New Crawler: SapphireWebCrawler/Nutch-1.0-dev (Sapphire Web Crawler using Nutch;...

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

diff --git a/lib/core.inc b/lib/core.inc
index c7dbbf2..cce4379 100644 (file)
--- a/lib/core.inc
+++ b/lib/core.inc
@@ -146,6 +146,18 @@ function is_spider()
       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'], 'Cityreview Robot') !== false) {
     if (isset($_SESSION['uid']))
       $_SESSION['robot'] = true;