php查找/过滤一段文字中的违禁词敏感词

 

 

 

<?php
   //定义编码  
  header( \'Content-Type:text/html;charset=utf-8 \');  
  
   $words=array(\'我\',\'你\',\'他\');
   $content="测一测我是不是违禁词";
   $banned=generateRegularExpression($words);
  //检查违禁词
   $res_banned=check_words($banned,$content);
   write_html($content,$res_banned);
  
  /**
     * @describe 数组生成正则表达式
     * @param array $words
     * @return string
     */
    function generateRegularExpression($words)
    {
        $regular = implode(\'|\', array_map(\'preg_quote\', $words));
        return "/$regular/i";
    }
	/**
     * @describe 字符串 生成正则表达式
     * @param array $words
     * @return string
     */
	function generateRegularExpressionString($string){
          $str_arr[0]=$string;
          $str_new_arr=  array_map(\'preg_quote\', $str_arr);
          return $str_new_arr[0];
    }
	/**
     * 检查敏感词
     * @param $banned
     * @param $string
     * @return bool|string
     */
	function check_words($banned,$string)
    {    $match_banned=array();
        //循环查出所有敏感词

        $new_banned=strtolower($banned);
        $i=0;
        do{
            $matches=null;
            if (!empty($new_banned) && preg_match($new_banned, $string, $matches)) {
                $isempyt=empty($matches[0]);
                if(!$isempyt){
                    $match_banned = array_merge($match_banned, $matches);
                    $matches_str=strtolower(generateRegularExpressionString($matches[0]));
                    $new_banned=str_replace("|".$matches_str."|","|",$new_banned);
                    $new_banned=str_replace("/".$matches_str."|","/",$new_banned);
                    $new_banned=str_replace("|".$matches_str."/","/",$new_banned);
                }
            }
            $i++;
            if($i>20){
                $isempyt=true;
                break;
            }
        }while(count($matches)>0 && !$isempyt);

        //查出敏感词
        if($match_banned){
            return $match_banned;
        }
        //没有查出敏感词
        return array();
    }
	
	  /**
     * 打印到页面上
     * @param $filepath
     * @param $res_mingan
     * @param $res_banned
     */
    function write_html($content,$res_banned){
      
            print_r($content);
            if($res_banned){
                print_r("  <font color=\'red\'>违禁词（".count($res_banned)."）：</font>".implode(\'|\',$res_banned));
            }
            echo "<br>";
       
    }