权重计算，稍加修改亦可用于分词，词频统计，全文和spam检测等

WBOY

发布： 2016-07-25 08:49:09

原创

940 人浏览过

效率非常客观，你要是改成其他用处那效率我就不保证了

/* 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
// Authors : latel
// +------------------------------------------------------------------------
//
/*外部调用示例*/
/*
$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);
*/
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 $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;
}
}
?>

复制代码