da_DK']['NAME'] = 'Danish';
333$languages['da_DK']['ALTNAME'] = 'Dansk';
334$languages['da_DK']['CHARSET'] = 'iso-8859-1';
335$languages['da']['ALIAS'] = 'da_DK';
336
337$languages['de_DE']['NAME'] = 'German';
338$languages['de_DE']['ALTNAME'] = 'Deutsch';
339$languages['de_DE']['CHARSET'] = 'iso-8859-1';
340$languages['de']['ALIAS'] = 'de_DE';
341
342$languages['el_GR']['NAME'] = 'Greek';
343$languages['el_GR']['ALTNAME'] = '&Epsilon;&lambda;&lambda;&eta;&nu;&iota;&kappa;&#940;';
344$languages['el_GR']['CHARSET'] = 'iso-8859-7';
345$languages['el']['ALIAS'] = 'el_GR';
a2a7852b 346
3bb3d83b 347$languages['en_GB']['NAME'] = 'British';
348$languages['en_GB']['CHARSET'] = 'iso-8859-15';
349$languages['en_GB']['LOCALE'] = 'en_GB.ISO-8859-15';
350
5c920668 351$languages['en_US']['NAME'] = 'English';
352$languages['en_US']['CHARSET'] = 'iso-8859-1';
353$languages['en']['ALIAS'] = 'en_US';
a2a7852b 354
a8fa8e33 355$languages['es_ES']['NAME'] = 'Spanish';
356$languages['es_ES']['ALTNAME'] = 'Espa&ntilde;ol';
357$languages['es_ES']['CHARSET'] = 'iso-8859-1';
358$languages['es']['ALIAS'] = 'es_ES';
359
360$languages['et_EE']['NAME'] = 'Estonian';
361$languages['et_EE']['CHARSET'] = 'iso-8859-15';
362$languages['et']['ALIAS'] = 'et_EE';
363
364$languages['fo_FO']['NAME'] = 'Faroese';
365$languages['fo_FO']['CHARSET'] = 'iso-8859-1';
366$languages['fo']['ALIAS'] = 'fo_FO';
367
368$languages['fi_FI']['NAME'] = 'Finnish';
369$languages['fi_FI']['ALTNAME'] = 'Suomi';
370$languages['fi_FI']['CHARSET'] = 'iso-8859-1';
371$languages['fi']['ALIAS'] = 'fi_FI';
372
373$languages['fr_FR']['NAME'] = 'French';
374$languages['fr_FR']['ALTNAME'] = 'Fran&#231;ais';
375$languages['fr_FR']['CHARSET'] = 'iso-8859-1';
376$languages['fr']['ALIAS'] = 'fr_FR';
377
378$languages['hr_HR']['NAME'] = 'Croatian';
379$languages['hr_HR']['CHARSET'] = 'iso-8859-2';
380$languages['hr']['ALIAS'] = 'hr_HR';
381
382$languages['hu_HU']['NAME'] = 'Hungarian';
383$languages['hu_HU']['ALTNAME'] = 'Magyar';
384$languages['hu_HU']['CHARSET'] = 'iso-8859-2';
385$languages['hu']['ALIAS'] = 'hu_HU';
386
387$languages['id_ID']['NAME'] = 'Indonesian';
388$languages['id_ID']['ALTNAME'] = 'Bahasa Indonesia';
389$languages['id_ID']['CHARSET'] = 'iso-8859-1';
390$languages['id']['ALIAS'] = 'id_ID';
391
392$languages['is_IS']['NAME'] = 'Icelandic';
393$languages['is_IS']['ALTNAME'] = '&Iacute;slenska';
394$languages['is_IS']['CHARSET'] = 'iso-8859-1';
395$languages['is']['ALIAS'] = 'is_IS';
396
397$languages['it_IT']['NAME'] = 'Italian';
398$languages['it_IT']['CHARSET'] = 'iso-8859-1';
399$languages['it']['ALIAS'] = 'it_IT';
400
401$languages['ja_JP']['NAME'] = 'Japanese';
402$languages['ja_JP']['ALTNAME'] = '&#26085;&#26412;&#35486;';
403$languages['ja_JP']['CHARSET'] = 'iso-2022-jp';
51468260 404$languages['ja_JP']['LOCALE'] = 'ja_JP.EUC-JP';
a8fa8e33 405$languages['ja_JP']['XTRA_CODE'] = 'japanese_charset_xtra';
