<?php
require_once dirname(__FILE__).'/config/config.php';
class Util_Keyword
{
	/**
	 * 编译后的字典数组
	 * 
	 * @var string
	 */
	private $mDictionaryWords = array();
	
	/**
	 * 特殊字符
	 * @var string
	 */
	private $mSpecialCharacters = array(); 
	
	/**
	 * 初始化
	 */
	public function __construct()
	{
		$this->mSpecialCharacters = array(
									0 => ' ',
									1 => '-',
									2 => '*',
									3 => '',
				);
	}
	
	/**
	 * 替换
	 * 
	 * @param string $article
	 * @return array
	 */
	public function replace($article)
	{
		$this->loadwords();
		$len = strlen($article);
		for($i=0; $i<$len; $i++)
		{
			$node = $this->searchKeyword($article, $this->mDictionaryWords, $i);
			if(is_array($node))
			{
				//禁止词
				if($node['value']== PROHIBITION_STATE)
				{
					//continue;
				}
				//替换词
				if($node['value']== REPLACE_STATE)
				{
					//continue;
				}				
				//审核词
				if($node['value']== CHECK_STATE)
				{
					continue;
				}
				if($node['value']!= CHECK_STATE)
				{
                	$this->convert($article, $i, $node['degree']);
				}	
				//替换词的最后位置
				$i--;	
			}
		}
		return $article;
	}
	
	/**
	 * 包含那种类型词
	 * 
	 * @param string $article
	 * @param array $nodeValue
	 * @return boolean
	 */
	public function comprise($article, $nodeValue = array(1,2,3))
	{
		$this->loadwords();
		$len = strlen($article);
		for($i=0; $i<$len; $i++)
		{
			$node = $this->searchKeyword($article, $this->mDictionaryWords, $i);
			if(is_array($node))
			{
				if(array_search($node['value'], $nodeValue)!==FALSE)
				{
					return true;
				}
				//替换词的最后位置
				$i = $i+$node['degree']-1;	
			}	
		}
		//没有那种类型词
		return false;
	}
	
	/**
	 * 全相同
	 * 
	 * @param string $article
	 * @param array $nodeValue
	 * @return boolean
	 */
	public function allSame($article, $nodeValue = array(1,2,3))
	{
		$this->loadwords();
		$len = strlen($article);
		$node = array(
					'degree'	=> -1,
					'value'		=> 0,	
				);
		for($i=0; $i<$len; $i++)
		{
			$node = $this->searchKeyword($article, $this->mDictionaryWords, $i, 1,false);
			if(is_array($node))
			{
				if(array_search($node['value'], $nodeValue)!==FALSE)
				{
					break;
				}
			}	
		}
		//长度和文章长度相等
		if($len == $node['degree'])
		{
			return true;
		}
		else
		{
			return false;
		}
	}
	
	/**
	 * 搜索关键字
	 * 
	 * @param string $content
	 * @param array $keywords
	 * @param int $pos
	 * @param int $degree
	 * @param boolean $semiMatch
	 * @return array
	 */
	public function searchKeyword($content, $keywords, $pos, $degree=1, $semiMatch=true)
	{
		/*	//过滤特殊字符
		if(array_search($content[$pos], $this->mSpecialCharacters)!==false) 
		{
			$this->searchKeyword($content,$keywords,$pos+1, $degree);
		}*/
		//节点
		if($pos<strlen($content))
		{
			$asc = ord($content{$pos});
		}
		else
		{
			return false;
		}
		if($degree>20)
		{
			return false;
		}
		if(!is_array($keywords) || (count($keywords)==0))
		{
		    return false;
		}
		if(!array_key_exists($asc,$keywords))
		{
			return false;
		}
		else
		{
			if(array_key_exists('val',$keywords[$asc]))
			{
				//半匹配模式
				if(intval($keywords[$asc]['val']) && $semiMatch) 
				{
					return array(
					'degree' => $degree,
					'value'  => $keywords[$asc]['val'],
					);
				}
				else if((strlen($content)==$degree) && intval($keywords[$asc]['val'])) //全匹配模式
				{
					return array(
					'degree' => $degree,
					'value'  => $keywords[$asc]['val'],
					);
				}
			}	
		}	
		if(empty($keywords[$asc]))  
		{	
			return false;
		}
		//深度,累加
		$degree++;
		//下一个节点
		$keywords = $keywords[$asc];
		return $this->searchKeyword($content, $keywords, $pos+1, $degree, $semiMatch);
	}
	
