vizi_whois 0.3.0 → 0.4.0

data/pkg/vizi_whois-0.1.0/lib/vizi/vizi_whois.rb

@@ -41,7 +41,7 @@ module Vizi
             i = i+1
           end
         end
-      elsif response.index("whois.afrinic.net")
+      elsif response.index("afrinic.net")
         s = "afrinic"
         response = ruby_whois(ipaddr, s)
       elsif response.length >250 and response.length < 500
@@ -100,28 +100,31 @@ module Vizi
   
 # This class will parse the response string from a Whois server
   class Formatter
+  
+		def initialize
+  		@cities = YAML.load_file("config/cities.yml")
+  	end	
  
 # This method will parse response string 
-    def parse (contents, country, server)
+#    def parse (contents, country, server)
+    def parse (contents, country)
       name_list = ["COUNTRY","NETNAME","ORGNAME","ORG-NAME","CUSTNAME","OWNER","ADDRESS","DESCR","CITY","STATEPROV","POSTALCODE"]
       country_list = ["GERMANY","UNITED KINGDOM","UK","ENGLAND","INDIA","NETHERLANDS","SPAIN","BELGIUM","PR CHINA","NORWAY"]
-      odd_list = ["EMAIL","CONTA","PHONE","ABUSE ","SENT ","SEND ","DATE ","ATTN ","INFRA","THIS ","ANTIA","*****","-----"]
-
+      odd_words = ["ABUSE", "PHISH", "SPAM", "ACTIVITY","PROVINCE NETWORK","COMMUNICATION DIVISION","IMPROPER","SERVICE PROVIDER","IP POOL","-----","*****","#####","IP ASSIGNMENT","FOR CUSTOMER","THIS SPACE"]
+			trim_orgname = ["LIMITED", "LTD", "INC", "CORPORATION", "LLC", "HOLDINGS", "UNICATIONS", "INFORMATION CENTER"]
 #      formatted_file = File.open('./log/formatted.log', 'a')
-
       a = contents.upcase.split(/\n/) 
       i = -1
       line = [nil,nil]
       lastname = nil
+      orgname = nil
+      cityflag = false
       addressflag = false
       country_count = 0
       da_count = 0
       netname_count = 0
       darray = []
-      outstring = "" 
-#      formatted_file.puts '+ ------------------------------------------------------'
-#      formatted_file.puts '+ '+ country + ' - ' + server.upcase
-#      formatted_file.puts '+ ------------------------------------------------------'         
+      outstring = ""         
       while i < a.length-1
         i = i + 1
         line = [nil, nil]
@@ -130,59 +133,142 @@ module Vizi
         if current.length < 2
           goodline = false        
         else
-		  goodline = false if current[0] == "%" or current[0] == "#"
+					goodline = false if current[0] == "%" or current[0] == "#"
         end
         if goodline
-          line = current.split(":")
-          if line[1].nil?
-            line[1] = "" 
-            if lastname == "ADDRESS" and line[0] != line[0].lstrip
-              line[0] = lastname
+          line = current.split(":")    
+          line_name = line[0]
+          line_value = line[1]
+          line_value = line_value.strip if not line_value.nil?
+          if line_value.nil?
+            line_value = "" 
+            if lastname == "ADDRESS" and line_name != line_name.lstrip
+              line_name = lastname
             end         
           else
-						addressflag = true if line[0] == "ADDRESS"
-          end 
-					line[1] = line[1].lstrip.gsub(".","")
-					line[1] = line[1].lstrip.gsub(", ",",")
-					line[1] = line[1].lstrip.gsub(" ,",",")
-					line[1] = line[1].chomp(",")	  		            
-          country_count = country_count + 1 if line[0] == "COUNTRY"
-          da_count = da_count + 1 if line[0] == "DESCR" or line[0] == "ADDRESS"
-          netname_count = netname_count + 1 if line[0] == "NETNAME"          
-          goodline = false if da_count > 10 or country_count > 4 or netname_count > 1 
-          goodline = false if country_count > 1 and line[0] == "COUNTRY"                
-          goodline = false if line[1][0] == "-"
-          goodline = false if country_list.index(line[1])
-          goodline = false if line[0] == "ADDRESS" and line[1] == country
-          goodline = false if odd_list.index(line[1][0..4]) 
-          goodline = false if addressflag and line[0] == "DESCR"
-          goodline = false if line[1].length < 2
-          goodline = false if darray.index(line[1])
-          darray << line[1] if goodline and (line[0] == "DESCR" or line[0] == "ADDRESS")
-          if goodline
-            country_count = country_count + 1 if line[0] == "COUNTRY"
-            da_count = da_count + 1 if line[0] == "DESCR" or line[0] == "ADDRESS"
-          end                            
+						addressflag = true if line_name == "ADDRESS"
+          end
+          odd_words.each {|w|
+						goodline = false if line_value.index(w)			
+					}	          
+            		             
+          if line_name == "COUNTRY"
+						country_count = country_count + 1
+						line_value = line_value[0..1] if not line_value.nil?          
+          end
+          if line_name == "NETNAME"
+						netname_count = netname_count + 1 
+						netname = line_value if netname_count == 1
+          end
