lib/wmap/host_tracker/primary_host.rb in wmap-2.4.6 vs lib/wmap/host_tracker/primary_host.rb in wmap-2.4.8

@@ -3,20 +3,20 @@
 #
 # A pure Ruby library for Internet web application discovery and tracking.
 #
 # Copyright (c) 2012-2015 Yang Li <[e-mail]>
 #++
-#require "singleton"		# Implement singleton pattern to avoid race condition under parallel engine
+require "singleton"		# Implement singleton pattern to avoid race condition under parallel engine
 
 
 module Wmap
   class HostTracker
 
 	# Class to differentiate the primary host-name from the potential aliases. This is needed in order to minimize the confusion on our final site inventory list, as it contains a large number of duplicates (aliases). More specifically, a filter could be built by using this class to track the primary url of a website.
 	class PrimaryHost < Wmap::HostTracker
 		include Wmap::Utils
-		#include Singleton
+		include Singleton
 
 		attr_accessor :hosts_file, :verbose, :data_dir
 		attr_reader :known_hosts, :known_ips
 
 		# Initialize the instance variables
@@ -33,16 +33,17 @@
 			de_duplicate
 		end
 
 		# Procedures to identify primary host-name from the site store SSL certificates. The assumption is that the CN used in the cert application must be primary hostname and used by the users.
 		def update_from_site_store!
-			puts "Invoke internal procedures to update the primary host-name table from the site store."
-			begin
-				# Step 1 - update the prime host table based on the SSL cert CN fields
+			#begin
+        puts "Invoke internal procedures to update the primary host-name table from the site store."
+        # Step 1 - update the prime host table based on the SSL cert CN fields
 				cns=Hash.new
 				checker=Wmap::UrlChecker.new(:data_dir=>@data_dir)
-        my_tracker = Wmap::SiteTracker.new(:data_dir=>@data_dir)
+        my_tracker = Wmap::SiteTracker.instance
+        my_tracker.data_dir = @data_dir
 				my_tracker.get_ssl_sites.map do |site|
 					puts "Exam SSL enabled site entry #{site} ..."
 					my_host=url_2_host(site)
 					next if @known_hosts.key?(my_host) # add the logic to optimize the process
 					puts "Pull SSL cert details on site: #{site}"
@@ -60,23 +61,26 @@
 				end
 				# Step 2 - Save the cache into the file
 				self.save!
         checker=nil
         my_tracker=nil
-			rescue Exception => ee
-				puts "Exception on method #{__method__}: #{ee}" if @verbose
-        checker=nil
-        my_tracker=nil
-				return nil
-			end
+			#rescue Exception => ee
+			#	puts "Exception on method #{__method__}: #{ee}" if @verbose
+      #  checker=nil
+      #  my_tracker=nil
+			#	return nil
+			#end
 		end
 		alias_method :update!, :update_from_site_store!
 
 		# Procedures to identify primary host-name from the site store redirection URLs. The assumption is that on site redirection, it must be directed to the well known primary site.
 		def update_from_site_redirections!
 			puts "Invoke internal procedures to update the primary host-name table from the site store."
 			begin
-				urls=Wmap::SiteTracker.new(:data_dir=>@data_dir).get_redirection_urls
+				my_tracker=Wmap::SiteTracker.instance
+        my_tracker.data_dir=@data_dir
+        urls = my_tracker.get_redirection_urls
+        my_tracker = nil 
 				urls.map do |url|
 					if is_url?(url)
 						host=url_2_host(url)
 						if is_fqdn?(host)
 							ip=host_2_ip(host)