	/**
	 * 生成字典词语
	 * 
	 * @param array $compileWords
	 * @param int $nodeValue
	 * @param string $dictionaryCompileFile
	 * @param string $dictionaryFile
	 * @param int $nodeValue
	 */
	public function compileDictionaryWord($compileWords, $nodeValue, $dictionaryCompileFile = COMPILEFILE, $dictionaryFile = DICTIONARYFILE)
	{
		require_once $dictionaryFile;
		$word = $dictionaryWords;
		$keywords = array(
							$nodeValue => $compileWords,
						);
		$dictionaryWords = $this->compileWord($keywords, $word);
		$content = "<?php \r\n \$dictionaryWords = ".var_export($dictionaryWords,1).";";
		//重新生成字典
		$this->writeOver($dictionaryFile, $content);
		require_once $dictionaryCompileFile;
		foreach($compileWords as $word)
		{
			$words[$nodeValue][] = $word;	
		}
		$content = "<?php \r\n \$words = ".var_export($words,1).";";
		//重新生成编译字典
		$this->writeOver($dictionaryCompileFile, $content);
		$this->mDictionaryWords = $dictionaryWords;
		unset($dictionaryWords, $keywords, $content,$words);
	}
	
	/**
	 * 写到字典文件
	 * 
	 * @param array $dictionaryWords
	 * @param string $dictionaryFile
	 * @return boolean
	 */
	public function writeDictionaryFile($dictionaryWords = array(), $dictionaryFile = DICTIONARYFILE)
	{
		if(!file_exists($dictionaryFile))
		{
			return false;
		}
		if(count($dictionaryWords))
		{
			$content = "<?php \r\n \$dictionaryWords = ".var_export($dictionaryWords,1).";";
			$this->writeOver($dictionaryFile, $content);
			return true;
		}
		else
		{
			return false;
		}
	}
	
	/**
	 * 根据字典词语生成字典
	 * 
	 * @param string $dictionaryCompileFile
	 * @param string $dictionaryFile
	 * @return int
	 */
	public function compileDictionaryByCompileFile($dictionaryCompileFile = COMPILEFILE, $dictionaryFile = DICTIONARYFILE)
	{
		if(!file_exists($dictionaryCompileFile))
		{
			return false;
		}
		require_once $dictionaryCompileFile;
		$keywords = $words;			
		$dictionaryWords = $this->compileWord($keywords);
		$content = "<?php \r\n \$dictionaryWords = ".var_export($dictionaryWords,1).";";
		$this->writeOver($dictionaryFile, $content);
		return $dictionaryWords;
	}
	
	/**
	 * 导入文本字典
	 * 
	 * @param string $dictionaryTxtFile
	 * @param string $dictionaryCompileFile
	 * @return boolean
	 */
	public function importTxtDictionary($dictionaryTxtFile = TXTFILE, $dictionaryCompileFile = COMPILEFILE)
	{
		if (file_exists($dictionaryTxtFile))
		{
			$words = array();
			$content = explode("\n",$this->readOver($dictionaryTxtFile));
			foreach($content as $key => $value)
			{			
				if($value)
				{					
					$word = trim(substr($value, 0, strpos($value,'|')));			
					$type = trim(substr(strrchr($value,'|'), 1));
					$type = intval($type);
					if(is_numeric($type))
					{
						$words[$type][] = $word;
					}
					else
					{
						continue;
					}	
    			}
			}
			$content = "<?php \r\n \$words = ".var_export($words,1).";";
			$this->writeOver($dictionaryCompileFile,$content);
			return true;
		}
		else
		{
			die('Error!');
			return false;
		}
	}	
	
