php权重计算方法代码分享

<?php
/* vim: set expandtab tabstop=4 shiftwidth=4: */
// +------------------------------------------------------------------------
//  name       :   权重计算                                        
//  description:   稍加修改，亦可用于分词，词频统计，全文检索和垃圾检测
//  date       :   2013/12/16 08:51

class weight {
    protected $adict = array(array());
    protected $aitems = array();
    protected $slastrule;
    protected $amatchs = array();
    protected $ashow = array();

 private function init() {
  //清空记录的匹配表和输出结果
  unset($this->ashow);
 }

    public function newitems($mitems) {
  //导入新的项目
  $this->aitems = (is_array($mitems))? $mitems: array($mitems);
  $this->init();
 }

 public function newtable(array $atable) {
        //导入新的对照表，并生成字典
        foreach($atable as $itablekey=>$stableline) {
            $atableline = explode(',', str_replace('|', ',', $stableline));
            $setter = function($v, $k, $parameter) {
                $k1 = $parameter[0]; $oweight = $parameter[1];
                $oweight->gendict($v, $k1);
            };
            array_walk($atableline, $setter, array($itablekey, $this));
        }
        $this->init();
 }

    public function getshow($srule = 'max') {
  //获取最终的显示结果
        if(empty($this->aitems) || empty($this->adict))
            return array();
  if (empty($this->ashow) || $srule != $this->slastrule)
            return $this->genshow($srule);
        return $this->ashow;
 }

    public function genshow($srule) {
        $ashow = array();
        $amatchs = array();
  $getter = function($v, $k, $oweight) use(&$ashow, &$amatchs, $srule) {
   $t = array_count_values($oweight->matchword($v));
            $amatchs[] = $t;
            switch ($srule) {
                case 'max':
                    $ashow[$k] = array_keys($t, max($t)); 
                    break;
            }
  };
  array_walk($this->aitems, $getter, $this);
  $this->ashow = $ashow;
  $this->amatchs = $amatchs;
  return $ashow;
    }

    private function gendict($mword, $ikey = '') {
        $iinsertponit = count($this->adict);
        $icur = 0; //当前节点号
        foreach (str_split($mword) as $ichar) {
            if (isset($this->adict[$icur][$ichar])) {
                $icur = $this->adict[$icur][$ichar];
                continue;
            }
            $this->adict[$iinsertponit] = array();
            $this->adict[$icur][$ichar] = $iinsertponit;
            $icur = $iinsertponit;
            $iinsertponit++;
        }
        $this->adict[$icur]['acc'][] = $ikey;

    }

        function matchword($sline) {
            $icur = $ioffset = $iposition = 0;
            $sline .= "\0";
            $ilen = strlen($sline);
            $areturn = array();
            while($ioffset < $ilen) {
                $schar = $sline{$ioffset};
                if(isset($this->adict[$icur][$schar])) {
                    $icur = $this->adict[$icur][$schar];
                    if(isset($this->adict[$icur]['acc'])) {
                        $areturn = array_merge($areturn, $this->adict[$icur]['acc']);

                        $iposition = $ioffset + 1;
                        $icur = 0;
                    }
                } else {
                    $icur = 0;
                    $ioffset = $iposition;
                    $iposition = $ioffset + 1;
                }
                ++$ioffset;
            }
            return $areturn;
        }
}

?>

$aitems = array(
    'chinaisbig',
    'whichisnot',
    'totalyrightforme',
);
$atable = array(
    'china,is|small',
    'china,big|me',
    'china,is|big,which|not,me',
    'totaly|right,for,me',
);

$oweight = new ttrie;
$oweight->newitems($aitems);
$aresult = $oweight->newtable($atable);