primary_host.rb in wmap-2.5.6

- old
+ new

@@ -14,23 +14,21 @@
 	# 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
 
-		attr_accessor :hosts_file, :verbose, :data_dir
-		attr_reader :known_hosts, :known_ips
+		attr_accessor :hosts_file, :verbose, :data_dir, :known_hosts, :known_ips
 
 		# Initialize the instance variables
 		def initialize (params = {})
 			@verbose=params.fetch(:verbose, false)
       @data_dir=params.fetch(:data_dir, File.dirname(__FILE__)+'/../../../data/')
 			# Set default instance variables
-			@file_hosts=@data_dir + 'prime_hosts'
-			file=params.fetch(:hosts_file, @file_hosts)
+			@hosts_file=params.fetch(:hosts_file, @data_dir + 'prime_hosts')
 			# Initialize the instance variables
-      File.write(@file_hosts, "") unless File.exist?(@file_hosts)
-			@known_hosts=load_known_hosts_from_file(file)
+      File.write(@hosts_file, "") unless File.exist?(@hosts_file)
+			@known_hosts=load_known_hosts_from_file(@hosts_file)
 			@known_ips=Hash.new
 			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.
@@ -39,11 +37,12 @@
         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.instance
-        my_tracker.data_dir = @data_dir
+        my_tracker.sites_file = @data_dir + "sites"
+        my_tracker.load_site_stores_from_file
 				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}"
@@ -75,12 +74,13 @@
 		# 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
 				my_tracker=Wmap::SiteTracker.instance
-        my_tracker.data_dir=@data_dir
+        my_tracker.sites_file=@data_dir + "sites"
+        my_tracker.load_site_stores_from_file
         urls = my_tracker.get_redirection_urls
-        my_tracker = nil 
+        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)