geerlingguy · May 7, 2014 18:48 · May 7, 2014
diff --git a/crawler_detect.php b/crawler_detect.php
@@ -0,0 +1,35 @@
+<?php
+
+/**
+ * Check if the given user agent string is one of a crawler, spider, or bot.
+ *
+ * @param string $user_agent
+ *   A user agent string (e.g. Googlebot/2.1 (+http://www.google.com/bot.html))
+ *
+ * @return bool
+ *   TRUE if the user agent is a bot, FALSE if not.
+ */
+function smart_ip_detect_crawler($user_agent) {
+  // User lowercase string for comparison.
+  $user_agent = strtolower($_SERVER['HTTP_USER_AGENT']);
+
+  // A list of some common words used only for bots and crawlers.
+  $bot_identifiers = array(
+    'bot',
+    'slurp',
+    'crawler',
+    'spider',
+    'curl',
+    'facebook',
+    'fetch',
+  );
+
+  // See if one of the identifiers is in the UA string.
+  foreach ($bot_identifiers as $identifier) {
+    if (strpos($user_agent, $identifier) !== FALSE) {
+      return TRUE;
+    }
+  }
+
+  return FALSE;
+}