[squirrelmail.git] / functions / i18n.php

<?php

/**
 * i18n.php
 *
 * Copyright (c) 1999-2003 The SquirrelMail Project Team
 * Licensed under the GNU GPL. For full terms see the file COPYING.
 *
 * This file contains variuos functions that are needed to do
 * internationalization of SquirrelMail.
 *
 * Internally the output character set is used. Other characters are
 * encoded using Unicode entities according to HTML 4.0.
 *
 * $Id$
 */

require_once(SM_PATH . 'functions/global.php');

/* Decodes a string to the internal encoding from the given charset */
function charset_decode ($charset, $string) {
    global $languages, $squirrelmail_language, $default_charset;

    if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
        function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) {
        $string = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $string);
    }

    $charset = strtolower($charset);

    set_my_charset();

    // Variables that allow to use functions without function_exist() calls
    $use_php_recode=false;
    $use_php_iconv=false;

    // Don't do conversion if charset is the same.
    if ( $charset == strtolower($default_charset) )
          return htmlspecialchars($string);

    // catch iso-8859-8-i thing
    if ( $charset == "iso-8859-8-i" )
              $charset = "iso-8859-8";

    /*
     * Recode converts html special characters automatically if you use 
     * 'charset..html' decoding. There is no documented way to put -d option 
     * into php recode function call.
     */
    if ( $use_php_recode ) {
      if ( $default_charset == "utf-8" ) {
	// other charsets can be converted to utf-8 without loss.
	// and output string is smaller
	$string = recode_string($charset . "..utf-8",$string);
	return htmlspecialchars($string);
      } else {
	$string = recode_string($charset . "..html",$string);
	// recode does not convert single quote, htmlspecialchars does.
	$string = str_replace("'", '&#039;', $string);
	return $string;
      }
    }

    // iconv functions does not have html target and can be used only with utf-8
    if ( $use_php_iconv && $default_charset=='utf-8') {
      $string = iconv($charset,$default_charset,$string);
      return htmlspecialchars($string);
    }

    // If we don't use recode and iconv, we'll do it old way.

    /* All HTML special characters are 7 bit and can be replaced first */
    
    $string = htmlspecialchars ($string);

    /* controls cpu and memory intensive decoding cycles */
    $agresive_decoding = false;

