blockbot/blockbot.php

   1 <?php
   2 /**
   3  * Name: blockbot
   4  * Description: Blocking bots based on detecting bots/crawlers/spiders via the user agent and http_from header.
   5  * Version: 0.1
   6  * Author: Philipp Holzer <admin@philipp.info>
   7  *
   8  */
   9
  10 use Friendica\App;
  11 use Friendica\Core\Hook;
  12 use Friendica\Core\System;
  13 use Jaybizzle\CrawlerDetect\CrawlerDetect;
  14 use Friendica\Core\Logger;
  15
  16 require_once __DIR__ . DIRECTORY_SEPARATOR . 'vendor' . DIRECTORY_SEPARATOR . 'autoload.php';
  17
  18 function blockbot_install() {
  19         Hook::register('init_1', __FILE__, 'blockbot_init_1');
  20 }
  21
  22
  23 function blockbot_uninstall() {
  24         Hook::unregister('init_1', __FILE__, 'blockbot_init_1');
  25 }
  26
  27 function blockbot_init_1(App $a) {
  28         $crawlerDetect = new CrawlerDetect();
  29
  30         // List of strings of known "good" agents
  31         $agents = ['diaspora-connection-tester', 'DiasporaFederation', 'Friendica', '(compatible; zot)',
  32                 'Micro.blog', 'Mastodon', 'hackney', 'GangGo', 'python/federation', 'GNU social', 'winHttp',
  33                 'Go-http-client', 'Mr.4x3 Powered', 'Test Certificate Info', 'WordPress.com', 'zgrab',
  34                 'curl/', 'StatusNet', 'OpenGraphReader/', 'Uptimebot/', 'python-opengraph-jaywink'];
  35
  36         if ($crawlerDetect->isCrawler()) {
  37                 foreach ($agents as $agent) {
  38                         if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
  39                                 // @ToDo: Report every false positive here: https://github.com/JayBizzle/Crawler-Detect/issues/326
  40                                 logger::notice('False positive', ['agent' => $_SERVER['HTTP_USER_AGENT']]);
  41                                 return;
  42                         }
  43                 }
  44                 logger::info('Blocked bot', ['agent' => $_SERVER['HTTP_USER_AGENT']]);
  45                 System::httpExit(403, 'Bots are not allowed');
  46         } else {
  47                 logger::debug('Good user agent detected', ['agent' => $_SERVER['HTTP_USER_AGENT']]);
  48         }
  49 }