githubFork/roundcubemail.git

alecpl

2010-09-25 e019f2d0f2dc2fbfa345ab5d7ae85e67bfdd76b8

commit \| author \| age
bdc07a	1	<?php
83dbb7	2	/*
T	3	utf8 1.0
	4	Copyright: Left
	5	---------------------------------------------------------------------------------
	6	Version: 1.0
	7	Date: 23 November 2004
	8	---------------------------------------------------------------------------------
	9	Author: Alexander Minkovsky (a_minkovsky@hotmail.com)
	10	---------------------------------------------------------------------------------
	11	License: Choose the more appropriated for You - I don't care.
	12	---------------------------------------------------------------------------------
	13	Description:
	14	Class provides functionality to convert single byte strings, such as CP1251
	15	ti UTF-8 multibyte format and vice versa.
	16	Class loads a concrete charset map, for example CP1251.
	17	(Refer to ftp://ftp.unicode.org/Public/MAPPINGS/ for map files)
	18	Directory containing MAP files is predefined as constant.
	19	Each charset is also predefined as constant pointing to the MAP file.
	20	---------------------------------------------------------------------------------
	21	Example usage:
	22	Pass the desired charset in the class constructor:
	23	$utfConverter = new utf8(CP1251); //defaults to CP1250.
	24	or load the charset MAP using loadCharset method like this:
	25	$utfConverter->loadCharset(CP1252);
	26	Then call
	27	$res = $utfConverter->strToUtf8($str);
	28	or
	29	$res = $utfConverter->utf8ToStr($utf);
	30	to get the needed encoding.
	31	---------------------------------------------------------------------------------
	32	Note:
	33	Rewrite or Override the onError method if needed. It's the error handler used from everywhere and takes 2 parameters:
	34	err_code and err_text. By default it just prints out a message about the error.
	35	*/
	36
	37	// Charset maps
e019f2	38	// Adapted to fit Roundcube
83dbb7	39	define("UTF8_MAP_DIR", "program/lib/encoding");
T	40
	41	//Error constants
ce72e0	42	define("ERR_OPEN_MAP_FILE", "ERR_OPEN_MAP_FILE");
83dbb7	43
T	44	//Class definition
ce72e0	45	Class utf8 {
83dbb7	46
58e360	47	var $charset = "ISO-8859-1";
83dbb7	48	var $ascMap = array();
T	49	var $utfMap = array();
ce72e0	50	var $aliases = array(
A	51	'KOI8-R' => 'KOI8R'
	52	);
	53	var $error = null;
58e360	54
ce72e0	55	function __construct($charset="ISO-8859-1") {
83dbb7	56	$this->loadCharset($charset);
T	57	}
	58
	59	//Load charset
ce72e0	60	function loadCharset($charset) {
83dbb7	61
ce72e0	62	$charset = preg_replace(array('/^WINDOWS-*125([0-8])$/', '/^CP-/'), array('CP125\\1', 'CP'), $charset);
A	63	if (isset($aliases[$charset]))
	64	$charset = $aliases[$charset];
	65
	66	$this->charset = $charset;
	67
83dbb7	68	if (empty($this->ascMap[$charset]))
T	69	{
ce72e0	70	$file = UTF8_MAP_DIR.'/'.$charset.'.map';
A	71
	72	if (!is_file($file)) {
	73	$this->onError(ERR_OPEN_MAP_FILE, "Failed to open map file for $charset");
	74	return;
	75	}
	76
	77	$lines = file_get_contents($file);
83dbb7	78	$lines = preg_replace("/#.*$/m","",$lines);
T	79	$lines = preg_replace("/\n\n/","",$lines);
	80	$lines = explode("\n",$lines);
ce72e0	81
83dbb7	82	foreach($lines as $line){
T	83	$parts = explode('0x',$line);
	84	if(count($parts)==3){
	85	$asc=hexdec(substr($parts[1],0,2));
	86	$utf=hexdec(substr($parts[2],0,4));
	87	$this->ascMap[$charset][$asc]=$utf;
	88	}
	89	}
ce72e0	90
A	91	$this->utfMap = array_flip($this->ascMap[$charset]);
83dbb7	92	}
T	93	}
	94
	95	//Error handler
	96	function onError($err_code,$err_text){
ce72e0	97	$this->error = $err_text;
A	98	return null;
83dbb7	99	}
T	100
	101	//Translate string ($str) to UTF-8 from given charset
	102	function strToUtf8($str){
ce72e0	103	if (empty($this->ascMap[$this->charset]))
A	104	return null;
	105
83dbb7	106	$chars = unpack('C*', $str);
T	107	$cnt = count($chars);
ce72e0	108	for($i=1; $i<=$cnt; $i++)
A	109	$this->_charToUtf8($chars[$i]);
	110
83dbb7	111	return implode("",$chars);
T	112	}
	113
	114	//Translate UTF-8 string to single byte string in the given charset
	115	function utf8ToStr($utf){
ce72e0	116	if (empty($this->ascMap[$this->charset]))
A	117	return null;
	118
83dbb7	119	$chars = unpack('C*', $utf);
T	120	$cnt = count($chars);
	121	$res = ""; //No simple way to do it in place... concatenate char by char
ce72e0	122
A	123	for ($i=1; $i<=$cnt; $i++)
83dbb7	124	$res .= $this->_utf8ToChar($chars, $i);
ce72e0	125
83dbb7	126	return $res;
T	127	}
	128
	129	//Char to UTF-8 sequence
	130	function _charToUtf8(&$char){
	131	$c = (int)$this->ascMap[$this->charset][$char];
	132	if ($c < 0x80){
	133	$char = chr($c);
	134	}
	135	else if($c<0x800) // 2 bytes
	136	$char = (chr(0xC0 \| $c>>6) . chr(0x80 \| $c & 0x3F));
	137	else if($c<0x10000) // 3 bytes
	138	$char = (chr(0xE0 \| $c>>12) . chr(0x80 \| $c>>6 & 0x3F) . chr(0x80 \| $c & 0x3F));
	139	else if($c<0x200000) // 4 bytes
	140	$char = (chr(0xF0 \| $c>>18) . chr(0x80 \| $c>>12 & 0x3F) . chr(0x80 \| $c>>6 & 0x3F) . chr(0x80 \| $c & 0x3F));
	141	}
	142
	143	//UTF-8 sequence to single byte character
	144	function _utf8ToChar(&$chars, &$idx){
	145	if(($chars[$idx] >= 240) && ($chars[$idx] <= 255)){ // 4 bytes
	146	$utf = (intval($chars[$idx]-240) << 18) +
	147	(intval($chars[++$idx]-128) << 12) +
	148	(intval($chars[++$idx]-128) << 6) +
	149	(intval($chars[++$idx]-128) << 0);
	150	}
	151	else if (($chars[$idx] >= 224) && ($chars[$idx] <= 239)){ // 3 bytes
	152	$utf = (intval($chars[$idx]-224) << 12) +
	153	(intval($chars[++$idx]-128) << 6) +
	154	(intval($chars[++$idx]-128) << 0);
	155	}
	156	else if (($chars[$idx] >= 192) && ($chars[$idx] <= 223)){ // 2 bytes
	157	$utf = (intval($chars[$idx]-192) << 6) +
	158	(intval($chars[++$idx]-128) << 0);
	159	}
	160	else{ // 1 byte
	161	$utf = $chars[$idx];
	162	}
	163	if(array_key_exists($utf,$this->utfMap))
	164	return chr($this->utfMap[$utf]);
	165	else
	166	return "?";
	167	}
	168
	169	}
58e360	170
6710a6	171	?>