[squirrelmail.git] / functions / encode / koi8_u.php

<?php

/**
 * koi8-u encoding functions
 *
 * takes a string of unicode entities and converts it to a koi8-u encoded string
 * Unsupported characters are replaced with ?.
 *
 * @copyright 2004-2023 The SquirrelMail Project Team
 * @license http://opensource.org/licenses/gpl-license.php GNU Public License
 * @version $Id$
 * @package squirrelmail
 * @subpackage encode
 */

/**
 * Converts string to koi8-u
 * @param string $string text with numeric unicode entities
 * @return string koi8-u encoded text
 */
function charset_encode_koi8_u ($string) {
   // don't run encoding function, if there is no encoded characters
   if (! preg_match("'&#[0-9]+;'",$string) ) return $string;

    $string=preg_replace_callback("/&#([0-9]+);/",'unicodetokoi8u',$string);

    return $string;
}

/**
 * Return koi8-u symbol when unicode character number is provided
 *
 * This function is used internally by charset_encode_koi8_u
 * function. It might be unavailable to other SquirrelMail functions.
 * Don't use it or make sure, that functions/encode/koi8_u.php is
 * included.
 *
 * @param array $matches array with first element a decimal unicode value
 * @return string koi8-u character
 */
function unicodetokoi8u($matches) {
    $var = $matches[1];

    $koi8uchars=array('160' => "\x9A",
                      '169' => "\xBF",
                      '176' => "\x9C",
                      '178' => "\x9D",
                      '183' => "\x9E",
                      '247' => "\x9F",
                      '1025' => "\xB3",
                      '1028' => "\xB4",
                      '1030' => "\xB6",
                      '1031' => "\xB7",
                      '1040' => "\xE1",
                      '1041' => "\xE2",
                      '1042' => "\xF7",
                      '1043' => "\xE7",
                      '1044' => "\xE4",
                      '1045' => "\xE5",
                      '1046' => "\xF6",
                      '1047' => "\xFA",
                      '1048' => "\xE9",
                      '1049' => "\xEA",
                      '1050' => "\xEB",
                      '1051' => "\xEC",
                      '1052' => "\xED",
                      '1053' => "\xEE",
                      '1054' => "\xEF",
                      '1055' => "\xF0",
                      '1056' => "\xF2",
                      '1057' => "\xF3",
                      '1058' => "\xF4",
                      '1059' => "\xF5",
                      '1060' => "\xE6",
                      '1061' => "\xE8",
                      '1062' => "\xE3",
                      '1063' => "\xFE",
                      '1064' => "\xFB",
                      '1065' => "\xFD",
                      '1066' => "\xFF",
                      '1067' => "\xF9",
                      '1068' => "\xF8",
                      '1069' => "\xFC",
                      '1070' => "\xE0",
                      '1071' => "\xF1",
                      '1072' => "\xC1",
                      '1073' => "\xC2",
                      '1074' => "\xD7",
                      '1075' => "\xC7",
                      '1076' => "\xC4",
                      '1077' => "\xC5",
                      '1078' => "\xD6",
                      '1079' => "\xDA",
                      '1080' => "\xC9",
                      '1081' => "\xCA",
                      '1082' => "\xCB",
                      '1083' => "\xCC",
                      '1084' => "\xCD",
                      '1085' => "\xCE",
                      '1086' => "\xCF",
                      '1087' => "\xD0",
                      '1088' => "\xD2",
                      '1089' => "\xD3",
                      '1090' => "\xD4",
                      '1091' => "\xD5",
                      '1092' => "\xC6",
                      '1093' => "\xC8",
                      '1094' => "\xC3",
                      '1095' => "\xDE",
                      '1096' => "\xDB",
                      '1097' => "\xDD",
                      '1098' => "\xDF",
                      '1099' => "\xD9",
                      '1100' => "\xD8",
                      '1101' => "\xDC",
                      '1102' => "\xC0",
                      '1103' => "\xD1",
                      '1105' => "\xA3",
                      '1108' => "\xA4",
                      '1110' => "\xA6",
                      '1111' => "\xA7",
                      '1168' => "\xBD",
                      '1169' => "\xAD",
                      '8729' => "\x95",
                      '8730' => "\x96",
                      '8776' => "\x97",
                      '8804' => "\x98",
                      '8805' => "\x99",
                      '8992' => "\x93",
                      '8993' => "\x9B",
                      '9472' => "\x80",
                      '9474' => "\x81",
                      '9484' => "\x82",
                      '9488' => "\x83",
                      '9492' => "\x84",
                      '9496' => "\x85",
                      '9500' => "\x86",
                      '9508' => "\x87",
                      '9516' => "\x88",
                      '9524' => "\x89",
                      '9532' => "\x8A",
                      '9552' => "\xA0",
                      '9553' => "\xA1",
                      '9554' => "\xA2",
                      '9556' => "\xA5",
                      '9559' => "\xA8",
                      '9560' => "\xA9",
                      '9561' => "\xAA",
                      '9562' => "\xAB",
                      '9563' => "\xAC",
                      '9565' => "\xAE",
                      '9566' => "\xAF",
                      '9567' => "\xB0",
                      '9568' => "\xB1",
                      '9569' => "\xB2",
                      '9571' => "\xB5",
                      '9574' => "\xB8",
                      '9575' => "\xB9",
                      '9576' => "\xBA",
                      '9577' => "\xBB",
                      '9578' => "\xBC",
                      '9580' => "\xBE",
                      '9600' => "\x8B",
                      '9604' => "\x8C",
                      '9608' => "\x8D",
                      '9612' => "\x8E",
                      '9616' => "\x8F",
                      '9617' => "\x90",
                      '9618' => "\x91",
                      '9619' => "\x92",
                      '9632' => "\x94");

