X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;ds=sidebyside;f=blockbot%2Fblockbot.php;h=0ff372c502862a3b8f1a4093f5f99cf9acaf2f40;hb=cff3dd67b065983dcbe13858a2a31f2de9992659;hp=b602bf63bbc7ac86030b8fd27232f4a017b1b90d;hpb=05c98064b4d624fcdf5e7481f66704056da007ea;p=friendica-addons.git

diff --git a/blockbot/blockbot.php b/blockbot/blockbot.php
index b602bf63..0ff372c5 100644
--- a/blockbot/blockbot.php
+++ b/blockbot/blockbot.php
@@ -15,6 +15,7 @@ use Friendica\DI;
 use Jaybizzle\CrawlerDetect\CrawlerDetect;
 use Friendica\Core\Logger;
 use Friendica\Core\Renderer;
+use Friendica\Network\HTTPException\ForbiddenException;
 
 require_once __DIR__ . DIRECTORY_SEPARATOR . 'vendor' . DIRECTORY_SEPARATOR . 'autoload.php';
 
@@ -49,7 +50,7 @@ function blockbot_init_1(App $a) {
 	// List of "good" crawlers
 	$good_agents = ['fediverse.space crawler', 'fediverse.network crawler', 'Active_Pods_CheckBot_3.0',
 		'Social-Relay/', 'Test Certificate Info', 'Uptimebot/', 'GNUSocialBot', 'UptimeRobot/',
-		'PTST/'];
+		'PTST/', 'Zabbix'];
 
 	// List of known crawlers.
 	$agents = ['SemrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
@@ -57,7 +58,7 @@ function blockbot_init_1(App $a) {
 		'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/',
+		'AwarioRssBot/', 'TweetmemeBot/', 'dcrawl/', 'PhantomJS/', 'Googlebot-Image/',
 		'CrowdTanglebot/', 'Mediapartners-Google', 'Baiduspider/', 'datagnionbot',
 		'MegaIndex.ru/', 'SMUrlExpander', 'Hatena-Favicon/', 'Wappalyzer', 'FlipboardProxy/',
 		'NetcraftSurveyAgent/', 'Dataprovider.com', 'SMTBot/', 'Nimbostratus-Bot/',
@@ -73,7 +74,8 @@ function blockbot_init_1(App $a) {
 		'SummalyBot/', 'DNSResearchBot/', 'PetalBot;', 'Nmap Scripting Engine;',
 		'Google-Apps-Script; beanserver;', 'woorankreview/', 'Seekport Crawler;', 'AHC/',
 		'SkypeUriPreview Preview/', 'Semanticbot/', 'Embed PHP library', 'XoviOnpageCrawler;',
-		'GetHPinfo.com-Bot/', 'BoardReader Favicon Fetcher'];
+		'GetHPinfo.com-Bot/', 'BoardReader Favicon Fetcher', 'Google-Adwords-Instant', 'newspaper/',
+		'YurichevBot/', 'Crawling at Home Project'];
 
 	if (!DI::config()->get('blockbot', 'good_crawlers')) {
 		$agents = array_merge($agents, $good_agents);
@@ -91,7 +93,7 @@ function blockbot_init_1(App $a) {
 
 	foreach ($agents as $agent) {
 		if (stristr($_SERVER['HTTP_USER_AGENT'], $agent)) {
-			System::httpExit(403, 'Bots are not allowed');
+			throw new ForbiddenException('Bots are not allowed');
 		}
 	}
 
@@ -114,7 +116,8 @@ function blockbot_init_1(App $a) {
 		'WordPress/', 'http.rb/', 'Apache-HttpClient/', 'WordPress.com;', 'Pleroma',
 		'Dispatch/', 'Ruby', 'Java/', 'libwww-perl/', 'Mastodon/', 'FeedlyApp/',
 		'lua-resty-http/', 'Tiny Tiny RSS/', 'Wget/', 'PostmanRuntime/',
-		'W3C_Validator/', 'NetNewsWire', 'FeedValidator/', 'theoldreader.com'];
+		'W3C_Validator/', 'NetNewsWire', 'FeedValidator/', 'theoldreader.com', 'axios/',
+		'Paw/', 'PeerTube/', 'fedi.inex.dev', 'FediDB/', 'index.community crawler'];
 
 	if (DI::config()->get('blockbot', 'good_crawlers')) {
 		$agents = array_merge($agents, $good_agents);
@@ -128,5 +131,5 @@ function blockbot_init_1(App $a) {
 	}
 
 	logger::info('Blocked bot', $logdata);
-	System::httpExit(403, 'Bots are not allowed');
+	throw new ForbiddenException('Bots are not allowed');
 }