- [1. 第三方过滤函数与类](#1-第三方过滤函数与类)
	- [1.1. discuz SQL安全过滤类分析](#11-discuz-sql安全过滤类分析)
	- [1.2. discuz XSS标签过滤函数分析](#12-discuz-xss标签过滤函数分析)
- [2. PHP内置过滤函数](#2-php内置过滤函数)
	- [2.1. SQL注入过滤函数](#21-sql注入过滤函数)
	- [2.2. XSS注入过滤函数](#22-xss注入过滤函数)
	- [2.3. 命令执行过滤函数](#23-命令执行过滤函数)

# 1. 第三方过滤函数与类

目前大多数应用都有一个参数过滤的统一入口，类似于dedecms的代码。

```/include/common.inc.php```

```php
foreach(Array('_GET','_POST','_COOKIE') as $_request)		// line 52
{
	foreach($$_request as $_k => $_v) ${$_k} = _RunMagicQuotes($_v);
}
```

_RunMagicQuotes 函数

```php
function _RunMagicQuotes(&$svar)
{
	if(!get_magic_quotes_gpc())
	{
		if( is_array($svar) )
		{
			foreach($svar as $_k => $_v) $svar[$_k] = _RunMagicQuotes($_v);
		}
		else
		{
			$svar = addslashes($svar);
		}
	}
	return $svar;
}
```

这里的确使用了 ```addslashes```函数对特定的字符进行转义，但是还不够完善。

## 1.1. discuz SQL安全过滤类分析

discuz专门有一个SQL注入过滤类来过滤SQL注入请求```/config/config global.php```中。

```php
// -------------------------  CONFIG SECURITY  -------------------------- //  line 57
$_config['security']['authkey'] = '2e4617iIZ7jIVt61';
$_config['security']['urlxssdefend'] = 1;
$_config['security']['attackevasive'] = '0';
$_config['security']['querysafe']['status'] = 1;
$_config['security']['querysafe']['dfunction']['0'] = 'load_file';
$_config['security']['querysafe']['dfunction']['1'] = 'hex';
$_config['security']['querysafe']['dfunction']['2'] = 'substring';
$_config['security']['querysafe']['dfunction']['3'] = 'if';
$_config['security']['querysafe']['dfunction']['4'] = 'ord';
$_config['security']['querysafe']['dfunction']['5'] = 'char';
$_config['security']['querysafe']['daction']['0'] = '@';
$_config['security']['querysafe']['daction']['1'] = 'intooutfile';
$_config['security']['querysafe']['daction']['2'] = 'intodumpfile';
$_config['security']['querysafe']['daction']['3'] = 'unionselect';
$_config['security']['querysafe']['daction']['4'] = '(select';
$_config['security']['querysafe']['daction']['5'] = 'unionall';
$_config['security']['querysafe']['daction']['6'] = 'uniondistinct';
$_config['security']['querysafe']['dnote']['0'] = '/*';
$_config['security']['querysafe']['dnote']['1'] = '*/';
$_config['security']['querysafe']['dnote']['2'] = '#';
$_config['security']['querysafe']['dnote']['3'] = '--';
$_config['security']['querysafe']['dnote']['4'] = '"';
$_config['security']['querysafe']['dlikehex'] = 1;
$_config['security']['querysafe']['afullnote'] = '0';
```

- ```$_config['security']['querysafe']['status'] = 1;```设置是否开启SQL注入防御，这个选项默认开启
- ```'daction']```和```['dnote']```都是SQL注入过滤类的过滤规则，规则里包含了常见的注入关键字
- Discuz 执行 SQL 语句之前会调用 \source\class\discuz\discuz_database.php 文件 discuz_database_safecheck类
- 下面的checkquery($sql)函数进行过滤

```php
	public static function checkquery($sql) {			// line 347
		if (self::$config === null) {
			self::$config = getglobal('config/security/querysafe');
		}
		if (self::$config['status']) {
			$check = 1;
			$cmd = strtoupper(substr(trim($sql), 0, 3));
			if(isset(self::$checkcmd[$cmd])) {
				$check = self::_do_query_safe($sql);
			} elseif(substr($cmd, 0, 2) === '/*') {
				$check = -1;
			}

			if ($check < 1) {
				throw new DbException('It is not safe to do this query', 0, $sql);
			}
		}
		return true;
	}
```

从代码中可以看到，程序首先加载配置文件中的config/security/querysafe，根据 $config['status']判断 SQL 注入防御是否开启，再到 $check = self::_do_query_safe($sql); 可以看到该函数又调用了同类下的_do_query_safe()函数对SQL语句进行过滤。

