$logdata = ['agent' => $_SERVER['HTTP_USER_AGENT'], 'uri' => $_SERVER['REQUEST_URI']];
- // 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.
+ // List of known crawlers.
$agents = ['SEMrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
'Diffbot/', 'Twitterbot/', 'YisouSpider/', 'evc-batch/', 'LivelapBot/', 'TrendsmapResolver/',
'PaperLiBot/', 'Nuzzel', 'um-LN/', 'Google Favicon', 'Datanyze', 'BLEXBot/', '360Spider',
'adscanner/', 'HeadlessChrome', 'wpif', 'startmebot/', 'Googlebot/', 'Applebot/',
'facebookexternalhit/', 'GoogleImageProxy', 'bingbot/', 'heritrix/', 'ldspider',
- 'AwarioRssBot/', 'Zabbix'];
+ 'AwarioRssBot/', 'Zabbix', 'TweetmemeBot/', 'dcrawl/', 'PhantomJS/', 'Googlebot-Image/',
+ 'CrowdTanglebot/', 'Mediapartners-Google', 'Baiduspider/', 'datagnionbot'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
}
}
- // This switch here is just meant for developers who want to add more bots to the list above
+ // This switch here is only meant for developers who want to add more bots to the list above, it is not safe for production.
if (!Config::get('blockbot', 'training')) {
return;
}
'curl', 'zgrab', 'Go-http-client', 'curb', 'github.com', 'reqwest', 'Feedly/',
'Python-urllib/', 'Liferea/', 'aiohttp/', 'WordPress.com Reader', 'hackney/',
'Faraday v', 'okhttp', 'UniversalFeedParser', 'PixelFedBot', 'python-requests',
- 'WordPress/', 'http.rb/', 'Apache-HttpClient/', 'WordPress.com;', 'Pleroma'];
+ 'WordPress/', 'http.rb/', 'Apache-HttpClient/', 'WordPress.com;', 'Pleroma',
+ 'Dispatch/'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {