Fixed #153

Author: Rafael Grigorian

src/app/code/community/JetRails/Cloudflare/Helper/Data.php

@@ -23,6 +23,22 @@ class JetRails_Cloudflare_Helper_Data extends Mage_Core_Helper_Abstract {
 		const XPATH_AUTH_ZONE = "cloudflare/configuration/auth_zone";
 		const XPATH_AUTH_TOKEN = "cloudflare/configuration/auth_token";
 
+		/**
+		 * This method returns the public suffix list either from cache based on
+		 * the user session, or downloads it directly
+		 * @return  array[string]                      Public Suffix List
+		 */
+		public function refreshPSL () {
+			$session = Mage::getSingleton ("core/session");
+			$data = $session->getPSL ();
+			if ( $data == null ) {
+				$psl = Mage::helper ("cloudflare/publicSuffixList");
+				$data = array_values ( $psl->getPSL () );
+				$session->setPSL ( $data );
+			}
+			return $data;
+		}
+
 		/**
 		 * This method gets the value for the authorization zone and decrypts
 		 * it. It then returns that result to the caller.
@@ -73,13 +89,12 @@ public function getAuthToken () {
 		 */
 		public function getDomainName () {
 			$session = Mage::getSingleton ("core/session");
+			$psl = Mage::helper ("cloudflare/publicSuffixList");
 			if ( !empty ( $session->getCloudflareSelectedDomain () ) ) {
 				return $session->getCloudflareSelectedDomain ();
 			}
 			$domain = Mage::getBaseUrl ( Mage_Core_Model_Store::URL_TYPE_WEB );
-			$domain = parse_url ( $domain ) ["host"];
-			preg_match ( "/([^.\s]+\.([^.\s]{3,}|[^.\s]{2}\.[^.\s]{2}|[^.\s]{2}))\b$/im", $domain, $matches );
-			return $matches [ 1 ];
+			return $psl->extract ( $domain, $this->refreshPSL () ) ["root_domain"];
 		}
 
 		/**
@@ -91,15 +106,15 @@ public function getDomainName () {
 		 * @return  array                             All domains for all stores
 		 */
 		public function getDomainNames () {
+			$psl = Mage::helper ("cloudflare/publicSuffixList");
 			$selection = $this->getDomainName ();
 			$domains = array ();
+			$pslData = $this->refreshPSL ();
 			foreach ( Mage::app ()->getWebsites () as $website ) {
 				foreach ( $website->getGroups () as $group ) {
 					$stores = $group->getStores ();
 					foreach ( $stores as $store ) {
-						$domain = parse_url ( $store->getBaseUrl () ) ["host"];
-						preg_match ( "/([^.\s]+\.([^.\s]{3,}|[^.\s]{2}\.[^.\s]{2}|[^.\s]{2}))\b$/im", $domain, $matches );
-						$domain = $matches [ 1 ];
+						$domain = $psl->extract ( $store->getBaseUrl (), $pslData ) ["root_domain"];
 						array_push ( $domains, $domain );
 					}
 				}

src/app/code/community/JetRails/Cloudflare/Helper/PublicSuffixList.php

@@ -0,0 +1,61 @@
+<?php
+
+	/**
+	 * This helper exists to download a list of domain suffixes that are used to then determine the
+	 * effective TLD and the root domain of a hostname.
+	 * @version     1.2.4
+	 * @package     JetRails® Cloudflare
+	 * @author      Rafael Grigorian <development@jetrails.com>
+	 * @copyright   © 2018 JETRAILS, All rights reserved
+	 * @license     MIT https://opensource.org/licenses/MIT
+	 */
+    class JetRails_Cloudflare_Helper_PublicSuffixList extends Mage_Core_Helper_Abstract {
+
+		protected function _endsWith ( $haystack, $needle ) {
+			$length = strlen ( $needle );
+			return $length > 0 ? substr ( $haystack, - $length ) === $needle : true;
+		}
+
+		public function getPSL () {
+			error_log ("Downloading PSL...");
+			$handle = curl_init ();
+			curl_setopt ( $handle, CURLOPT_URL, "https://publicsuffix.org/list/public_suffix_list.dat" );
+			curl_setopt ( $handle, CURLOPT_RETURNTRANSFER, 1 );
+			curl_setopt ( $handle, CURLOPT_TIMEOUT, 3 );
+			$output = curl_exec ( $handle );
+			$code = curl_getinfo ( $handle, CURLINFO_HTTP_CODE );
+			curl_close ( $handle );
+			if ( $code == 200 ) {
+				$lines = explode ( "\n", $output );
+				$lines = array_filter ( $lines, function ( $line ) { return $line != "" && !in_array ( $line [0], [ "/", " ", "\t" ] ); } );
+				$lines = array_map ( function ( $line ) { return trim ( $line, ".!* \t" ); }, $lines );
+				return $lines;
+			}
+			return [];
+		}
+
+		public function extract ( $url, $suffixes = null ) {
+			$host_name = parse_url ( $url ) ["host"];
+			$max_segments = null;
+			$best_match = null;
+			if ( $suffixes == null ) {
+				$suffixes = $this->getPSL ();
+			}
+			foreach ( $suffixes as $suffix ) {
+				if ( $this->_endsWith ( $host_name, ".$suffix" ) ) {
+					$count = substr_count ( $suffix, "." );
+					if ( $max_segments == null || $count > $max_segments ) {
+						$max_segments = $count;
+						$best_match = $suffix;
+					}
+				}
+			}
+			$root_domain = implode ( ".", array_slice ( explode ( ".", $host_name ), -1 * ( $max_segments + 2 ) ) );
+			return [
+				"host_name" => $host_name,
+				"root_domain" => $root_domain,
+				"effective_tld" => $best_match ? $best_match : "",
+			];
+		}
+
+	}