406$languages['ja']['ALIAS'] = 'ja_JP';
407
408$languages['ko_KR']['NAME'] = 'Korean';
409$languages['ko_KR']['CHARSET'] = 'euc-KR';
410$languages['ko_KR']['XTRA_CODE'] = 'korean_charset_xtra';
411$languages['ko']['ALIAS'] = 'ko_KR';
412
413$languages['lt_LT']['NAME'] = 'Lithuanian';
414$languages['lt_LT']['ALTNAME'] = 'Lietuvi&#371;';
415$languages['lt_LT']['CHARSET'] = 'utf-8';
416$languages['lt_LT']['LOCALE'] = 'lt_LT.UTF-8';
417$languages['lt']['ALIAS'] = 'lt_LT';
418
419$languages['nl_NL']['NAME'] = 'Dutch';
420$languages['nl_NL']['ALTNAME'] = 'Nederlands';
421$languages['nl_NL']['CHARSET'] = 'iso-8859-1';
422$languages['nl']['ALIAS'] = 'nl_NL';
423
424$languages['ms_MY']['NAME'] = 'Malay';
425$languages['ms_MY']['ALTNAME'] = 'Bahasa Melayu';
426$languages['ms_MY']['CHARSET'] = 'iso-8859-1';
427$languages['my']['ALIAS'] = 'ms_MY';
428
850db3c8 429$languages['nb_NO']['NAME'] = 'Norwegian (Bokm&aring;l)';
430$languages['nb_NO']['ALTNAME'] = 'Norsk (Bokm&aring;l)';
431$languages['nb_NO']['CHARSET'] = 'iso-8859-1';
432$languages['nb']['ALIAS'] = 'nb_NO';
a8fa8e33 433
434$languages['nn_NO']['NAME'] = 'Norwegian (Nynorsk)';
435$languages['nn_NO']['ALTNAME'] = 'Norsk (Nynorsk)';
436$languages['nn_NO']['CHARSET'] = 'iso-8859-1';
437
438$languages['pl_PL']['NAME'] = 'Polish';
439$languages['pl_PL']['ALTNAME'] = 'Polski';
440$languages['pl_PL']['CHARSET'] = 'iso-8859-2';
441$languages['pl']['ALIAS'] = 'pl_PL';
442
443$languages['pt_PT']['NAME'] = 'Portuguese (Portugal)';
444$languages['pt_PT']['CHARSET'] = 'iso-8859-1';
445$languages['pt']['ALIAS'] = 'pt_PT';
446
447$languages['pt_BR']['NAME'] = 'Portuguese (Brazil)';
448$languages['pt_BR']['ALTNAME'] = 'Portugu&ecirc;s do Brasil';
449$languages['pt_BR']['CHARSET'] = 'iso-8859-1';
450
451$languages['ro_RO']['NAME'] = 'Romanian';
452$languages['ro_RO']['ALTNAME'] = 'Rom&acirc;n&#259;';
453$languages['ro_RO']['CHARSET'] = 'iso-8859-2';
454$languages['ro']['ALIAS'] = 'ro_RO';
455
456$languages['ru_RU']['NAME'] = 'Russian';
457$languages['ru_RU']['ALTNAME'] = '&#1056;&#1091;&#1089;&#1089;&#1082;&#1080;&#1081;';
458$languages['ru_RU']['CHARSET'] = 'utf-8';
459$languages['ru_RU']['LOCALE'] = 'ru_RU.UTF-8';
460$languages['ru']['ALIAS'] = 'ru_RU';
461
462$languages['sk_SK']['NAME'] = 'Slovak';
463$languages['sk_SK']['CHARSET'] = 'iso-8859-2';
464$languages['sk']['ALIAS'] = 'sk_SK';
465
466$languages['sl_SI']['NAME'] = 'Slovenian';
467$languages['sl_SI']['ALTNAME'] = 'Sloven&scaron;&#269;ina';
468$languages['sl_SI']['CHARSET'] = 'iso-8859-2';
469$languages['sl']['ALIAS'] = 'sl_SI';
470
471$languages['sr_YU']['NAME'] = 'Serbian';
472$languages['sr_YU']['ALTNAME'] = 'Srpski';
473$languages['sr_YU']['CHARSET'] = 'iso-8859-2';
474$languages['sr']['ALIAS'] = 'sr_YU';
475
476$languages['sv_SE']['NAME'] = 'Swedish';
477$languages['sv_SE']['ALTNAME'] = 'Svenska';
