Detect more robots

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

diff --git a/lib/core.inc b/lib/core.inc
index d7be078..9a39761 100644 (file)
--- a/lib/core.inc
+++ b/lib/core.inc
@@ -141,7 +141,7 @@ 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/1.1') !== 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 ||
@@ -168,6 +168,18 @@ function is_spider()
       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'], 'Cityreview Robot') !== false) {
     if (isset($_SESSION['uid']))
       $_SESSION['robot'] = true;