/**
* Decode a UTF-8 string to an entity encoded string
*
* @param string $string Encoded string
* @return string Decoded string
*/
protected static function _decodeUTF8($string)
{
// don't do decoding when there are no 8bit symbols
if (!QuickBooks_Cast::_is8Bit($string, 'utf-8')) {
return $string;
}
// decode four byte unicode characters
$string = preg_replace_callback("/([ð-÷])([€-¿])([€-¿])([€-¿])/", function ($arr) {
$val = (ord($arr[1]) - 240) * 262144 + (ord($arr[2]) - 128) * 4096 + (ord($arr[3]) - 128) * 64 + (ord($arr[4]) - 128);
return "&#" . $val . ";";
}, $string);
// decode three byte unicode characters
$string = preg_replace_callback("/([à-ï])([€-¿])([€-¿])/", function ($arr) {
$val = (ord($arr[1]) - 224) * 4096 + (ord($arr[2]) - 128) * 64 + (ord($arr[3]) - 128);
return "&#" . $val . ";";
}, $string);
// decode two byte unicode characters
$string = preg_replace_callback("/([À-ß])([€-¿])/", function ($arr) {
$val = (ord($arr[1]) - 192) * 64 + (ord($arr[2]) - 128);
return "&#" . $val . ";";
}, $string);
// remove broken unicode
$string = preg_replace("/[€-Ÿ]| |[¡-ÿ]/", '?', $string);
return $string;
}
