hankg/friendica-addons
1
0
Fork 0
forked from friendica/friendica-addons
Code Pull requests Activity

Added training mode

Browse source
This commit is contained in:
Michael 2019-05-29 18:51:07 +00:00
parent 02fdb7c654
commit 9e8ab6ccf0
1 changed files with 23 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

40
blockbot/blockbot.php
View file

@ -8,6 +8,7 @@
*/ */
use Friendica\App; use Friendica\App;
use Friendica\Core\Config;
use Friendica\Core\Hook; use Friendica\Core\Hook;
use Friendica\Core\System; use Friendica\Core\System;
use Jaybizzle\CrawlerDetect\CrawlerDetect; use Jaybizzle\CrawlerDetect\CrawlerDetect;
@ -25,14 +26,33 @@ function blockbot_uninstall() {
} }
function blockbot_init_1(App $a) { function blockbot_init_1(App $a) {
$crawlerDetect = new CrawlerDetect();
if (empty($_SERVER['HTTP_USER_AGENT'])) { if (empty($_SERVER['HTTP_USER_AGENT'])) {
return; return;
} }
$logdata = ['agent' => $_SERVER['HTTP_USER_AGENT'], 'uri' => $_SERVER['REQUEST_URI']]; $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.
$agents = ['SEMrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
'Diffbot/', 'Twitterbot/', 'YisouSpider/', 'evc-batch/', 'LivelapBot/', 'TrendsmapResolver/',
'PaperLiBot/', 'Nuzzel', 'um-LN/', 'Google Favicon', 'Datanyze', 'BLEXBot/', '360Spider',
'adscanner/', 'HeadlessChrome', 'wpif', 'startmebot/', 'Googlebot/', 'Applebot/',
'facebookexternalhit/', 'GoogleImageProxy', 'bingbot/', 'heritrix/', 'ldspider'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
System::httpExit(403, 'Bots are not allowed');
}
}
// This switch here is just meant for developers who want to add more bots to the list above
if (!Config::get('blockbot', 'training')) {
return;
}
$crawlerDetect = new CrawlerDetect();
if (!$crawlerDetect->isCrawler()) { if (!$crawlerDetect->isCrawler()) {
logger::debug('Good user agent detected', $logdata); logger::debug('Good user agent detected', $logdata);
return; return;
@ -52,20 +72,6 @@ function blockbot_init_1(App $a) {
} }
} }
// 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.
$agents = ['SEMrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
'Diffbot/', 'Twitterbot/', 'YisouSpider/', 'evc-batch/', 'LivelapBot/', 'TrendsmapResolver/',
'PaperLiBot/', 'Nuzzel', 'um-LN/', 'Google Favicon', 'Datanyze', 'BLEXBot/', '360Spider',
'adscanner/', 'HeadlessChrome', 'wpif', 'startmebot/', 'Googlebot/', 'Applebot/',
'facebookexternalhit/', 'GoogleImageProxy', 'bingbot/', 'heritrix/', 'ldspider'];
foreach ($agents as $agent) {
if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
System::httpExit(403, ['title' => 'Bots are not allowed']);
}
}
logger::info('Blocked bot', $logdata); logger::info('Blocked bot', $logdata);
System::httpExit(403, ['title' => 'Bots are not allowed']); System::httpExit(403, 'Bots are not allowed');
} }