    if (ereg('iso-8859-([[:digit:]]+)', $charset, $res)) {
        if ($res[1] == '1') {
	    include_once(SM_PATH . 'functions/decode/iso8859-1.php');
            $ret = charset_decode_iso8859_1 ($string);
        } else if ($res[1] == '2') {
	    include_once(SM_PATH . 'functions/decode/iso8859-2.php');
            $ret = charset_decode_iso8859_2 ($string);
        } else if ($res[1] == '3') {
	    include_once(SM_PATH . 'functions/decode/iso8859-3.php');
            $ret = charset_decode_iso8859_3 ($string);
        } else if ($res[1] == '4') {
	    include_once(SM_PATH . 'functions/decode/iso8859-4.php');
            $ret = charset_decode_iso8859_4 ($string);
        } else if ($res[1] == '5') {
	    include_once(SM_PATH . 'functions/decode/iso8859-5.php');
            $ret = charset_decode_iso8859_5 ($string);
        } else if ($res[1] == '6') {
	    include_once(SM_PATH . 'functions/decode/iso8859-6.php');
            $ret = charset_decode_iso8859_6 ($string);
        } else if ($res[1] == '7') {
	    include_once(SM_PATH . 'functions/decode/iso8859-7.php');
            $ret = charset_decode_iso8859_7 ($string);
        } else if ($res[1] == '8') {
	    include_once(SM_PATH . 'functions/decode/iso8859-8.php');
            $ret = charset_decode_iso8859_8 ($string);
        } else if ($res[1] == '9') {
	    include_once(SM_PATH . 'functions/decode/iso8859-9.php');
            $ret = charset_decode_iso8859_9 ($string);
        } else if ($res[1] == '10') {
	    include_once(SM_PATH . 'functions/decode/iso8859-10.php');
            $ret = charset_decode_iso8859_10 ($string);
        } else if ($res[1] == '11') {
	    include_once(SM_PATH . 'functions/decode/iso8859-11.php');
            $ret = charset_decode_iso8859_11 ($string);
        } else if ($res[1] == '13') {
	    include_once(SM_PATH . 'functions/decode/iso8859-13.php');
            $ret = charset_decode_iso8859_13 ($string);
        } else if ($res[1] == '14') {
	    include_once(SM_PATH . 'functions/decode/iso8859-14.php');
            $ret = charset_decode_iso8859_14 ($string);
        } else if ($res[1] == '15') {
	    include_once(SM_PATH . 'functions/decode/iso8859-15.php');
            $ret = charset_decode_iso8859_15 ($string);
        } else if ($res[1] == '16') {
	    include_once(SM_PATH . 'functions/decode/iso8859-16.php');
            $ret = charset_decode_iso8859_16 ($string);
        } else {
            $ret = charset_decode_iso_8859_default ($string);
        }
    } else if ($charset == 'ns_4551-1') {
        $ret = charset_decode_ns_4551_1 ($string);
    } else if ($charset == 'koi8-r') {
        include_once(SM_PATH . 'functions/decode/koi8-r.php');
        $ret = charset_decode_koi8r ($string);
    } else if ($charset == 'koi8-u') {
        include_once(SM_PATH . 'functions/decode/koi8-u.php');
        $ret = charset_decode_koi8u ($string);
    } else if ($charset == 'windows-1250') {
        include_once(SM_PATH . 'functions/decode/cp1250.php');
        $ret = charset_decode_cp1250 ($string);
    } else if ($charset == 'windows-1251') {
        include_once(SM_PATH . 'functions/decode/cp1251.php');
        $ret = charset_decode_cp1251 ($string);
    } else if ($charset == 'windows-1252') {
        include_once(SM_PATH . 'functions/decode/cp1252.php');