478$languages['sv_SE']['CHARSET'] = 'iso-8859-1';
479$languages['sv']['ALIAS'] = 'sv_SE';
480
481$languages['th_TH']['NAME'] = 'Thai';
482$languages['th_TH']['CHARSET'] = 'tis-620';
483$languages['th']['ALIAS'] = 'th_TH';
484
551a09c7 485$languages['tl_PH']['NAME'] = 'Tagalog';
486$languages['tl_PH']['CHARSET'] = 'iso-8859-1';
487$languages['tl_PH']['LOCALE'] = 'tl_PH.ISO-8859-1';
488$languages['tl']['ALIAS'] = 'tl_PH';
489
a8fa8e33 490$languages['tr_TR']['NAME'] = 'Turkish';
491$languages['tr_TR']['CHARSET'] = 'iso-8859-9';
492$languages['tr']['ALIAS'] = 'tr_TR';
493
494$languages['zh_TW']['NAME'] = 'Chinese Trad';
495$languages['zh_TW']['CHARSET'] = 'big5';
496$languages['tw']['ALIAS'] = 'zh_TW';
497
498$languages['zh_CN']['NAME'] = 'Chinese Simp';
499$languages['zh_CN']['CHARSET'] = 'gb2312';
500$languages['cn']['ALIAS'] = 'zh_CN';
060c9483 501
87c6b544 502/*
a8fa8e33 503$languages['uk_UA']['NAME'] = 'Ukrainian';
504$languages['uk_UA']['CHARSET'] = 'koi8-u';
505$languages['uk']['ALIAS'] = 'uk_UA';
87c6b544 506*/
87c6b544 507/*
850db3c8 508$languages['vi_VN']['NAME'] = 'Vietnamese';
509$languages['vi_VN']['CHARSET'] = 'utf-8';
510$languages['vi']['ALIAS'] = 'vi_VN';
87c6b544 511*/
a8fa8e33 512
d3b57948 513// Right to left languages
a8fa8e33 514$languages['ar']['NAME'] = 'Arabic';
515$languages['ar']['CHARSET'] = 'windows-1256';
516$languages['ar']['DIR'] = 'rtl';
517
4417eead 518$languages['fa_IR']['NAME'] = 'Farsi';
519$languages['fa_IR']['CHARSET'] = 'utf-8';
520$languages['fa_IR']['DIR'] = 'rtl';
521$languages['fa_IR']['LOCALE'] = 'fa_IR.UTF-8';
522$languages['fa']['ALIAS'] = 'fa_IR';
523
a8fa8e33 524$languages['he_IL']['NAME'] = 'Hebrew';
525$languages['he_IL']['CHARSET'] = 'windows-1255';
526$languages['he_IL']['DIR'] = 'rtl';
527$languages['he']['ALIAS'] = 'he_IL';
d3b57948 528
5c920668 529/* Detect whether gettext is installed. */
a2a7852b 530$gettext_flags = 0;
531if (function_exists('_')) {
532 $gettext_flags += 1;
533}
534if (function_exists('bindtextdomain')) {
535 $gettext_flags += 2;
536}
537if (function_exists('textdomain')) {
538 $gettext_flags += 4;
539}
540
5c920668 541/* If gettext is fully loaded, cool */
a2a7852b 542if ($gettext_flags == 7) {
543 $use_gettext = true;
544}
5c920668 545/* If we can fake gettext, try that */
a2a7852b 546elseif ($gettext_flags == 0) {
547 $use_gettext = true;
e7ab8c9d 548 include_once(SM_PATH . 'functions/gettext.php');
a2a7852b 549} else {
5c920668 550 /* Uh-ho. A weird install */
a2a7852b 551 if (! $gettext_flags & 1) {
552 function _($str) {
553 return $str;
554 }
555 }
556 if (! $gettext_flags & 2) {
557 function bindtextdomain() {
558 return;
559 }
560 }
561 if (! $gettext_flags & 4) {
562 function textdomain() {
563 return;
564 }
565 }
566}
567
1d33e35e 568
51468260 569/**
1d33e35e 570 * Japanese charset extra function
571 *
51468260 572 * Action performed by function is defined by first argument.
