PHP--去掉所有的HTML标记和JavaScript标记

 

/* *
 * 去掉所有的HTML标记和JavaScript标记
  */
function  replaceHtmlAndJs( $document )
{
  $document   =   trim ( $document );
  if  ( strlen ( $document )  <=   0 )
 {
   return   $document ;
 }
  $search   =   array  ( " '<script[^>]*?>.*?</script>'si " ,    //  去掉 javascript
                   " '<[\/\!]*?[^<>]*?>'si " ,            //  去掉 HTML 标记
                   " '([\r\n])[\s]+' " ,                  //  去掉空白字符
                   " '&(quot|#34);'i " ,                  //  替换 HTML 实体
                   " '&(amp|#38);'i " ,
                   " '&(lt|#60);'i " ,
                   " '&(gt|#62);'i " ,
                   " '&(nbsp|#160);'i "
                 );                     //  作为 PHP 代码运行

  $replace   =   array  ( "" ,
                    "" ,
                    " \\1 " ,
                    " \ "" ,
                    " & " ,
                    " < " ,
                    " > " ,
                    "   "
                  );

 return @preg_replace ( $search ,  $replace ,  $document );
}