You are here

AntiCrawler.php in Anti Spam by CleanTalk 9.1.x

Same filename and directory in other branches
  1. 8.4 src/lib/Cleantalk/Common/Firewall/Modules/AntiCrawler.php

File

src/lib/Cleantalk/Common/Firewall/Modules/AntiCrawler.php
View source
<?php

namespace Cleantalk\Common\Firewall\Modules;

use Cleantalk\Common\DB;
use Cleantalk\Common\Helper;
use Cleantalk\Common\Firewall\FirewallModule;
use Cleantalk\Common\Variables\Server;
use Drupal\cleantalk\CleantalkFuncs;
class AntiCrawler extends FirewallModule {

  /**
   * Public module name
   *
   * @var string
   */
  public $module_name = 'ANTICRAWLER';

  /**
   * Signature - User-Agent + Protocol
   *
   * @var string
   */
  private $sign;
  private $store_interval = 60;

  /**
   * @var false|string
   */
  private $antibot_cookie_value;

  /**
   * @var string
   */
  private $ua_bl_table_name;

  /**
   * @var string|null
   */
  private $db__table__ac_logs;

  /**
   * @var string|null
   */
  private $db__table__sfw_logs;
  private $db__table__sfw;
  public function __construct($data_table = 'cleantalk_ac_logs', $params = []) {
    $db = DB::getInstance();
    $this->ua_bl_table_name = $db->prefix . $data_table;
    $this->db__table__ac_logs = $params['db__table__ac_logs'] ? $db->prefix . $params['db__table__ac_logs'] : null;
    $this->db__table__sfw_logs = $params['db__table__sfw_logs'] ? $db->prefix . $params['db__table__sfw_logs'] : null;
    $this->db__table__sfw = $params['db__table__sfw'] ? $db->prefix . $params['db__table__sfw'] : null;
    $this->sign = md5(Server::get('HTTP_USER_AGENT') . Server::get('HTTPS') . Server::get('HTTP_HOST'));
    $this->antibot_cookie_value = CleantalkFuncs::create_ac_cookie_value();
  }

  /**
   * Use this method to execute main logic of the module.
   *
   * @return array  Array of the check results
   */
  public function check() {
    $results = array();
    foreach ($this->ip_array as $current_ip) {

      // Skip by 301 response code
      if ($this
        ->is_redirected()) {
        $results[] = array(
          'ip' => $current_ip,
          'is_personal' => false,
          'status' => 'PASS_ANTICRAWLER',
        );
        return $results;
      }

      // UA check
      $ua_bl_results = $this->db
        ->fetch_all("SELECT * FROM " . $this->ua_bl_table_name . " ORDER BY `ua_status` DESC;");
      if (!empty($ua_bl_results)) {
        $is_blocked = false;
        foreach ($ua_bl_results as $ua_bl_result) {
          if (!empty($ua_bl_result['ua_template']) && preg_match("%" . str_replace('"', '', $ua_bl_result['ua_template']) . "%i", Server::get('HTTP_USER_AGENT'))) {
            if ($ua_bl_result['ua_status'] == 1) {

              // Whitelisted
              $results[] = array(
                'ip' => $current_ip,
                'is_personal' => false,
                'status' => 'PASS_ANTICRAWLER_UA',
              );
              return $results;
            }
            else {

              // Blacklisted
              $results[] = array(
                'ip' => $current_ip,
                'is_personal' => false,
                'status' => 'DENY_ANTICRAWLER_UA',
              );
              $is_blocked = true;
              break;
            }
          }
        }
        if (!$is_blocked) {
          $results[] = array(
            'ip' => $current_ip,
            'is_personal' => false,
            'status' => 'PASS_ANTICRAWLER_UA',
          );
        }
      }

      // Skip by cookie
      if (CleantalkFuncs::apbct_getcookie('apbct_antibot') == $this->antibot_cookie_value) {
        if (CleantalkFuncs::apbct_getcookie('apbct_anticrawler_passed') == 1) {
          if (!headers_sent()) {
            CleantalkFuncs::apbct_setcookie('apbct_anticrawler_passed', '0');
          }
        }
        $results[] = array(
          'ip' => $current_ip,
          'is_personal' => false,
          'status' => 'PASS_ANTICRAWLER',
        );
        return $results;
      }
    }