+
+					if not ["NETNAME","POSTALCODE"].index(line_name)
+						line_value = line_value.lstrip.gsub(",",";")
+						line_value = line_value.lstrip.gsub(".","")
+						line_value = line_value.lstrip.gsub("; ",";")			
+						line_value = line_value.lstrip.gsub(" ;",";")
+						line_value = line_value.lstrip.gsub("/",";")
+#						line_value = line_value.lstrip.gsub("-",";")																		
+						line_value = line_value.chomp(";")
+					end
+          if ["ORGNAME","CUSTNAME","OWNER","ORG-NAME"].index(line_name)
+						orgname = (line_value + ";").split(";")[0]
+					end					
+          if line_name == "DESCR" or line_name == "ADDRESS"
+						goodline = false if line_value.length == 0
+						goodline = false if da_count > 6          
+            goodline = false if darray.index(line_value)         
+						if goodline
+							darray << line_value
+							da_count = da_count + 1
+						end
+					end
+						          
+          goodline = false if country_count > 1 and line_name == "COUNTRY"
+          goodline = false if netname_count > 1 and line_name == "NETNAME"                          
+          goodline = false if line_value[0] == "-"
+          goodline = false if country_list.index(line_value)
+          goodline = false if line_name == "ADDRESS" and line_value == country
+          goodline = false if addressflag and line_name == "DESCR"
+          goodline = false if line_value.length < 2 
+
+          cityflag = true if line_name == "CITY" and line_value.length > 0
         else
-          line[0] = current
-          line[1] = ""
+          line_name = current
+          line_value = ""
         end         
-        lastname = line[0]  
+        lastname = line_name  
         if goodline
-          if name_list.index(line[0])
-            newline = line[0] + ': ' + line[1] + '\n'
-#            formatted_file.puts '+ ' + newline
-            outstring = outstring + newline
-          else
-#            formatted_file.puts '. ' + line[0] + ':: ' + line[1]         
+          if name_list.index(line_name)
+            newline = line_name + ': ' + line_value + '\n'
+            outstring = outstring + newline       
           end
-        else          
-#          formatted_file.puts '- ' + line[0]
         end      
+      end 
+      
+#     At the end of regular parse, find organization NAME if needed    
+			if orgname.nil?
+				if darray[0].nil?
+					orgname = 'NOT FOUND'
+				else
+					orgname = (darray[0] + ";").split(";")[0]				
+				end
+			end     
+       
+			if country != "CA" and country != "US" and not orgname[/[0123456789]/].nil? 
+				orgname = netname if not netname.nil?
+			end 
+      
+			trim_orgname.each {|p|
+				orgname = orgname.sub(p, "")
+			}
+			orgname = orgname.sub(".","")
+			orgname = orgname[0..29]
+			orgname = orgname[0..28-orgname.reverse.index(" ")] if not orgname.reverse.index(" ").nil?    
+      
+			newline = 'NAME: ' + orgname + '\n'
+			outstring = outstring + newline
+
+# 		Search for city name, if not identified  		
+      if country != "CA" and country != "US"
+				if not cityflag             
+					candidates = @cities["."+country]
+					city = "UNKNOWN"
+					if not darray[0].nil? and not candidates.nil?
+						darray.each {|line|
+							candidates.each {|c|
+								city = c.delete("!") if c.class != Hash and line.index(c.delete("!"))
+								break if city != 'UNKNOWN'
+							}
+						}
+					end
+					if city == "UNKNOWN" and not candidates.nil?
+						candidates.each {|c|
+							if c.class == String
+								city = c.delete("!") if c.index("!")
+								break if city != 'UNKNOWN'
+							end
+						}
+					end
+					newline = 'CITY: ' + city + '\n'
+					outstring = outstring + newline       
+				end
       end
-#      formatted_file.puts '+ ' + country_count.to_s + ': ' + da_count.to_s
+      
       return outstring         
     end
+    
+    def createhash(formatted)
+			farray = formatted.split('\n')
+			hashout = Hash.new
+			farray.each {|x|
+	      line = x.split(': ')		
+			  hashout[line[0]] = line[1]
+			}
+			hashout['NAME'] = '??' if hashout['NAME'].nil?	
+			hashout['CITY'] = '??' if hashout['CITY'].nil?				
+			hashout['COUNTRY'] = '??' if hashout['COUNTRY'].nil?						
+			hashout
+    end
   end
       
 end

data/testit.rb

@@ -5,11 +5,13 @@
 #
 # Author::    Al Kivi <al.kivi@vizitrax.com>
 
-## require './lib/vizi_whois'
+#require 'c:\rails\vizi_whois\lib\vizi_whois'
+require 'rubygems'   # needed for ruby 1.8.7
 require 'vizi_whois'
 
 require 'logger'
 require 'socket'
+require 'yaml'
  
 syslog = Logger.new('./log/system.log',shift_age = 'weekly')
 syslog.info "Starting IP address test file ... >>> "+Time.now.to_s
@@ -19,10 +21,13 @@ out_file = File.new('./log/output.log', 'w')
 File.delete('./log/formatted.log') if File.exist?('./log/formatted.log') 
 parse_file = File.new('./log/formatted.log', 'w')
 
