BuildExtendedChar.php

var $UTF8SpecialChars = array(
  "|([\xC2-\xDF])([\x80-\xBF])|e",
  "|(\xE0)([\xA0-\xBF])([\x80-\xBF])|e",
  "|([\xE1-\xEF])([\x80-\xBF])([\x80-\xBF])|e",
  "|(\xF0)([\x90-\xBF])([\x80-\xBF])([\x80-\xBF])|e",
  "|([\xF1-\xF3])([\x80-\xBF])([\x80-\xBF])([\x80-\xBF])|e",
  "|(\xF4)([\x80-\x8F])([\x80-\xBF])([\x80-\xBF])|e"
);

var $UTF8HTMLEntities = array(
  "\$this->BuildExtendedChar('\\1','\\2')",
  "\$this->BuildExtendedChar('\\1','\\2','\\3')",
  "\$this->BuildExtendedChar('\\1','\\2','\\3')",
  "\$this->BuildExtendedChar('\\1','\\2','\\3','\\4')",
  "\$this->BuildExtendedChar('\\1','\\2','\\3','\\4')",
  "\$this->BuildExtendedChar('\\1','\\2','\\3','\\4')"
);

function BuildExtendedChar( $byteOne, $byteTwo = "\x00", $byteThree = "\x00", $byteFour = "\x00" ) {
  if( ord( $byteTwo ) >= 128 ) {
    $tempChar = substr(decbin(ord($byteTwo)), -6);
    if (ord($byteThree) >= 128) {
      $tempChar .= substr(decbin(ord($byteThree)), -6);
      if (ord($byteFour) >= 128) {
        $tempChar .= substr(decbin(ord($byteFour)), -6);
        $tempChar = substr(decbin(ord($byteOne)), -3) . $tempChar;
      } else {
        $tempChar = substr(decbin(ord($byteOne)), -4) . $tempChar;
      }
    } else {
      $tempChar = substr(decbin(ord($byteOne)), -5) . $tempChar;
    }
  } else $tempChar = $byteOne;
  $tempChar = '&#' . bindec($tempChar) . ';';
  return $tempChar;
}