/除去字串中的重復詞,生成索引字符串,字符串參數為已經分詞處理的串
//--------------------------------------------------
function getindextext($okstr,$ilen=-1)
{
if($okstr=="") return "";
$ws = explode(" ",$okstr);
$okstr = "";
$wks = "";
foreach($ws as $w)
{
$w = trim($w);
//排除小於2的字符
if(strlen($w)<2) continue;
//排除數字或日期
if(!ereg("[^0-9:-]",$w)) continue;
if(strlen($w)==2&&ord($w[0])>0x80) continue;
if(isset($wks[$w])) $wks[$w]++;
else $wks[$w] = 1;
}
if(is_array($wks))
{
arsort($wks);
if($ilen==-1)
{ foreach($wks as $w=>$v) $okstr .= $w." "; }
else
{
foreach($wks as $w=>$v){
if((strlen($okstr)+strlen($w)+1)<$ilen) $okstr .= $w." ";
else break;
}
}
}
return trim($okstr);
}?>