    // Common check
    foreach ($this->ip_array as $current_ip) {

      // IP check
      $result = $this->db
        ->fetch("SELECT ip" . ' FROM `' . $this->db__table__ac_logs . '`' . " WHERE `ip` = '{$current_ip}'" . " AND `ua` = '{$this->sign}' AND " . rand(1, 100000) . ";");
      if (isset($result['ip'])) {
        if (CleantalkFuncs::apbct_getcookie('apbct_antibot') !== $this->antibot_cookie_value) {
          $results[] = array(
            'ip' => $current_ip,
            'is_personal' => false,
            'status' => 'DENY_ANTICRAWLER',
          );
        }
        else {
          if (CleantalkFuncs::apbct_getcookie('apbct_anticrawler_passed') === '1') {
            if (!headers_sent()) {
              CleantalkFuncs::apbct_setcookie('apbct_anticrawler_passed', '0');
            }

            // Do logging an one passed request
            $this
              ->update_log($current_ip, 'PASS_ANTICRAWLER');
            $results[] = array(
              'ip' => $current_ip,
              'is_personal' => false,
              'status' => 'PASS_ANTICRAWLER',
            );
            return $results;
          }
        }
      }
      else {
        if (!CleantalkFuncs::apbct_getcookie('apbct_antibot')) {
          $this
            ->update_ac_log();
        }
      }
    }
    return $results;
  }
  public function actionsForDenied($result) {

    // TODO: Implement actionsForDenied() method.
  }
  public function actionsForPassed($result) {

    // TODO: Implement actionsForPassed() method.
  }
  private function is_redirected() {
    $is_redirect = false;
    if (Server::get('HTTP_REFERER') !== '' && Server::get('HTTP_HOST') !== '' && $this
      ->is_cloudflare()) {
      $parse_referer = parse_url(Server::get('HTTP_REFERER'));
      if ($parse_referer && isset($parse_referer['host'])) {
        $is_redirect = Server::get('HTTP_HOST') !== $parse_referer['host'];
      }
    }
    return http_response_code() === 301 || http_response_code() === 302 || $is_redirect;
  }
  private function is_cloudflare() {
    return Server::get('HTTP_CF_RAY') && Server::get('HTTP_CF_CONNECTING_IP') && Server::get('HTTP_CF_REQUEST_ID');
  }

  /**
   * Add entry to SFW log.
   * Writes to database.
   *
   * @param string $ip
   * @param string $status
   * @param string $network
   * @param string $source
   */
  public function update_log($ip, $status, $network = null, $source = null) {
    $id = md5($ip . $this->module_name);
    $time = time();
    if (!$source) {
      $source = 'NULL';
    }
    if (!$network) {
      $network = 'NULL';
    }
    $query = "INSERT INTO `{$this->db__table__sfw_logs}`\n\t\tSET\n\t\t\t`id` = '{$id}',\n\t\t\t`ip` = '{$ip}',\n\t\t\t`status` = '{$status}',\n\t\t\t`all_entries` = 1,\n\t\t\t`blocked_entries` = " . (strpos($status, 'DENY') !== false ? 1 : 0) . ",\n\t\t\t`entries_timestamp` = '{$time}',\n\t\t\t`ua_name` = '" . Server::get('HTTP_USER_AGENT') . "',\n\t\t\t`source` = {$source},\n\t\t\t`network` = '{$network}',\n      `first_url` = '" . substr(Server::get('HTTP_HOST') . Server::get('REQUEST_URI'), 0, 100) . "',\n      `last_url` = '" . substr(Server::get('HTTP_HOST') . Server::get('REQUEST_URI'), 0, 100) . "'\n\t\tON DUPLICATE KEY\n\t\tUPDATE\n\t\t\t`status` = '{$status}',\n\t\t\t`source` = {$source},\n\t\t\t`all_entries` = `all_entries` + 1,\n\t\t\t`blocked_entries` = `blocked_entries`" . (strpos($status, 'DENY') !== false ? ' + 1' : '') . ",\n\t\t\t`entries_timestamp` = '{$time}',\n\t\t\t`ua_name` = '" . Server::get('HTTP_USER_AGENT') . "',\n\t\t\t`network` = '{$network}',\n      `last_url` = '" . substr(Server::get('HTTP_HOST') . Server::get('REQUEST_URI'), 0, 100) . "'";
    $this->db
      ->execute(str_replace(';', '', $query));
  }