	/**
	 * 设置字典数组
	 * 
	 * @param array $words
	 */
	public function setDictionaryWords($words)
	{
		if(is_array($words))
		{
			$this->mDictionaryWords = $words;
			unset($words);
		}
	}
	
	/**
	 * 生成词库
	 * 
	 * @param array $keywords
	 * @param array $word
	 * @return array
	 */
	public function compileWord($keywords, $word=array())
	{
		foreach($keywords as $key => $words)
		{
			foreach ($words as $keyword)
			{
				$temp = '$word';
				for($i=0;$i<strlen($keyword);$i++)
				{
					$asc = ord($keyword{$i});
					$temp.= "['".$asc."']";
				}
				eval($temp.="['val']=".$key.";");
			}
		}	
		return $word;
	}
	
	private function loadwords()
	{
		//如果有值，就不加载
		if(count($this->mDictionaryWords))
		{
			return;
		}
		$dictionaryFile = DICTIONARYFILE;
		//加载字典
		if(file_exists($dictionaryFile))
		{
			require_once $dictionaryFile;
			$this->mDictionaryWords = $dictionaryWords;
		}
		else
		{
			//编译字典
			$dictionaryCompileFile = COMPILEFILE;
			$this->mDictionaryWords = $this->compileDictionary($dictionaryCompileFile, $dictionaryFile);
			if(!is_array($this->mDictionaryWords))
			{
				$dictionaryTxtFile = TXTFILE;
				$this->importTxtDictionary($dictionaryTxtFile, $dictionaryCompileFile);
				$this->mDictionaryWords = $this->compileDictionary($dictionaryCompileFile, $dictionaryFile);
			}
		}	
	}
	/**
	 * 读取内容
	 * 
	 * @param string $fileName
	 * @param string $method
	 * @return string
	 */
	public function readOver($fileName,$method="rb")
	{  
    	//由于strpos函数返回是整数值，所以使用!==（不全等）符号判断是否$filename是否包含..，包含则退出显示 
    	strpos($fileName,'..')!==false && exit();  
    	if($handle=@fopen($fileName,$method))
    	{ 
        	//轻便的咨询文件锁定，设置共享锁定   
       	 	flock($handle,LOCK_SH);
       	 	//读取所有内容
        	$filedata=@fread($handle,filesize($fileName));
        	fclose($handle); 
   		 }  
    	//返回文件内容  
   	 	return $filedata;   
	}
	
	/**
	 * 转化
	 * 
	 * @param string $article
	 * @param int $pos
	 * @param int $len
	 * @return string 
	 */
	private function convert(&$article, &$pos, $len)
	{
		$temp = $pos+$len;
		for($len;$len>0;$len--)
		{
			//判断是否是汉字
			if((ord($article{$pos})>127) && (ord($article{$pos+1})>127) &&
				(ord($article{$pos+2})>127) && ($pos+2<=$temp))
			{
				$article = substr($article,0, $pos).'*'.substr($article, $pos+3);
				$len-=2;
			}
			else
			{
				$article{$pos} = '*';
			}	
			$pos++;
		}
		unset($temp, $len);
	}
	
	/**
	 * 写入内容
	 * 
	 * @param string $fileName
	 * @param string $method
	 * @return boolean
	 */
	public function writeOver($fileName, $content, $method="wb")
	{  
    	//由于strpos函数返回是整数值，所以使用!==（不全等）符号判断是否$filename是否包含..，包含则退出显示 
    	strpos($fileName,'..')!==false && exit();  
    	if($handle=@fopen($fileName,$method))
    	{ 
        	//轻便的咨询文件锁定，设置独占锁定   
       	 	flock($handle,LOCK_EX);
       	 	//写入所有内容
        	fwrite($handle, $content, strlen($content));
        	fclose($handle); 
        	return true;
   		 } 
   		 else
   		 {
   		 	return false;
   		 } 
    	  
	}
	
	/**
	 * 析构
	 */
	public function __destruct()
	{
		unset($this->mDictionaryWords, $this->mSpecialCharacters);
	}
	
}
?>