573 * Default return value is defined by second argument.
574 * Use of third argument depends on action.
575 *
576 * @param string action performed by this function.
577 * possible values:
578 * decode - convert returned string to euc-jp. third argument unused
579 * encode - convert returned string to jis. third argument unused
580 * strimwidth - third argument=$width. trims string to $width symbols.
581 * encodeheader - create base64 encoded header in iso-2022-jp. third argument unused
582 * decodeheader - return human readable string from mime header. string is returned in euc-jp. third argument unused
583 * downloadfilename - third argument $useragent. Arguments provide browser info. Returns shift-jis or euc-jp encoded file name
584 * wordwrap - third argument=$wrap. wraps text at $wrap symbols
585 * utf7-imap_encode - returns string converted from euc-jp to utf7-imap. third argument unused
586 * utf7-imap_decode - returns string converted from utf7-imap to euc-jp. third argument unused
587 * @param string default return value
1d33e35e 588 */
589function japanese_charset_xtra() {
590 $ret = func_get_arg(1); /* default return value */
591 if (function_exists('mb_detect_encoding')) {
592 switch (func_get_arg(0)) { /* action */
593 case 'decode':
e842b215 594 $detect_encoding = @mb_detect_encoding($ret);
1d33e35e 595 if ($detect_encoding == 'JIS' ||
596 $detect_encoding == 'EUC-JP' ||
e842b215 597 $detect_encoding == 'SJIS' ||
598 $detect_encoding == 'UTF-8') {
1d33e35e 599
e842b215 600 $ret = mb_convert_kana(mb_convert_encoding($ret, 'EUC-JP', 'AUTO'), "KV");
1d33e35e 601 }
602 break;
603 case 'encode':
e842b215 604 $detect_encoding = @mb_detect_encoding($ret);
1d33e35e 605 if ($detect_encoding == 'JIS' ||
606 $detect_encoding == 'EUC-JP' ||
e842b215 607 $detect_encoding == 'SJIS' ||
608 $detect_encoding == 'UTF-8') {
1d33e35e 609
e842b215 610 $ret = mb_convert_encoding(mb_convert_kana($ret, "KV"), 'JIS', 'AUTO');
1d33e35e 611 }
612 break;
613 case 'strimwidth':
614 $width = func_get_arg(2);
615 $ret = mb_strimwidth($ret, 0, $width, '...');
616 break;
617 case 'encodeheader':
8ba05cbe 618 $result = '';
619 if (strlen($ret) > 0) {
620 $tmpstr = mb_substr($ret, 0, 1);
621 $prevcsize = strlen($tmpstr);
622 for ($i = 1; $i < mb_strlen($ret); $i++) {
623 $tmp = mb_substr($ret, $i, 1);
624 if (strlen($tmp) == $prevcsize) {
625 $tmpstr .= $tmp;
626 } else {
627 if ($prevcsize == 1) {
628 $result .= $tmpstr;
629 } else {
e842b215 630 $result .= str_replace(' ', '',
631 mb_encode_mimeheader($tmpstr,'iso-2022-jp','B',''));
8ba05cbe 632 }
633 $tmpstr = $tmp;
634 $prevcsize = strlen($tmp);
635 }
636 }
637 if (strlen($tmpstr)) {
638 if (strlen(mb_substr($tmpstr, 0, 1)) == 1)
639 $result .= $tmpstr;
640 else
e842b215 641 $result .= str_replace(' ', '',
642 mb_encode_mimeheader($tmpstr,'iso-2022-jp','B',''));
8ba05cbe 643 }
644 }
645 $ret = $result;
1d33e35e 646 break;
647 case 'decodeheader':
648 $ret = str_replace("\t", "", $ret);
649 if (eregi('=\\?([^?]+)\\?(q|b)\\?([^?]+)\\?=', $ret))
e842b215 650 $ret = @mb_decode_mimeheader($ret);
651 $ret = @mb_convert_encoding($ret, 'EUC-JP', 'AUTO');
1d33e35e 652 break;
653 case 'downloadfilename':
654 $useragent = func_get_arg(2);
655 if (strstr($useragent, 'Windows') !== false ||
656 strstr($useragent, 'Mac_') !== false) {
657 $ret = mb_convert_encoding($ret, 'SJIS', 'AUTO');
658 } else {
659 $ret = mb_convert_encoding($ret, 'EUC-JP', 'AUTO');
660}
661 break;
e842b215 662 case 'wordwrap':
663 $no_begin = "\x21\x25\x29\x2c\x2e\x3a\x3b\x3f\x5d\x7d\xa1\xf1\xa1\xeb\xa1" .
