function mb_word_split($str, $lang)
{
	$m = array();

	switch ($lang) {
		case 'chinese_big5':
			preg_match_all('!((?:[A-Za-z]+) | (?:[\xa1-\xfe] [\x40-\x7e] | [\xa1-\xfe] ) )!xs', $str, $m);
			break;
		case 'chinese': /* bg2312 */
			preg_match_all('!((?:[A-Za-z]+) | (?:[\xa1-\xf7] [\xa1-\xfe] ) | ( [\xa1-\xfe] [\x40-\x7e] | [\xa1-\xfe] ) )!xs', $str, $m);
			break;
		case 'japanese': /* utf-8 */
		case 'korean':
			preg_match_all('!((?:[\x0-\x7f]+) | (?:[\xc0-\xfd]{1}[\x80-\xbf]+) )!xs', $str, $m);
			break;
	}

	if (!$m) {
		return array();
	}

	$m2 = array();
	$tmp="";
	foreach (array_unique($m[0]) as $v) {
		if (isset($v[1])) 
		{
			if(preg_match('!(?:[A-Za-z]+)!xs',$v))
			{
				$m2[] = _esc($v);
			}
			else
			{
				if($tmp!="")
				{
					$m2[] = _esc($tmp.$v);
				}
				$tmp=$v;
			}
		}
	}
	return $m2;
}