        $ret = charset_decode_cp1252 ($string);
    } else if ($charset == 'windows-1253') {
        include_once(SM_PATH . 'functions/decode/cp1253.php');
	$ret = charset_decode_cp1253 ($string);
    } else if ($charset == 'windows-1254') {
        include_once(SM_PATH . 'functions/decode/cp1254.php');
	$ret = charset_decode_cp1254 ($string);
    } else if ($charset == 'windows-1255') {
        include_once(SM_PATH . 'functions/decode/cp1255.php');
	$ret = charset_decode_cp1255 ($string);
    } else if ($charset == 'windows-1256') {
        include_once(SM_PATH . 'functions/decode/cp1256.php');
	$ret = charset_decode_cp1256 ($string);
    } else if ($charset == 'windows-1257') {
        include_once(SM_PATH . 'functions/decode/cp1257.php');
        $ret = charset_decode_cp1257 ($string);
    } else if ($charset == 'windows-1258') {
        include_once(SM_PATH . 'functions/decode/cp1258.php');
        $ret = charset_decode_cp1258 ($string);
    } else if ($charset == 'x-mac-roman') {
        include_once(SM_PATH . 'functions/decode/cp10000.php');
        $ret = charset_decode_cp10000 ($string);
    } else if ($charset == 'x-mac-greek') {
        include_once(SM_PATH . 'functions/decode/cp10006.php');
        $ret = charset_decode_cp10006 ($string);
    } else if ($charset == 'x-mac-cyrillic') {
        include_once(SM_PATH . 'functions/decode/cp10007.php');
        $ret = charset_decode_cp10007 ($string);
    } else if ($charset == 'x-mac-ukrainian') {
        include_once(SM_PATH . 'functions/decode/cp10017.php');
        $ret = charset_decode_cp10017 ($string);
    } else if ($charset == 'x-mac-centraleurroman') {
        include_once(SM_PATH . 'functions/decode/cp10029.php');
        $ret = charset_decode_cp10029 ($string);
    } else if ($charset == 'x-mac-icelandic') {
        include_once(SM_PATH . 'functions/decode/cp10079.php');
        $ret = charset_decode_cp10079 ($string);
    } else if ($charset == 'x-mac-turkish') {
        include_once(SM_PATH . 'functions/decode/cp10081.php');
        $ret = charset_decode_cp10081 ($string);
    } else if ($charset == 'ibm855') {
        include_once(SM_PATH . 'functions/decode/cp855.php');
        $ret = charset_decode_cp855 ($string);
    } else if ($charset == 'ibm866') {
        include_once(SM_PATH . 'functions/decode/cp866.php');
        $ret = charset_decode_cp866 ($string);
    } else if ($charset == 'iso-ir-111') {
        include_once(SM_PATH . 'functions/decode/iso-ir-111.php');
        $ret = charset_decode_iso_ir_111 ($string);
    } else if ($charset == 'tis-620') {
        include_once(SM_PATH . 'functions/decode/tis620.php');
        $ret = charset_decode_tis620 ($string);
    } else if ($charset == 'big5' and $agresive_decoding ) {
        include_once(SM_PATH . 'functions/decode/big5.php');
        $ret = charset_decode_big5 ($string);
    } else if ($charset == 'gb2312' and $agresive_decoding ) {
        include_once(SM_PATH . 'functions/decode/gb2312.php');
        $ret = charset_decode_gb2312 ($string);
    } else if ($charset == 'utf-8') {
        include_once(SM_PATH . 'functions/decode/utf-8.php');
	$ret = charset_decode_utf8 ($string);
    } else {
        $ret = $string;
    }
    return( $ret );
}


