/**
* word segmentation
*
* @param string $string
* @param string $autoVariant
* @return string
*/
function normalizeForSearch($string, $autoVariant = 'gan-hans')
{
// LanguageZh::normalizeForSearch
return parent::normalizeForSearch($string, $autoVariant);
}
function stripForSearch($string)
{
wfProfileIn(__METHOD__);
// eventually this should be a word segmentation
// for now just treat each character as a word
// @fixme only do this for Han characters...
$t = preg_replace("/([\\xc0-\\xff][\\x80-\\xbf]*)/", " \$1", $string);
//always convert to gan-hans before indexing. it should be
//better to use gan-hans for search, since conversion from
//Traditional to Simplified is less ambiguous than the
//other way around
$t = $this->mConverter->autoConvert($t, 'gan-hans');
$t = parent::stripForSearch($t);
wfProfileOut(__METHOD__);
return $t;
}
