除去字串中的重复词,生成索引字符串

作者:袖梨 2022-07-02

  //--------------------------------------------------
  function GetIndexText($okstr,$ilen=-1)
  {
    if($okstr=="") return "";
    $ws = explode(" ",$okstr);
    $okstr = "";
    $wks = "";
    foreach($ws as $w)
    {
      $w = trim($w);
      //排除小于2的字符
      if(strlen($w)<2) continue;
      //排除数字或日期
      if(!ereg("[^0-9:-]",$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);
  }

相关文章

精彩推荐