/* Remove all 8 bit characters from all other ISO-8859 character sets */
function charset_decode_iso_8859_default ($string) {
    return (strtr($string, "\240\241\242\243\244\245\246\247".
                    "\250\251\252\253\254\255\256\257".
                    "\260\261\262\263\264\265\266\267".
                    "\270\271\272\273\274\275\276\277".
                    "\300\301\302\303\304\305\306\307".
                    "\310\311\312\313\314\315\316\317".
                    "\320\321\322\323\324\325\326\327".
                    "\330\331\332\333\334\335\336\337".
                    "\340\341\342\343\344\345\346\347".
                    "\350\351\352\353\354\355\356\357".
                    "\360\361\362\363\364\365\366\367".
                    "\370\371\372\373\374\375\376\377",
                    "????????????????????????????????????????".
                    "????????????????????????????????????????".
                    "????????????????????????????????????????".
                    "????????"));

}

/*
 * This is the same as ISO-646-NO and is used by some
 * Microsoft programs when sending Norwegian characters
 */
function charset_decode_ns_4551_1 ($string) {
    /*
     * These characters are:
     * Latin capital letter AE
     * Latin capital letter O with stroke
     * Latin capital letter A with ring above
     * and the same as small letters
     */
Commit	Line	Data
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');
961ca3d8	19
a2a7852b	20	/* Decodes a string to the internal encoding from the given charset */
a2a7852b	21	function charset_decode ($charset, $string) {
3ec81e63	22	global $languages, $squirrelmail_language, $default_charset;
a2a7852b	23
3714db45	24	if (isset($languages[$squirrelmail_language]['XTRA_CODE']) &&
3714db45	25	function_exists($languages[$squirrelmail_language]['XTRA_CODE'])) {
6fbd125b	26	$string = $languages[$squirrelmail_language]['XTRA_CODE']('decode', $string);
6fbd125b	27	}
b05c8961	28
3ec81e63	29	$charset = strtolower($charset);
	30
	31	set_my_charset();
	32
	33	// Variables that allow to use functions without function_exist() calls
	34	$use_php_recode=false;
	35	$use_php_iconv=false;
	36
	37	// Don't do conversion if charset is the same.
	38	if ( $charset == strtolower($default_charset) )
	39	return htmlspecialchars($string);
	40
	41	// catch iso-8859-8-i thing
	42	if ( $charset == "iso-8859-8-i" )
	43	$charset = "iso-8859-8";
	44
	45	/*
	46	* Recode converts html special characters automatically if you use
	47	* 'charset..html' decoding. There is no documented way to put -d option
	48	* into php recode function call.
	49	*/
	50	if ( $use_php_recode ) {
	51	if ( $default_charset == "utf-8" ) {
	52	// other charsets can be converted to utf-8 without loss.
	53	// and output string is smaller
	54	$string = recode_string($charset . "..utf-8",$string);
	55	return htmlspecialchars($string);
	56	} else {
	57	$string = recode_string($charset . "..html",$string);
	58	// recode does not convert single quote, htmlspecialchars does.
	59	$string = str_replace("'", ''', $string);
	60	return $string;
	61	}
	62	}
	63
	64	// iconv functions does not have html target and can be used only with utf-8
	65	if ( $use_php_iconv && $default_charset=='utf-8') {
	66	$string = iconv($charset,$default_charset,$string);
	67	return htmlspecialchars($string);
	68	}
	69
	70	// If we don't use recode and iconv, we'll do it old way.
	71
a2a7852b	72	/* All HTML special characters are 7 bit and can be replaced first */
cef054e4	73
098ea084	74	$string = htmlspecialchars ($string);
a2a7852b	75
5dd23dac	76	/* controls cpu and memory intensive decoding cycles */
	77	$agresive_decoding = false;
	78
a2a7852b	79	if (ereg('iso-8859-([[:digit:]]+)', $charset, $res)) {
a2a7852b	80	if ($res[1] == '1') {
5dd23dac	81	include_once(SM_PATH . 'functions/decode/iso8859-1.php');
5dd23dac	82	$ret = charset_decode_iso8859_1 ($string);
