Add more bots

[infodrom.org/www.zeitungsliste.de] / lib / core.inc

diff --git a/lib/core.inc b/lib/core.inc
index c7dbbf2..05680a5 100644 (file)
--- a/lib/core.inc
+++ b/lib/core.inc
@@ -141,11 +141,74 @@ function is_spider()
       strpos($_SERVER['HTTP_USER_AGENT'], 'holmes/3.12') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'NoteworthyBot/0.1') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'Eurosoft-Bot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Eurobot/1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Yandex') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'msnbot') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'Twiceler-0.9') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'ScoutJet') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'penthesilea/0.3') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'VisBot/2.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Netluchs/Nutch-1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Yanga WorldSearch Bot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Semager') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'DotBot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Baiduspider') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'MJ12bot/') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Yahoo-MMCrawler') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Charlotte/1.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Gaisbot/3.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'psbot/0.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Ask Jeeves/Teoma') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SapphireWebCrawler/Nutch') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'kalooga/KaloogaBot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SimilarPages/Nutch-1.0-dev') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Cogentbot/1.') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'CatchBot/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'GingerCrawler/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'yacybot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'www.yacy.net') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Nutch-1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Tagoobot/3.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SapphireWebCrawler/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'BotOnParade') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SurveyBot/2.3') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'XmarksFetch/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'spbot/2.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'TinEye/1.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'TurnitinBot/2.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'CligooRobot/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'libwww-perl') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Wget') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Python-urllib') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'CamontSpider/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'TwengaBot-Discover') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Hailoobot/1.2') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'bingbot/2.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Exabot/3.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'spbot/2.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Search17Bot/1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Linguee Bot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'findlinks/2') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SiteBot/0.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Purebot/1.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Ezooms/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'discobot/1.1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'AhrefsBot/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'suggybot v0.01a') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'DomainCrawler/2.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Plukkie/1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], '/Nutch-') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'aiHitBot/1.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'AhrefsBot/') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Crawler') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'Acoon') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'aiHitBot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SeznamBot') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'CareerBot/1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SiteExplorer/1') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'discoverybot/2.0') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'SemrushBot/') !== false ||
+      strpos($_SERVER['HTTP_USER_AGENT'], 'archive.org_bot') !== false ||
       strpos($_SERVER['HTTP_USER_AGENT'], 'Cityreview Robot') !== false) {
     if (isset($_SESSION['uid']))
       $_SESSION['robot'] = true;
@@ -198,4 +261,4 @@ function format_info_bookmarks()
   return $ret;
 }
 
-?>
\ No newline at end of file
+?>