[Plugin: PC Robots.txt] Secure robots.txt

Hey!

When I started to use your plugin I simply loved it, but I kinda wanted to hide it and all its contents revealing that it’s a WordPress site to the general user browsing with one of the common browsers.
First I had the idea of doing it via .htaccess but that would have meant a real robots.txt file had to be put in the root folder of the site in order for Apache’s mod_rewrite to redirect to another custom 404 php page. So I came up with the idea of building this feature directly into your plugin.

What it basically does is check whether or not the user is a bot or a casual user by checking the user agent and if it is a bot it creates the robots.txt as it should and if it’s a user it makes a 404 and sends the user back to the homepage. The user agents are based on this site here: http://www.useragentstring.com/pages/useragentstring.php

Here is the DIFF:

--- pc-robotstxt.php
+++ pc-robotstxt.php
@@ -74,24 +74,45 @@

 		if ( is_robots() ) {

-			$options = $this->get_options();
-
-			$output = "# This virtual robots.txt file was created by the PC Robots.txt WordPress plugin.\n";
-			$output .= "# For more info visit: http://petercoughlin.com/robotstxt-wordpress-plugin/\n\n";
-
-			if ( '' != $options['user_agents'] )
-				$output .= stripslashes($options['user_agents']);
-
-			// if there's an existing sitemap file or we're using pc-xml-sitemap plugin add a reference..
-			if ( file_exists($_SERVER['DOCUMENT_ROOT'].'/sitemap.xml.gz') )
-				$output .= "\n\n".'Sitemap: http://'.$_SERVER['HTTP_HOST'].'/sitemap.xml.gz';
-			elseif ( class_exists('pc_xml_sitemap') || file_exists($_SERVER['DOCUMENT_ROOT'].'/sitemap.xml') )
-				$output .= "\n\n".'Sitemap: http://'.$_SERVER['HTTP_HOST'].'/sitemap.xml';
-
-			header('Status: 200 OK', true, 200);
-			header('Content-type: text/plain; charset='.get_bloginfo('charset'));
-			echo $output;
-			exit;
+			$rawData = $_SERVER['HTTP_USER_AGENT'];
+			$trueBrowsers = array('Mozilla','Opera','Links','Lynx','Nokia','Samsung','MOT','SonyEricsson','Doris','HTC','Bunjalloo','PSP','wii','Amiga','ELinks','Cyberdog','Dillo','Dooble','Enigma','Galaxy','HotJava','IBM','LeechCraft','NCSA','NetSurf','retawq','Surf','Webkit','Uzbl','Vimprobable','w3m','WorldWideweb');
+			$mozBots = array('Googlebot','Yahoo! Slurp','Ask Jeeves', 'Twiceler','Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 4.0)');
+			foreach($trueBrowsers as $key => $search_browser) {
+			   if(stristr($rawData, $search_browser) == TRUE) {
+				if(stristr($rawData, $search_browser) == 'Mozilla') {
+					foreach($mozBots as $key => $search_moz_bot) {
+						if(stristr($rawData, $search_moz_bot) == TRUE) {
+							$isBrowser = FALSE;
+						}
+					}
+				}
+				else {
+				$isBrowser = TRUE;
+				}
+			   }
+			}
+			if ($isBrowser == FALSE) {
+				$options = $this->get_options();
+
+				if ( '' != $options['user_agents'] )
+					$output .= stripslashes($options['user_agents']);
+
+				// if there's an existing sitemap file or we're using pc-xml-sitemap plugin add a reference..
+				if ( file_exists($_SERVER['DOCUMENT_ROOT'].'/sitemap.xml.gz') )
+					$output .= "\n\n".'Sitemap: http://'.$_SERVER['HTTP_HOST'].'/sitemap.xml.gz';
+				elseif ( class_exists('pc_xml_sitemap') || file_exists($_SERVER['DOCUMENT_ROOT'].'/sitemap.xml') )
+					$output .= "\n\n".'Sitemap: http://'.$_SERVER['HTTP_HOST'].'/sitemap.xml';
+
+				header('Status: 200 OK', true, 200);
+				header('Content-type: text/plain; charset='.get_bloginfo('charset'));
+				echo $output;
+				exit;
+			}
+			else {
+				header('HTTP/1.0 404 Not Found', true, 404);
+				header('Location: '.site_url());
+				exit();
+			}

 		}// end if

http://wordpress.org/extend/plugins/pc-robotstxt/