a2a7852b	83	} else if ($res[1] == '2') {
5dd23dac	84	include_once(SM_PATH . 'functions/decode/iso8859-2.php');
5dd23dac	85	$ret = charset_decode_iso8859_2 ($string);
3a66bed2	86	} else if ($res[1] == '3') {
	87	include_once(SM_PATH . 'functions/decode/iso8859-3.php');
	88	$ret = charset_decode_iso8859_3 ($string);
9be313d5	89	} else if ($res[1] == '4') {
3a66bed2	90	include_once(SM_PATH . 'functions/decode/iso8859-4.php');
3a66bed2	91	$ret = charset_decode_iso8859_4 ($string);
94965562	92	} else if ($res[1] == '5') {
3a66bed2	93	include_once(SM_PATH . 'functions/decode/iso8859-5.php');
3a66bed2	94	$ret = charset_decode_iso8859_5 ($string);
ef82d2d5	95	} else if ($res[1] == '6') {
5dd23dac	96	include_once(SM_PATH . 'functions/decode/iso8859-6.php');
5dd23dac	97	$ret = charset_decode_iso8859_6 ($string);
a2a7852b	98	} else if ($res[1] == '7') {
5dd23dac	99	include_once(SM_PATH . 'functions/decode/iso8859-7.php');
5dd23dac	100	$ret = charset_decode_iso8859_7 ($string);
3a66bed2	101	} else if ($res[1] == '8') {
	102	include_once(SM_PATH . 'functions/decode/iso8859-8.php');
	103	$ret = charset_decode_iso8859_8 ($string);
3ab35042	104	} else if ($res[1] == '9') {
5dd23dac	105	include_once(SM_PATH . 'functions/decode/iso8859-9.php');
5dd23dac	106	$ret = charset_decode_iso8859_9 ($string);
3a66bed2	107	} else if ($res[1] == '10') {
	108	include_once(SM_PATH . 'functions/decode/iso8859-10.php');
	109	$ret = charset_decode_iso8859_10 ($string);
	110	} else if ($res[1] == '11') {
	111	include_once(SM_PATH . 'functions/decode/iso8859-11.php');
	112	$ret = charset_decode_iso8859_11 ($string);
9be313d5	113	} else if ($res[1] == '13') {
3a66bed2	114	include_once(SM_PATH . 'functions/decode/iso8859-13.php');
	115	$ret = charset_decode_iso8859_13 ($string);
	116	} else if ($res[1] == '14') {
	117	include_once(SM_PATH . 'functions/decode/iso8859-14.php');
	118	$ret = charset_decode_iso8859_14 ($string);
a2a7852b	119	} else if ($res[1] == '15') {
5dd23dac	120	include_once(SM_PATH . 'functions/decode/iso8859-15.php');
5dd23dac	121	$ret = charset_decode_iso8859_15 ($string);
3a66bed2	122	} else if ($res[1] == '16') {
	123	include_once(SM_PATH . 'functions/decode/iso8859-16.php');
	124	$ret = charset_decode_iso8859_16 ($string);
a2a7852b	125	} else {
	126	$ret = charset_decode_iso_8859_default ($string);
	127	}
	128	} else if ($charset == 'ns_4551-1') {
	129	$ret = charset_decode_ns_4551_1 ($string);
	130	} else if ($charset == 'koi8-r') {
5dd23dac	131	include_once(SM_PATH . 'functions/decode/koi8-r.php');
a2a7852b	132	$ret = charset_decode_koi8r ($string);
1c0e847f	133	} else if ($charset == 'koi8-u') {
5dd23dac	134	include_once(SM_PATH . 'functions/decode/koi8-u.php');
1c0e847f	135	$ret = charset_decode_koi8u ($string);
5dd23dac	136	} else if ($charset == 'windows-1250') {
	137	include_once(SM_PATH . 'functions/decode/cp1250.php');
	138	$ret = charset_decode_cp1250 ($string);
a2a7852b	139	} else if ($charset == 'windows-1251') {
5dd23dac	140	include_once(SM_PATH . 'functions/decode/cp1251.php');
	141	$ret = charset_decode_cp1251 ($string);
	142	} else if ($charset == 'windows-1252') {
	143	include_once(SM_PATH . 'functions/decode/cp1252.php');
	144	$ret = charset_decode_cp1252 ($string);
3ab35042	145	} else if ($charset == 'windows-1253') {
5dd23dac	146	include_once(SM_PATH . 'functions/decode/cp1253.php');
5dd23dac	147	$ret = charset_decode_cp1253 ($string);
3ab35042	148	} else if ($charset == 'windows-1254') {
5dd23dac	149	include_once(SM_PATH . 'functions/decode/cp1254.php');
5dd23dac	150	$ret = charset_decode_cp1254 ($string);
c48a8ca7	151	} else if ($charset == 'windows-1255') {
5dd23dac	152	include_once(SM_PATH . 'functions/decode/cp1255.php');
5dd23dac	153	$ret = charset_decode_cp1255 ($string);
c48a8ca7	154	} else if ($charset == 'windows-1256') {
5dd23dac	155	include_once(SM_PATH . 'functions/decode/cp1256.php');
5dd23dac	156	$ret = charset_decode_cp1256 ($string);