```php
	private static function _do_query_safe($sql) {
		$sql = str_replace(array('\\\\', '\\\'', '\\"', '\'\''), '', $sql);
		$mark = $clean = '';
		if (strpos($sql, '/') === false && strpos($sql, '#') === false && strpos($sql, '-- ') === false && strpos($sql, '@') === false && strpos($sql, '`') === false) {
			$clean = preg_replace("/'(.+?)'/s", '', $sql);
		} else {
			$len = strlen($sql);
			$mark = $clean = '';
			for ($i = 0; $i < $len; $i++) {
				$str = $sql[$i];
				switch ($str) {
					case '`':
						if(!$mark) {
							$mark = '`';
							$clean .= $str;
						} elseif ($mark == '`') {
							$mark = '';
						}
						break;
					case '\'':
						if (!$mark) {
							$mark = '\'';
							$clean .= $str;
						} elseif ($mark == '\'') {
							$mark = '';
						}
						break;
					case '/':
						if (empty($mark) && $sql[$i + 1] == '*') {
							$mark = '/*';
							$clean .= $mark;
							$i++;
						} elseif ($mark == '/*' && $sql[$i - 1] == '*') {
							$mark = '';
							$clean .= '*';
						}
						break;
					case '#':
						if (empty($mark)) {
							$mark = $str;
							$clean .= $str;
						}
						break;
					case "\n":
						if ($mark == '#' || $mark == '--') {
							$mark = '';
						}
						break;
					case '-':
						if (empty($mark) && substr($sql, $i, 3) == '-- ') {
							$mark = '-- ';
							$clean .= $mark;
						}
						break;

					default:

						break;
				}
				$clean .= $mark ? '' : $str;
			}
		}

		if(strpos($clean, '@') !== false) {
			return '-3';
		}

		$clean = preg_replace("/[^a-z0-9_\-\(\)#\*\/\"]+/is", "", strtolower($clean));

		if (self::$config['afullnote']) {
			$clean = str_replace('/**/', '', $clean);
		}

		if (is_array(self::$config['dfunction'])) {
			foreach (self::$config['dfunction'] as $fun) {
				if (strpos($clean, $fun . '(') !== false)
					return '-1';
			}
		}

		if (is_array(self::$config['daction'])) {
			foreach (self::$config['daction'] as $action) {
				if (strpos($clean, $action) !== false)
					return '-3';
			}
		}

		if (self::$config['dlikehex'] && strpos($clean, 'like0x')) {
			return '-2';
		}

		if (is_array(self::$config['dnote'])) {
			foreach (self::$config['dnote'] as $note) {
				if (strpos($clean, $note) !== false)
					return '-4';
			}
		}

		return 1;
	}
```

该函数首先使用```$sql = str_replace(array('\\\\', '\\\'', '\\"', '\'\''), '', $sql);```将SQL语句中的\\、\'、\"以及'、"替换为空紧接着是一个if else判断逻辑来选择过滤的方式：

```$clean = preg_replace("/'(.+?)'/s", '', $sql);```

- 这段代码表示当SQL语句里存在'/'、'#'、'-- '、'@'、'`'这些字符时，则直接调用 preg_replace()函数将单引号(')中间的内容替换为空
- 正则表达式中？匹配前面的子表达式零次或一次。
- /s：默认情况下的圆点 . 是匹配除换行符 \n 之外的任何字符，加上 s 修饰符之后，圆点 . 中包含换行符 \n。

else条件中是对整段SQL语句进行逐个字符进行判断:

```php
else {
			$len = strlen($sql);
			$mark = $clean = '';
			for ($i = 0; $i < $len; $i++) {
```

```php
					case '/':
						if (empty($mark) && $sql[$i + 1] == '*') {
							$mark = '/*';
							$clean .= $mark;
							$i++;
						} elseif ($mark == '/*' && $sql[$i - 1] == '*') {
							$mark = '';
							$clean .= '*';
						}
						break;
```

这段代码的逻辑是：当检查到SQL语句中存在斜杠(/)时，则去判断下一个字符是不是星号(*),如果是星号(*)就把这两个字符拼接起来，即/*,然后继续判断下一个字符是不是星号(*)，如果是星号则再继续拼接起来，得到/**

最后在如下代码中判断是否存在原来拦截规则里面定义的字符，如果存在则拦截SQL语句执行：

```php
		if (is_array(self::$config['dnote'])) {
			foreach (self::$config['dnote'] as $note) {
				if (strpos($clean, $note) !== false)
					return '-4';
			}
		}
```

