$ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $address);
        curl_setopt($ch, CURLOPT_HEADER, 1);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
        curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
        curl_setopt($ch, CURLOPT_USERAGENT,
                    'Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.36 (KHTML, like Gecko) Ubuntu Chromium/34.0.1847.116 Chrome/34.0.1847.116 Safari/537.36');

        $dirty_html = curl_exec($ch);

        $response = (object)curl_getinfo($ch);

        if(stripos($response->content_type, 'windows-1251')){;
          $dirty_html = mb_convert_encoding($dirty_html, 'utf-8', 'cp1251');
        }
        curl_close($ch);

        $prf = new HTMLPurifier();
        $clean_html = $prf->process($dirty_html,[
              'AutoFormat.AutoParagraph' => true,
              'AutoFormat.RemoveEmpty' => true,
              'AutoFormat.RemoveEmpty.RemoveNbsp' => true,
              'HTML.AllowedElements' => ['p','a','br','table','tbody','tr','th','td','h1','span','big'],
              'HTML.ForbiddenAttributes' => [
                                             'a@href',
                                             'a@style',
                                             'a@rel',
                                             'a@class',
                                             'a@title',
                                             'p@class',
                                             'p@style',
                                             'table@style',
                                             'table@border',
                                             'table@width',
                                             'td@class',
                                             'th@class',
                                             'td@valign',
                                             'td@style',
                                             'td@width',
                                             'td@colspan',
                                             'td@align',
                                             'td@rowspan',
                                             'span@class'
                                            ]
          ]);
  $search = ['<p>','</p>','&amp;','<a>','</a>','<table>','<tbody>','</tbody>','</table>','<tr>','</tr>','<th>','</th>','<td>','</td>','<h1>','</h1>','<span>','</span>','<big>','</big>'];
        $replace = ['',"\r\n",'','',"\r\n",'','','','','',"\r\n",'',' ','','','',"\r\n",'',' ','',' '];
        $clean_html = preg_replace("#(?:<p>\s+</p>)#", '', $clean_html);
        $clean_html = str_replace(["\r","\n","\r\n","\t",'  '], '', $clean_html);
        $clean_html = str_replace($search, $replace, $clean_html);
        $clean_html = preg_replace("#(?:\r\n?|\n){2,}#", "\r\n", $clean_html);
        $clean_html = str_replace(['<br>', '<br />'], " ", $clean_html);
        $clean_html = preg_replace("#(?:\"\'\+;)+#", '', $clean_html);

        echo $clean_html