c37a12f8	157	} else if ($charset == 'windows-1257') {
3a66bed2	158	include_once(SM_PATH . 'functions/decode/cp1257.php');
3a66bed2	159	$ret = charset_decode_cp1257 ($string);
5dd23dac	160	} else if ($charset == 'windows-1258') {
	161	include_once(SM_PATH . 'functions/decode/cp1258.php');
	162	$ret = charset_decode_cp1258 ($string);
e9a71964	163	} else if ($charset == 'x-mac-roman') {
	164	include_once(SM_PATH . 'functions/decode/cp10000.php');
	165	$ret = charset_decode_cp10000 ($string);
	166	} else if ($charset == 'x-mac-greek') {
	167	include_once(SM_PATH . 'functions/decode/cp10006.php');
	168	$ret = charset_decode_cp10006 ($string);
	169	} else if ($charset == 'x-mac-cyrillic') {
	170	include_once(SM_PATH . 'functions/decode/cp10007.php');
	171	$ret = charset_decode_cp10007 ($string);
	172	} else if ($charset == 'x-mac-ukrainian') {
	173	include_once(SM_PATH . 'functions/decode/cp10017.php');
	174	$ret = charset_decode_cp10017 ($string);
	175	} else if ($charset == 'x-mac-centraleurroman') {
	176	include_once(SM_PATH . 'functions/decode/cp10029.php');
	177	$ret = charset_decode_cp10029 ($string);
	178	} else if ($charset == 'x-mac-icelandic') {
	179	include_once(SM_PATH . 'functions/decode/cp10079.php');
	180	$ret = charset_decode_cp10079 ($string);
	181	} else if ($charset == 'x-mac-turkish') {
	182	include_once(SM_PATH . 'functions/decode/cp10081.php');
	183	$ret = charset_decode_cp10081 ($string);
	184	} else if ($charset == 'ibm855') {
	185	include_once(SM_PATH . 'functions/decode/cp855.php');
	186	$ret = charset_decode_cp855 ($string);
	187	} else if ($charset == 'ibm866') {
	188	include_once(SM_PATH . 'functions/decode/cp866.php');
	189	$ret = charset_decode_cp866 ($string);
0e4337e7	190	} else if ($charset == 'iso-ir-111') {
	191	include_once(SM_PATH . 'functions/decode/iso-ir-111.php');
	192	$ret = charset_decode_iso_ir_111 ($string);
7af26ef8	193	} else if ($charset == 'tis-620') {
	194	include_once(SM_PATH . 'functions/decode/tis620.php');
	195	$ret = charset_decode_tis620 ($string);
5dd23dac	196	} else if ($charset == 'big5' and $agresive_decoding ) {
	197	include_once(SM_PATH . 'functions/decode/big5.php');
	198	$ret = charset_decode_big5 ($string);
	199	} else if ($charset == 'gb2312' and $agresive_decoding ) {
	200	include_once(SM_PATH . 'functions/decode/gb2312.php');
	201	$ret = charset_decode_gb2312 ($string);
3ab35042	202	} else if ($charset == 'utf-8') {
5dd23dac	203	include_once(SM_PATH . 'functions/decode/utf-8.php');
3ab35042	204	$ret = charset_decode_utf8 ($string);
a2a7852b	205	} else {
	206	$ret = $string;
	207	}
	208	return( $ret );
	209	}
	210
a2a7852b	211
	212	/* Remove all 8 bit characters from all other ISO-8859 character sets */
	213	function charset_decode_iso_8859_default ($string) {
	214	return (strtr($string, "\240\241\242\243\244\245\246\247".
1fd97780	215	"\250\251\252\253\254\255\256\257".
	216	"\260\261\262\263\264\265\266\267".
	217	"\270\271\272\273\274\275\276\277".
	218	"\300\301\302\303\304\305\306\307".
	219	"\310\311\312\313\314\315\316\317".
	220	"\320\321\322\323\324\325\326\327".
	221	"\330\331\332\333\334\335\336\337".
	222	"\340\341\342\343\344\345\346\347".
	223	"\350\351\352\353\354\355\356\357".
	224	"\360\361\362\363\364\365\366\367".
a2a7852b	225	"\370\371\372\373\374\375\376\377",
1fd97780	226	"????????????????????????????????????????".
	227	"????????????????????????????????????????".
	228	"????????????????????????????????????????".
	229	"????????"));
a2a7852b	230
	231	}
	232
	233	/*
	234	* This is the same as ISO-646-NO and is used by some
	235	* Microsoft programs when sending Norwegian characters
	236	*/
	237	function charset_decode_ns_4551_1 ($string) {
	238	/*
	239	* These characters are:
	240	* Latin capital letter AE
	241	* Latin capital letter O with stroke
	242	* Latin capital letter A with ring above
	243	* and the same as small letters
	244	*/
	245