## 1.2. discuz XSS标签过滤函数分析

在XSS的防御上，只要过滤掉尖括号以及单、 双引号就能干掉绝大部分的payload。

下面是discuz的HTML标签过滤代码在/source/function/function_blog.php文件中

```php
function checkhtml($html) {			// line 271
	if(!checkperm('allowhtml')) {

		preg_match_all("/\<([^\<]+)\>/is", $html, $ms);

		$searchs[] = '<';
		$replaces[] = '&lt;';
		$searchs[] = '>';
		$replaces[] = '&gt;';

		if($ms[1]) {
			$allowtags = 'img|a|font|div|table|tbody|caption|tr|td|th|br|p|b|strong|i|u|em|span|ol|ul|li|blockquote';
			$ms[1] = array_unique($ms[1]);
			foreach ($ms[1] as $value) {
				$searchs[] = "&lt;".$value."&gt;";

				$value = str_replace('&amp;', '_uch_tmp_str_', $value);
				$value = dhtmlspecialchars($value);
				$value = str_replace('_uch_tmp_str_', '&amp;', $value);

				$value = str_replace(array('\\','/*'), array('.','/.'), $value);
				$skipkeys = array('onabort','onactivate','onafterprint','onafterupdate','onbeforeactivate','onbeforecopy','onbeforecut','onbeforedeactivate',
						'onbeforeeditfocus','onbeforepaste','onbeforeprint','onbeforeunload','onbeforeupdate','onblur','onbounce','oncellchange','onchange',
						'onclick','oncontextmenu','oncontrolselect','oncopy','oncut','ondataavailable','ondatasetchanged','ondatasetcomplete','ondblclick',
						'ondeactivate','ondrag','ondragend','ondragenter','ondragleave','ondragover','ondragstart','ondrop','onerror','onerrorupdate',
						'onfilterchange','onfinish','onfocus','onfocusin','onfocusout','onhelp','onkeydown','onkeypress','onkeyup','onlayoutcomplete',
						'onload','onlosecapture','onmousedown','onmouseenter','onmouseleave','onmousemove','onmouseout','onmouseover','onmouseup','onmousewheel',
						'onmove','onmoveend','onmovestart','onpaste','onpropertychange','onreadystatechange','onreset','onresize','onresizeend','onresizestart',
						'onrowenter','onrowexit','onrowsdelete','onrowsinserted','onscroll','onselect','onselectionchange','onselectstart','onstart','onstop',
						'onsubmit','onunload','javascript','script','eval','behaviour','expression','style','class');
				$skipstr = implode('|', $skipkeys);
				$value = preg_replace(array("/($skipstr)/i"), '.', $value);
				if(!preg_match("/^[\/|\s]?($allowtags)(\s+|$)/is", $value)) {
					$value = '';
				}
				$replaces[] = empty($value)?'':"<".str_replace('&quot;', '"', $value).">";
			}
		}
		$html = str_replace($searchs, $replaces, $html);
	}

	return $html;
}
```

```preg_match_all("/\<([^\<]+)\>/is", $html, $ms);```正则取岀来尖括号中间的内容然后在if($ms[l])这个if条件里对这些标签中的关键字进行筛选。

```$skipkeys```变量定义了很多on事件的敏感字符。

最后拼接正则将这些字符串替换掉。

```php
				$skipstr = implode('|', $skipkeys);
				$value = preg_replace(array("/($skipstr)/i"), '.', $value);
```

# 2. PHP内置过滤函数

## 2.1. SQL注入过滤函数

- SQL 注入过滤函数有 addslashes()、mysql_real_escape_string()以及 mysql_escape_string(),
- 它们的作用都是给字符串添加反斜杠(\)来转义掉单引号(')、双引号('')、反斜线(\)以及空字符 NULL。

## 2.2. XSS注入过滤函数

htmlspecialchars()函数的作用是将字符串中的特殊字符转换成HTML实体编码，

- 如```&``` 转换成```&amp;``` 

- ```"``` 转换成```&quot;```

- ```'``` 转换成```&#039;```

- ```<``` 转换成```&lt;```

- ```>``` 转换成```&gt;```

strip_tags()函数则是用来去掉HTML及PHP标记比如给这个函数传入  ```<h1>xxxxx</h1>```经过它处理后返回的字符串为 ```xxxxx```。

## 2.3. 命令执行过滤函数

PHP为了防止系统命令注入的漏洞，提供了 [escapeshellcmd()](https://www.php.net/manual/zh/function.escapeshellcmd.php)和[escapeshellarg()]()两个函数对参数进行过滤。