全站違禁詞掃描下載地址
https://files.cnblogs.com/files/kingchou/%E8%BF%9D%E7%A6%81%E8%AF%8D%E7%B1%BB%E6%96%87%E4%BB%B6.rar
全站掃描違禁詞代碼
require_once ROOT_PATH."/Banned.php";
$banned=new Banned();
$banned->write_html=true;
$banned->write_log=true;
$banned->check_type="file";
$banned->checkFileAll();
類文件
<?php
date_default_timezone_set('Asia/Shanghai');
/*error_reporting(E_WARNING);*/
/**
* 違禁詞：
* Created by PhpStorm.
* Author: zhouzj -周宗君
* Date: 2017/11/17 15:52
*/
class Banned
{
private $data;//數據
private $match_banned;//違禁詞規則
private $match_mingan;//敏感詞規則
private $match_field;//字符類型規則
private $finish_path=array();//已完成的路徑
private $finish_table=array();//已完成的表
private $logtime;
private $path;
private $check_sub_dir;
private $check_type;
private $banned_from;
private $clean;
private $write_html;
private $write_log;
private $fornum;
private $bannedword;
public $document_root;
public $_config=array(
"path"=>'',
"check_sub_dir"=>array('web'),
'check_type'=>'file',
'banned_from'=>'table',
'clean'=>false,
'fornum'=>20,
'write_html'=>false,
'write_log'=>false,
'bannedword'=>true,);
/**
*初始化方法
*/
public function __construct($config=array())
{
if(!empty($config)) {
$this->_config = array_merge($this->_config, $config);
}
$this->BannedInit();
}
/**
* 設置key
* @param $key
* @param $value
*/
public function __set($key, $value)
{
$this->_config[$key] = $value;
}
/**
* 讀取key
* @param $key
* @return mixed
*/
public function __get($key)
{
return $this->_config[$key];
}
/**
* 初始化違禁詞和敏感詞
*/
private function BannedInit(){
$this->logtime=date("YmdHis");
$this->document_root=$_SERVER['DOCUMENT_ROOT'];
//初始化加載違禁詞 否則默認讀取文件中的違禁詞
//敏感詞正則規則
$mingan_words= $this->getMinganWords();
$this->match_mingan=$this->generateRegularExpression($mingan_words);
//違禁詞正則規則
$banned_words= $this->getBannedWords();
$this->match_banned=$this->generateRegularExpression($banned_words);
}
/**
* 檢測所有內容
*/
public function checkAll(){
//判斷類型是數據庫
if($this->_config['check_type']=="file"){
return $this->checkFileAll();//檢測文件
}
}
/**
* 檢查文件
*/
public function checkFileAll(){
$path=$this->_config['path'];
if(!$path){
$path=$this->document_root."";
}
$laststr= substr($path, -1);
if($laststr!="/"){
$path.="/";
}
$dir_list= $this->getDir($path);
//判斷是否包含子目錄
if(count($dir_list)>0){
foreach ($dir_list as $dirkey=>$dirvalue){
if(empty($dirvalue)){
continue;
}
//判斷是否是允許檢測的子目錄
// if(in_array($dirvalue,$this->_config['check_sub_dir'])){
$subpath=$path.$dirvalue.'/';
$this->check_sub_dir($subpath);
// }
}
}
return true;
}
/**
* 檢測單條內容
*/
public function check($content){
if($this->_config['check_type']=="file"){
return $this->check_file($content);//檢測文件
}
}
/**
* 檢測單條文本內容
* @param $content
*/
public function check_text($content){
$res_mingan=array();$res_banned=array();
$this->__check_content($content,$res_mingan,$res_banned);
$data['mingan']=$res_mingan;
$data['banned']=$res_banned;
return $data;
}
/**
* 檢測單條文件內容
* @param $content
*/
public function check_file($content){
$res= $this->__check_file($content);
$this->set_finish_path($content);
return $res;
}
/**
* 迭代檢查子文件目錄
* @param $path
*/
function check_sub_dir($path){
$file_list= $this->getFile($path);//獲取文件目錄
if(count($file_list)>0){
foreach ($file_list as $filekey=>$filevalue){
if(empty($filevalue))
continue;
$this->__check_file($filevalue);//執行檢查文件
}
}
$dir_list= $this->getDir($path);//獲取文件夾目錄
if(count($dir_list)>0){
foreach ($dir_list as $dirkey=>$dirvalue){
if(empty($dirvalue)){
continue;
}
$subpath=$path.$dirvalue.'/';
$this->check_sub_dir($subpath);
}
}
}
/**
* 驗證單個文件
* @param $filepath
*/
function __check_file($filepath){
//判斷文件是否已經完成檢查，如果已經完成則不需要檢查
if(!$filepath) return;
if(in_array($filepath,$this->finish_path))
return;
if(!file_exists($filepath)) return;
if(stripos($filepath,"mingan_words.txt")>0 || stripos($filepath,"banned_words.txt")>0 ){
return ;
}
//判斷文件如果是 違禁詞或敏感詞文件則跳過不處理
$content = file_get_contents($filepath);
$res_mingan=array();$res_banned=array();
$this->__check_content($content,$res_mingan,$res_banned);
$data=array();
if($res_mingan || $res_banned){//如果有敏感詞或違禁詞則寫日志
$this->write_log($filepath,$res_mingan,$res_banned);
$this->write_html($filepath,$res_mingan,$res_banned);
$data['mingan']=$res_mingan;
$data['banned']=$res_banned;
}
//執行保存文件路徑
return $data;
}
/**
* 檢查內容
* @param $content
* @param $res_mingan
* @param $res_banned
*/
private function __check_content($content,&$res_mingan,&$res_banned){