    if (array_key_exists($var,$koi8uchars)) {
        $ret=$koi8uchars[$var];
    } else {
        $ret='?';
    }
    return $ret;
}
Commit	Line	Data
9911dc64	1	<?php
4b4abf93	2
9911dc64	3	/**
	4	* koi8-u encoding functions
	5	*
	6	* takes a string of unicode entities and converts it to a koi8-u encoded string
	7	* Unsupported characters are replaced with ?.
	8	*
4b28020b	9	* @copyright 2004-2023 The SquirrelMail Project Team
4b4abf93	10	* @license http://opensource.org/licenses/gpl-license.php GNU Public License
9911dc64	11	* @version $Id$
9911dc64	12	* @package squirrelmail
	13	* @subpackage encode
	14	*/
	15
	16	/**
	17	* Converts string to koi8-u
	18	* @param string $string text with numeric unicode entities
	19	* @return string koi8-u encoded text
	20	*/
	21	function charset_encode_koi8_u ($string) {
	22	// don't run encoding function, if there is no encoded characters
	23	if (! preg_match("'&#[0-9]+;'",$string) ) return $string;
	24
6c4e2e9b	25	$string=preg_replace_callback("/&#([0-9]+);/",'unicodetokoi8u',$string);
91e0dccc	26
9911dc64	27	return $string;
	28	}
	29
	30	/**
	31	* Return koi8-u symbol when unicode character number is provided
91e0dccc	32	*
91e0dccc	33	* This function is used internally by charset_encode_koi8_u
598294a7	34	* function. It might be unavailable to other SquirrelMail functions.
91e0dccc	35	* Don't use it or make sure, that functions/encode/koi8_u.php is
91e0dccc	36	* included.
9911dc64	37	*
6c4e2e9b	38	* @param array $matches array with first element a decimal unicode value
9911dc64	39	* @return string koi8-u character
9911dc64	40	*/
6c4e2e9b	41	function unicodetokoi8u($matches) {
6c4e2e9b	42	$var = $matches[1];
9911dc64	43
	44	$koi8uchars=array('160' => "\x9A",
	45	'169' => "\xBF",
	46	'176' => "\x9C",
	47	'178' => "\x9D",
	48	'183' => "\x9E",
	49	'247' => "\x9F",
	50	'1025' => "\xB3",
	51	'1028' => "\xB4",
	52	'1030' => "\xB6",
	53	'1031' => "\xB7",
	54	'1040' => "\xE1",
	55	'1041' => "\xE2",
	56	'1042' => "\xF7",
	57	'1043' => "\xE7",
	58	'1044' => "\xE4",
	59	'1045' => "\xE5",
	60	'1046' => "\xF6",
	61	'1047' => "\xFA",
	62	'1048' => "\xE9",
	63	'1049' => "\xEA",
	64	'1050' => "\xEB",
	65	'1051' => "\xEC",
	66	'1052' => "\xED",
	67	'1053' => "\xEE",
	68	'1054' => "\xEF",
	69	'1055' => "\xF0",
	70	'1056' => "\xF2",
	71	'1057' => "\xF3",
	72	'1058' => "\xF4",