  /**
   * Update ac logs table
   */
  public function update_ac_log() {
    $interval_time = Helper::time__get_interval_start($this->store_interval);
    foreach ($this->ip_array as $current_ip) {
      $id = md5($current_ip . $this->sign . $interval_time);
      $this->db
        ->execute("INSERT INTO " . $this->db__table__ac_logs . " SET\n\t\t\t\t\tid = '{$id}',\n\t\t\t\t\tip = '{$current_ip}',\n\t\t\t\t\tua = '{$this->sign}',\n\t\t\t\t\tentries = 1,\n\t\t\t\t\tinterval_start = {$interval_time}\n\t\t\t\tON DUPLICATE KEY UPDATE\n\t\t\t\t\tip = ip,\n\t\t\t\t\tentries = entries + 1,\n\t\t\t\t\tinterval_start = {$interval_time};");
    }
  }

  /**
   * @inheritdoc
   */
  public function _die($result) {
    parent::_die($result);

    // File exists?
    if (file_exists(__DIR__ . "/die_page_anticrawler.html")) {
      $die_page = file_get_contents(__DIR__ . "/die_page_anticrawler.html");
      $net_count = $this->db
        ->fetch('SELECT COUNT(*) as net_count FROM ' . $this->db__table__sfw)['net_count'];

      // Translation
      $replaces = array(
        '{SFW_DIE_NOTICE_IP}' => 'Anti-Crawler Protection is activated for your IP ',
        '{SFW_DIE_MAKE_SURE_JS_ENABLED}' => 'To continue working with the web site, please make sure that you have enabled JavaScript.',
        '{SFW_DIE_YOU_WILL_BE_REDIRECTED}' => 'You will be automatically redirected to the requested page after 3 seconds.<br>' . 'Don\'t close this page. Please, wait for 3 seconds to pass to the page.',
        '{CLEANTALK_TITLE}' => $this
          ->__('Antispam by CleanTalk', 'cleantalk-spam-protect'),
        '{REMOTE_ADDRESS}' => $result['ip'],
        '{SERVICE_ID}' => $net_count,
        '{HOST}' => '',
        '{COOKIE_ANTICRAWLER}' => $this->antibot_cookie_value,
        '{COOKIE_ANTICRAWLER_PASSED}' => '1',
        '{GENERATED}' => '<p>The page was generated at&nbsp;' . date('D, d M Y H:i:s') . "</p>",
        '{USE_ALT_COOKIES}' => \Drupal::config('cleantalk.settings')
          ->get('cleantalk_alternative_cookies_session') ? 1 : 0,
      );
      foreach ($replaces as $place_holder => $replace) {
        $die_page = str_replace($place_holder, $replace, $die_page);
      }
      if (isset($_GET['debug'])) {
        $debug = '<h1>Headers</h1>' . str_replace("\n", "<br>", print_r(\apache_request_headers(), true)) . '<h1>$_SERVER</h1>' . str_replace("\n", "<br>", print_r($_SERVER, true)) . '<h1>IPS</h1>' . str_replace("\n", "<br>", print_r($this->ip_array, true));
      }
      else {
        $debug = '';
      }
      $die_page = str_replace("{DEBUG}", $debug, $die_page);
      die($die_page);
    }
    die("IP BLACKLISTED. Blocked by AntiCrawler " . $result['ip']);
  }

}

Classes

Namesort descending Description
AntiCrawler