Merge pull request #1218 from annando/morebots
Blockbot: add more bots to the allow and deny listspull/1220/head
commit
8495420eb7
|
@ -50,7 +50,7 @@ function blockbot_init_1(App $a) {
|
||||||
// List of "good" crawlers
|
// List of "good" crawlers
|
||||||
$good_agents = ['fediverse.space crawler', 'fediverse.network crawler', 'Active_Pods_CheckBot_3.0',
|
$good_agents = ['fediverse.space crawler', 'fediverse.network crawler', 'Active_Pods_CheckBot_3.0',
|
||||||
'Social-Relay/', 'Test Certificate Info', 'Uptimebot/', 'GNUSocialBot', 'UptimeRobot/',
|
'Social-Relay/', 'Test Certificate Info', 'Uptimebot/', 'GNUSocialBot', 'UptimeRobot/',
|
||||||
'PTST/', 'Zabbix'];
|
'PTST/', 'Zabbix', 'Poduptime/'];
|
||||||
|
|
||||||
// List of known crawlers.
|
// List of known crawlers.
|
||||||
$agents = ['SemrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
|
$agents = ['SemrushBot', 's~feedly-nikon3', 'Qwantify/Bleriot/', 'ltx71', 'Sogou web spider/',
|
||||||
|
@ -75,7 +75,7 @@ function blockbot_init_1(App $a) {
|
||||||
'Google-Apps-Script; beanserver;', 'woorankreview/', 'Seekport Crawler;', 'AHC/',
|
'Google-Apps-Script; beanserver;', 'woorankreview/', 'Seekport Crawler;', 'AHC/',
|
||||||
'SkypeUriPreview Preview/', 'Semanticbot/', 'Embed PHP library', 'XoviOnpageCrawler;',
|
'SkypeUriPreview Preview/', 'Semanticbot/', 'Embed PHP library', 'XoviOnpageCrawler;',
|
||||||
'GetHPinfo.com-Bot/', 'BoardReader Favicon Fetcher', 'Google-Adwords-Instant', 'newspaper/',
|
'GetHPinfo.com-Bot/', 'BoardReader Favicon Fetcher', 'Google-Adwords-Instant', 'newspaper/',
|
||||||
'YurichevBot/', 'Crawling at Home Project'];
|
'YurichevBot/', 'Crawling at Home Project', 'InfoTigerBot/'];
|
||||||
|
|
||||||
if (!DI::config()->get('blockbot', 'good_crawlers')) {
|
if (!DI::config()->get('blockbot', 'good_crawlers')) {
|
||||||
$agents = array_merge($agents, $good_agents);
|
$agents = array_merge($agents, $good_agents);
|
||||||
|
@ -117,7 +117,8 @@ function blockbot_init_1(App $a) {
|
||||||
'Dispatch/', 'Ruby', 'Java/', 'libwww-perl/', 'Mastodon/', 'FeedlyApp/',
|
'Dispatch/', 'Ruby', 'Java/', 'libwww-perl/', 'Mastodon/', 'FeedlyApp/',
|
||||||
'lua-resty-http/', 'Tiny Tiny RSS/', 'Wget/', 'PostmanRuntime/',
|
'lua-resty-http/', 'Tiny Tiny RSS/', 'Wget/', 'PostmanRuntime/',
|
||||||
'W3C_Validator/', 'NetNewsWire', 'FeedValidator/', 'theoldreader.com', 'axios/',
|
'W3C_Validator/', 'NetNewsWire', 'FeedValidator/', 'theoldreader.com', 'axios/',
|
||||||
'Paw/', 'PeerTube/', 'fedi.inex.dev', 'FediDB/', 'index.community crawler'];
|
'Paw/', 'PeerTube/', 'fedi.inex.dev', 'FediDB/', 'index.community crawler',
|
||||||
|
'Slackbot-LinkExpanding'];
|
||||||
|
|
||||||
if (DI::config()->get('blockbot', 'good_crawlers')) {
|
if (DI::config()->get('blockbot', 'good_crawlers')) {
|
||||||
$agents = array_merge($agents, $good_agents);
|
$agents = array_merge($agents, $good_agents);
|
||||||
|
|
Loading…
Reference in New Issue