And again

[friendica-addons.git/.git] / blockbot / blockbot.php

diff --git a/blockbot/blockbot.php b/blockbot/blockbot.php
index 424ffb7..4473bbe 100644 (file)
--- a/blockbot/blockbot.php
+++ b/blockbot/blockbot.php
@@ -8,6 +8,7 @@
  */
 
 use Friendica\App;
+use Friendica\Core\Config;
 use Friendica\Core\Hook;
 use Friendica\Core\System;
 use Jaybizzle\CrawlerDetect\CrawlerDetect;
@@ -25,48 +26,51 @@ function blockbot_uninstall() {
 }
 
 function blockbot_init_1(App $a) {
-       $crawlerDetect = new CrawlerDetect();
+       if (empty($_SERVER['HTTP_USER_AGENT'])) {
+               return;
+       }
 
        $logdata = ['agent' => $_SERVER['HTTP_USER_AGENT'], 'uri' => $_SERVER['REQUEST_URI']];
 
-       if (!$crawlerDetect->isCrawler()) {
-               logger::debug('Good user agent detected', $logdata);
-               return;
-       }
+       // List of known crawlers.
+       $agents = ['SEMrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
+               'Diffbot/', 'Twitterbot/', 'YisouSpider/', 'evc-batch/', 'LivelapBot/', 'TrendsmapResolver/',
+               'PaperLiBot/', 'Nuzzel', 'um-LN/', 'Google Favicon', 'Datanyze', 'BLEXBot/', '360Spider',
+               'adscanner/', 'HeadlessChrome', 'wpif', 'startmebot/', 'Googlebot/', 'Applebot/',
+               'facebookexternalhit/', 'GoogleImageProxy', 'bingbot/', 'heritrix/', 'ldspider',
+               'AwarioRssBot/', 'Zabbix', 'TweetmemeBot/', 'dcrawl/', 'PhantomJS/', 'Googlebot-Image/',
+               'CrowdTanglebot/', 'Mediapartners-Google', 'Baiduspider/', 'datagnionbot'];
 
-       // List of strings of reported false positives
-       $agents = ['hackney/', 'Faraday v', 'okhttp', 'UniversalFeedParser', 'PixelFedBot', 'python-requests',
-               'WordPress/', 'http.rb/'];
        foreach ($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('Reported false positive', $logdata);
-                       return;
+                       System::httpExit(403, 'Bots are not allowed');
                }
        }
 
-       // List of false positives' strings of known "good" agents we haven't reported (yet)
-       $agents = ['fediverse.network crawler', 'Active_Pods_CheckBot_3.0', 'Social-Relay/',
-               'curl', 'zgrab', 'Go-http-client', 'curb', 'github.com', 'reqwest', 'Feedly/',
-               'Python-urllib/'];
+       // This switch here is only meant for developers who want to add more bots to the list above, it is not safe for production.
+       if (!Config::get('blockbot', 'training')) {
+               return;
+       }
 
-       foreach ($agents as $agent) {
-               if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
-                       logger::notice('Unreported falsely detected agent', $logdata);
-                       return;
-               }
+       $crawlerDetect = new CrawlerDetect();
+
+       if (!$crawlerDetect->isCrawler()) {
+               logger::debug('Good user agent detected', $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 = ['SEMrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
-               'Diffbot/', 'Twitterbot/', 'YisouSpider/', 'evc-batch/', 'LivelapBot/', 'TrendsmapResolver/',
-               'PaperLiBot/', 'Nuzzel', 'um-LN/', 'Google Favicon', 'Datanyze', 'BLEXBot/', '360Spider',
-               'adscanner/', 'HeadlessChrome', 'wpif', 'startmebot/', 'Googlebot/'];
+       // List of false positives' strings of known "good" agents.
+       $agents = ['fediverse.network crawler', 'Active_Pods_CheckBot_3.0', 'Social-Relay/',
+               'curl', 'zgrab', 'Go-http-client', 'curb', 'github.com', 'reqwest', 'Feedly/',
+               'Python-urllib/', 'Liferea/', 'aiohttp/', 'WordPress.com Reader', 'hackney/',
+               'Faraday v', 'okhttp', 'UniversalFeedParser', 'PixelFedBot', 'python-requests',
+               'WordPress/', 'http.rb/', 'Apache-HttpClient/', 'WordPress.com;', 'Pleroma',
+               'Dispatch/'];
 
        foreach ($agents as $agent) {
                if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
-                       System::httpExit(403, 'Bots are not allowed');
+                       logger::notice('False positive', $logdata);
+                       return;
                }
        }