//檢查敏感詞
$res_mingan=$this->check_words($this->match_mingan,$content);
//檢查違禁詞
$res_banned=$this->check_words($this->match_banned,$content);
}
/**
* 檢查敏感詞
* @param $banned
* @param $string
* @return bool|string
*/
private function check_words($banned,$string)
{ $match_banned=array();
//循環查出所有敏感詞
$new_banned=strtolower($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);
$matches_str=strtolower($this->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>$this->_config['fornum']){
$isempyt=true;
break;
}
}while(count($matches)>0 && !$isempyt);
//查出敏感詞
if($match_banned){
return $match_banned;
}
//沒有查出敏感詞
return array();
}
/**
* @describe 生成正則表達式
* @param array $words
* @return string
*/
private function generateRegularExpression($words)
{
$regular = implode('|', array_map('preg_quote', $words));
return "/$regular/i";
}
/**
* @describe 生成正則表達式
* @param array $words
* @return string
*/
private function generateRegularExpressionString($string){
$str_arr[0]=$string;
$str_new_arr= array_map('preg_quote', $str_arr);
return $str_new_arr[0];
}
/**
* 寫日志
* @param $path
* @param $content
*/
private function write_log($location,$contentarr,$weijinciarr){
if($this->_config['write_log']) {
if (!$contentarr && !$weijinciarr) {
return;
}
$content = $location;
if (count($contentarr) > 0) {
$content .= "," . count($contentarr) . "," . implode('|', $contentarr);
} else {
$content .= ",,";
}
if (count($weijinciarr) > 0) {
$content .= "," . count($weijinciarr) . "," . implode('|', $weijinciarr);
} else {
$content .= ",,";
}
$content .= "\r\n";
$filename =$this->document_root."/logs/file" . $this->logtime . "/file_bannwords.csv";
/* 文件日志路徑 */
// $file = './' . $filename;
$file = $filename;
if (!file_exists($file)) {
$pathdir = dirname($file);
if (!is_dir($pathdir)) {
mkdir($pathdir, 0775, true);
}
$content_title = "位置,敏感詞數量,敏感詞,違禁詞數量,違禁詞" . "\r\n";
error_log(iconv('UTF-8', 'GB2312', $content_title), 3, $file);
}
error_log(iconv('UTF-8', 'GB2312', $content), 3, $file);
}
}
/**
* 打印到頁面上
* @param $filepath
* @param $res_mingan
* @param $res_banned
*/
private function write_html($location,$res_mingan,$res_banned){
if($this->_config['write_html']){
print_r(iconv('GB2312','UTF-8',$location));
if($res_mingan){
print_r(" <font color='red'>敏感詞（".count($res_mingan)."）：</font>".implode('|',$res_mingan));
}
if($res_banned){
print_r(" <font color='red'>違禁詞（".count($res_banned)."）：</font>".implode('|',$res_banned));
}
echo "<br>";
}
}
/**
* 保存已完成文件
* @param $path
*/
private function set_finish_path($path){
if(!$path){
return;
}
$content =$path. "\r\n";
$filename=$this->document_root."/logs/file" . $this->logtime . "/banned_finish_path.txt";
/* 文件日志路徑 */
// $file ='./' . $filename;
$file = $filename;
if (!file_exists($file)) {
mkdir(dirname($file), 0775, true);
}
error_log(iconv('GB2312','UTF-8',$content), 3, $file);
}
//重置已完成文件
/* function clean_finish_file(){
$filename=$this->document_root."/logs/banned_finish_path.txt";
file_put_contents($filename,'');
}
*/
//獲取文件目錄列表,該方法返回數組
private function getDir($dir) {
$dirArray[]=NULL;
if (is_dir($dir)) {
try{
if (false != ($handle = @opendir($dir))) {
$i = 0;
while (false !== ($file = @readdir($handle))) {
//去掉"“.”、“..”以及帶“.xxx”后綴的文件
if ($file != "." && $file != ".." && !strpos("*" . $file, ".")) {
$dirArray[$i] = $file;
$i++;
}
}
//關閉句柄
@closedir($handle);
}
}catch (Exception $ex){
}
}
return $dirArray;
}
//獲取文件列表
private function getFile($dir) {
$fileArray[]=NULL;
if (false != ($handle = @opendir ( $dir ))) {
$i=0;
while ( false !== ($file = @readdir ( $handle )) ) {
//去掉"“.”、“..”以及帶“.xxx”后綴的文件
if ($file != "." && $file != ".." && (strpos($file,".php") || strpos($file,".html"))) {
$fileArray[$i]=$dir.$file;
if($i==1000){//當同一個文件下超出1000個文件則跳出循環
break;
}
$i++;
}
}
//關閉句柄
@closedir ( $handle );
}
return $fileArray;
}
//獲取敏感詞文件
private function getMinganWords(){
$shehuangwords=file_get_contents($this->document_root."/words/forbidden.txt");
// $shehuangwords=iconv("GBK","UTF-8",$shehuangwords);
$shehuangword_arr=explode("\r\n",$shehuangwords);
return $shehuangword_arr;
}
//獲取違禁詞文件
private function getBannedWords(){
if($this->_config['bannedword']){
$guanggaowords=file_get_contents($this->document_root."/words/banned.txt");
// $guanggaowords=iconv("GBK","UTF-8",$guanggaowords);
$guanggaowords_arr=explode("\r\n",$guanggaowords);
return $guanggaowords_arr;
}else{
return array();
}
}
}