664 "\xc7\xa1\xc9\xa2\xf3\xa1\xec\xa1\xed\xa1\xee\xa1\xa2\xa1\xa3\xa1\xb9" .
665 "\xa1\xd3\xa1\xd5\xa1\xd7\xa1\xd9\xa1\xdb\xa1\xcd\xa4\xa1\xa4\xa3\xa4" .
666 "\xa5\xa4\xa7\xa4\xa9\xa4\xc3\xa4\xe3\xa4\xe5\xa4\xe7\xa4\xee\xa1\xab" .
667 "\xa1\xac\xa1\xb5\xa1\xb6\xa5\xa1\xa5\xa3\xa5\xa5\xa5\xa7\xa5\xa9\xa5" .
668 "\xc3\xa5\xe3\xa5\xe5\xa5\xe7\xa5\xee\xa5\xf5\xa5\xf6\xa1\xa6\xa1\xbc" .
669 "\xa1\xb3\xa1\xb4\xa1\xaa\xa1\xf3\xa1\xcb\xa1\xa4\xa1\xa5\xa1\xa7\xa1" .
670 "\xa8\xa1\xa9\xa1\xcf\xa1\xd1";
671 $no_end = "\x5c\x24\x28\x5b\x7b\xa1\xf2\x5c\xa1\xc6\xa1\xc8\xa1\xd2\xa1" .
672 "\xd4\xa1\xd6\xa1\xd8\xa1\xda\xa1\xcc\xa1\xf0\xa1\xca\xa1\xce\xa1\xd0\xa1\xef";
673 $wrap = func_get_arg(2);
674
675 if (strlen($ret) >= $wrap &&
676 substr($ret, 0, 1) != '>' &&
677 strpos($ret, 'http://') === FALSE &&
678 strpos($ret, 'https://') === FALSE &&
679 strpos($ret, 'ftp://') === FALSE) {
680
681 $ret = mb_convert_kana($ret, "KV");
682
683 $line_new = '';
684 $ptr = 0;
685
686 while ($ptr < strlen($ret) - 1) {
687 $l = mb_strcut($ret, $ptr, $wrap);
688 $ptr += strlen($l);
689 $tmp = $l;
690
691 $l = mb_strcut($ret, $ptr, 2);
692 while (strlen($l) != 0 && mb_strpos($no_begin, $l) !== FALSE ) {
693 $tmp .= $l;
694 $ptr += strlen($l);
695 $l = mb_strcut($ret, $ptr, 1);
696 }
697 $line_new .= $tmp;
698 if ($ptr < strlen($ret) - 1)
699 $line_new .= "\n";
700 }
701 $ret = $line_new;
702 }
703 break;
704 case 'utf7-imap_encode':
705 $ret = mb_convert_encoding($ret, 'UTF7-IMAP', 'EUC-JP');
706 break;
707 case 'utf7-imap_decode':
708 $ret = mb_convert_encoding($ret, 'EUC-JP', 'UTF7-IMAP');
709 break;
1d33e35e 710 }
711 }
712 return $ret;
713}
714
715
51468260 716/**
717 * Korean charset extra functions
718 *
719 * Action performed by function is defined by first argument.
720 * Default return value is defined by second argument.
721 *
722 * @param string action performed by this function.
723 * possible values:
724 * downloadfilename - Hangul(Korean Character) Attached File Name Fix.
