friendica-addons/blockbot/blockbot.php

78 lines
2.7 KiB
PHP
Raw Normal View History

2019-04-20 12:15:45 +00:00
<?php
/**
* Name: blockbot
2019-04-20 12:15:45 +00:00
* Description: Blocking bots based on detecting bots/crawlers/spiders via the user agent and http_from header.
* Version: 0.1
* Author: Philipp Holzer <admin@philipp.info>
*
*/
use Friendica\App;
use Friendica\Core\Hook;
use Friendica\Core\System;
use Jaybizzle\CrawlerDetect\CrawlerDetect;
2019-04-27 11:51:44 +00:00
use Friendica\Core\Logger;
2019-04-20 12:15:45 +00:00
2019-04-21 10:35:33 +00:00
require_once __DIR__ . DIRECTORY_SEPARATOR . 'vendor' . DIRECTORY_SEPARATOR . 'autoload.php';
function blockbot_install() {
2019-04-22 08:49:40 +00:00
Hook::register('init_1', __FILE__, 'blockbot_init_1');
2019-04-20 12:15:45 +00:00
}
function blockbot_uninstall() {
2019-04-22 08:49:40 +00:00
Hook::unregister('init_1', __FILE__, 'blockbot_init_1');
2019-04-20 12:15:45 +00:00
}
function blockbot_init_1(App $a) {
2019-04-20 12:15:45 +00:00
$crawlerDetect = new CrawlerDetect();
$logdata = ['agent' => $_SERVER['HTTP_USER_AGENT'], 'uri' => $_SERVER['REQUEST_URI']];
if (!$crawlerDetect->isCrawler()) {
logger::debug('Good user agent detected', $logdata);
return;
}
// List of strings of reported false positives
$agents = ['Mastodon', 'hackney', 'Faraday', 'okhttp', 'UniversalFeedParser', 'PixelFedBot', 'python-requests',
2019-04-27 13:50:25 +00:00
'WordPress', 'http.rb'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
// The agents had been reported to https://github.com/JayBizzle/Crawler-Detect/issues/
logger::notice('Already reported wrong detection', $logdata);
return;
}
}
2019-04-27 11:51:44 +00:00
// List of strings of known "good" agents
$agents = ['diaspora-connection-tester', 'DiasporaFederation', 'Friendica', '(compatible; zot)',
'Micro.blog', 'GangGo', 'python/federation', 'GNU social', 'winHttp',
2019-04-27 11:51:44 +00:00
'Go-http-client', 'Mr.4x3 Powered', 'Test Certificate Info', 'WordPress.com', 'zgrab',
'curl/', 'StatusNet', 'OpenGraphReader/', 'Uptimebot/', 'python-opengraph-jaywink',
'fediverse.network crawler', 'Active_Pods_CheckBot_3.0', 'Social-Relay'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
// Report every false positive here: https://github.com/JayBizzle/Crawler-Detect/issues/
// After report move it into the array above
logger::notice('False positive', $logdata);
return;
}
}
// List of known crawlers. They are added here to avoid having them logged at the end of the function.
// This helps to detect false positives
$agents = ['Mozilla/5.0 (compatible; SemrushBot/3~bl; +http://www.semrush.com/bot.html)', 'SEMrushBot',
'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/65.0.3325.181 Safari/537.36 AppEngine-Google; (+http://code.google.com/appengine; appid: s~feedly-nikon3)'];
foreach ($agents as $agent) {
if ($_SERVER['HTTP_USER_AGENT'] == $agent) {
System::httpExit(403, 'Bots are not allowed');
2019-04-27 11:51:44 +00:00
}
2019-04-20 12:15:45 +00:00
}
logger::info('Blocked bot', $logdata);
System::httpExit(403, 'Bots are not allowed');
2019-04-20 12:15:45 +00:00
}