	73	'1059' => "\xF5",
	74	'1060' => "\xE6",
	75	'1061' => "\xE8",
	76	'1062' => "\xE3",
	77	'1063' => "\xFE",
	78	'1064' => "\xFB",
	79	'1065' => "\xFD",
	80	'1066' => "\xFF",
	81	'1067' => "\xF9",
	82	'1068' => "\xF8",
	83	'1069' => "\xFC",
	84	'1070' => "\xE0",
	85	'1071' => "\xF1",
	86	'1072' => "\xC1",
	87	'1073' => "\xC2",
	88	'1074' => "\xD7",
	89	'1075' => "\xC7",
	90	'1076' => "\xC4",
	91	'1077' => "\xC5",
	92	'1078' => "\xD6",
	93	'1079' => "\xDA",
	94	'1080' => "\xC9",
	95	'1081' => "\xCA",
	96	'1082' => "\xCB",
	97	'1083' => "\xCC",
	98	'1084' => "\xCD",
	99	'1085' => "\xCE",
	100	'1086' => "\xCF",
	101	'1087' => "\xD0",
	102	'1088' => "\xD2",
	103	'1089' => "\xD3",
	104	'1090' => "\xD4",
	105	'1091' => "\xD5",
	106	'1092' => "\xC6",
107	'1093' => "\xC8",
108	'1094' => "\xC3",
109	'1095' => "\xDE",
110	'1096' => "\xDB",
111	'1097' => "\xDD",
112	'1098' => "\xDF",
113	'1099' => "\xD9",
114	'1100' => "\xD8",
115	'1101' => "\xDC",
116	'1102' => "\xC0",
117	'1103' => "\xD1",
118	'1105' => "\xA3",
119	'1108' => "\xA4",
120	'1110' => "\xA6",
121	'1111' => "\xA7",
122	'1168' => "\xBD",
123	'1169' => "\xAD",
124	'8729' => "\x95",
125	'8730' => "\x96",
126	'8776' => "\x97",
127	'8804' => "\x98",
128	'8805' => "\x99",
129	'8992' => "\x93",
130	'8993' => "\x9B",
131	'9472' => "\x80",
132	'9474' => "\x81",
133	'9484' => "\x82",
134	'9488' => "\x83",
135	'9492' => "\x84",
136	'9496' => "\x85",
137	'9500' => "\x86",
138	'9508' => "\x87",
139	'9516' => "\x88",
140	'9524' => "\x89",
141	'9532' => "\x8A",
142	'9552' => "\xA0",
143	'9553' => "\xA1",
144	'9554' => "\xA2",
145	'9556' => "\xA5",
146	'9559' => "\xA8",
147	'9560' => "\xA9",
148	'9561' => "\xAA",
149	'9562' => "\xAB",
150	'9563' => "\xAC",
151	'9565' => "\xAE",
152	'9566' => "\xAF",
153	'9567' => "\xB0",
154	'9568' => "\xB1",
155	'9569' => "\xB2",
156	'9571' => "\xB5",
157	'9574' => "\xB8",
158	'9575' => "\xB9",
159	'9576' => "\xBA",
160	'9577' => "\xBB",
161	'9578' => "\xBC",
162	'9580' => "\xBE",
163	'9600' => "\x8B",
164	'9604' => "\x8C",
165	'9608' => "\x8D",
166	'9612' => "\x8E",
167	'9616' => "\x8F",
168	'9617' => "\x90",
169	'9618' => "\x91",
170	'9619' => "\x92",
171	'9632' => "\x94");
172
173	if (array_key_exists($var,$koi8uchars)) {
174	$ret=$koi8uchars[$var];
175	} else {
176	$ret='?';
177	}
178	return $ret;
179	}