725 * @param string default return value
1d33e35e 726 */
727function korean_charset_xtra() {
728
729 $ret = func_get_arg(1); /* default return value */
730 if (func_get_arg(0) == 'downloadfilename') { /* action */
731 $ret = str_replace("\x0D\x0A", '', $ret); /* Hanmail's CR/LF Clear */
732 for ($i=0;$i<strlen($ret);$i++) {
733 if ($ret[$i] >= "\xA1" && $ret[$i] <= "\xFE") { /* 0xA1 - 0XFE are Valid */
734 $i++;
735 continue;
736 } else if (($ret[$i] >= 'a' && $ret[$i] <= 'z') || /* From Original ereg_replace in download.php */
737 ($ret[$i] >= 'A' && $ret[$i] <= 'Z') ||
738 ($ret[$i] == '.') || ($ret[$i] == '-')) {
739 continue;
740 } else {
741 $ret[$i] = '_';
742 }
743 }
744
745 }
1d33e35e 746 return $ret;
747}
748
db08d0c3 749/**
750 * Replaces non-braking spaces inserted by some browsers with regular space
751 *
9af9c0a2 752 * This function can be used to replace non-braking space symbols
753 * that are inserted in forms by some browsers instead of normal
754 * space symbol.
db08d0c3 755 *
756 * @param string $string Text that needs to be cleaned
757 * @param string $charset Charset used in text
758 * @return string Cleaned text
9af9c0a2 759 */
760function cleanup_nbsp($string,$charset) {
761
762 // reduce number of case statements
763 if (stristr('iso-8859-',substr($charset,0,9))){
764 $output_charset="iso-8859-x";
765 }
766 if (stristr('windows-125',substr($charset,0,11))){
767 $output_charset="cp125x";
768 }
769 if (stristr('koi8',substr($charset,0,4))){
770 $output_charset="koi8-x";
771 }
772 if (! isset($output_charset)){
773 $output_charset=strtolower($charset);
774 }
775
776// where is non-braking space symbol
777switch($output_charset):
778 case "iso-8859-x":
97b9c02f 779 case "cp125x":
780 case "iso-2022-jp":
9af9c0a2 781 $nbsp="\xA0";
782 break;
9af9c0a2 783 case "koi8-x":
784 $nbsp="\x9A";
785 break;
786 case "utf-8":
787 $nbsp="\xC2\xA0";
788 break;
9af9c0a2 789 default:
790 // don't change string if charset is unmatched
791 return $string;
792endswitch;
793
794// return space instead of non-braking space.
795 return str_replace($nbsp,' ',$string);
796}
4e519821 797
db08d0c3 798/**
799 * Function informs if it is safe to convert given charset to the one that is used by user.
800 *
801 * It is safe to use conversion only if user uses utf-8 encoding and when
802 * converted charset is similar to the one that is used by user.
803 *
804 * @param string $input_charset Charset of text that needs to be converted
805 * @return bool is it possible to convert to user's charset
806 */
4e519821 807function is_conversion_safe($input_charset) {
808 global $languages, $sm_notAlias, $default_charset;
809
810 // convert to lower case
811 $input_charset = strtolower($input_charset);
812
813 // Is user's locale Unicode based ?
814 if ( $default_charset == "utf-8" ) {
815 return true;
816 }
817
818 // Charsets that are similar
819switch ($default_charset):
820case "windows-1251":
821 if ( $input_charset == "iso-8859-5" ||
822 $input_charset == "koi8-r" ||
823 $input_charset == "koi8-u" ) {
824 return true;
825 } else {
826 return false;
827 }
828case "windows-1257":
829 if ( $input_charset == "iso-8859-13" ||
830 $input_charset == "iso-8859-4" ) {
831 return true;
832 } else {
833 return false;
834 }
835case "iso-8859-4":
836 if ( $input_charset == "iso-8859-13" ||
837 $input_charset == "windows-1257" ) {
838 return true;
839 } else {
840 return false;
841 }
842case "iso-8859-5":
843 if ( $input_charset == "windows-1251" ||
844 $input_charset == "koi8-r" ||
845 $input_charset == "koi8-u" ) {
846 return true;
847 } else {
848 return false;
849 }
850case "iso-8859-13":
851 if ( $input_charset == "iso-8859-4" ||
852 $input_charset == "windows-1257" ) {
853 return true;
854 } else {
855 return false;
856 }
857case "koi8-r":
858 if ( $input_charset == "windows-1251" ||
859 $input_charset == "iso-8859-5" ||
860 $input_charset == "koi8-u" ) {
861 return true;
862 } else {
863 return false;
864 }
865case "koi8-u":
866 if ( $input_charset == "windows-1251" ||
867 $input_charset == "iso-8859-5" ||
868 $input_charset == "koi8-r" ) {
869 return true;
870 } else {
871 return false;
872 }
873default:
874 return false;
875endswitch;
876}
51468260 877?>
Unnamed repository; edit this file 'description' to name the repository.