+p 'starting'
+
 # Open test file for reading
 File.open('./data/testfile.txt', 'r') do |file|
   rec_count = 0
   while(line = file.gets) # Read each line of the test file, one IP address per line
+  p line
     @whoisresult = Vizi::Gowhois.new
     p line.chomp
     rarray = @whoisresult.query(line.chomp)
@@ -34,7 +39,8 @@ File.open('./data/testfile.txt', 'r') do |file|
     out_file.puts '----------------------------------------------------------------'    
     out_file.puts @contents
     @result = Vizi::Formatter.new
-    @formatted = @result.parse(@contents, rarray[1], rarray[2]) 
+    @formatted = @result.parse(@contents, rarray[1]) 
+    @formatted = @formatted.gsub('\n','<br/>') 
     p @formatted   
     rec_count = rec_count + 1 
   end

data/vizi_whois.gemspec

@@ -2,7 +2,7 @@ require 'rubygems'
 
 spec = Gem::Specification.new do |s|
   s.name = 'vizi_whois'
-  s.version = '0.3.0'
+  s.version = '0.4.0'
   s.summary = "Global whois capability to select the right whois server and get response data"
   s.description = "This gem module provides a classes to find the right Regional Internet Registry
     for a given IP Address. The query method will navigate each major RIR until a response is found.
@@ -14,5 +14,5 @@ spec = Gem::Specification.new do |s|
   s.homepage = "http://www.vizitrax.com"
   s.email = "al.kivi@yahoo.com"
   s.has_rdoc = true
-  s.required_ruby_version = '>= 1.8.2'
+  s.required_ruby_version = '>= 1.8.7'
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: vizi_whois
 version: !ruby/object:Gem::Version 
-  hash: 19
+  hash: 15
   prerelease: 
   segments: 
   - 0
-  - 3
+  - 4
   - 0
-  version: 0.3.0
+  version: 0.4.0
 platform: ruby
 authors: 
 - Al Kivi
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-01-09 00:00:00 -05:00
+date: 2012-05-05 00:00:00 -04:00
 default_executable: 
 dependencies: []
 
@@ -32,6 +32,7 @@ extensions: []
 extra_rdoc_files: []
 
 files: 
+- vizi_whois-0.3.0.gem
 - doc/Visit.html
 - doc/testit_rb.html
 - doc/Rakefile.html
@@ -105,16 +106,15 @@ files:
 - build_gem3.bat
 - README.rdoc
 - Rakefile
-- log/system.log.20110903
 - log/system.log
 - log/output.log
-- log/system.log.20110820
-- log/formatted.log
 - test/parser_test.rb
 - test/test_helper.rb
+- config/cities.yml
 - testit.rb
 - build_gem2.bat
 - vizi_whois-0.2.0.gem
+- vizi_whois-0.4.0.gem
 has_rdoc: true
 homepage: http://www.vizitrax.com
 licenses: []
@@ -129,12 +129,12 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 51
+      hash: 57
       segments: 
       - 1
       - 8
-      - 2
-      version: 1.8.2
+      - 7
+      version: 1.8.7
 required_rubygems_version: !ruby/object:Gem::Requirement 
   none: false
   requirements: 

data/log/formatted.log

@@ -1,52 +0,0 @@
-+ ------------------------------------------------------
-+ US - ARIN
-+ ------------------------------------------------------
-+ NETNAME: NET74
-+ ORGNAME: AMERICAN REGISTRY FOR INTERNET NUMBERS
-+ ADDRESS: 3635 CONCORDE PARKWAY
-+ ADDRESS: SUITE 200
-+ CITY: CHANTILLY
-+ STATEPROV: VA
-+ POSTALCODE: 20151
-+ COUNTRY: US
-+ ------------------------------------------------------
-+ US - ARIN
-+ ------------------------------------------------------
-+ NETNAME: NET144
-+ ORGNAME: VARIOUS REGISTRIES (MAINTAINED BY ARIN)
-+ ADDRESS: 3635 CONCORD PARKWAY,SUITE 200
-+ CITY: CHANTILLY
-+ STATEPROV: VA
-+ POSTALCODE: 20151
-+ COUNTRY: US
-+ ------------------------------------------------------
-+ US - ARIN
-+ ------------------------------------------------------
-+ NETNAME: COMCAST-VOIP-3
-+ ORGNAME: COMCAST CABLE COMMUNICATIONS,LLC
-+ ADDRESS: 1800 BISHOPS GATE BLVD
-+ CITY: MT LAUREL
-+ STATEPROV: NJ
-+ POSTALCODE: 08054
-+ COUNTRY: US
-+ ------------------------------------------------------
-+ US - ARIN
-+ ------------------------------------------------------
-+ NETNAME: PRIVATE-ADDRESS-ABLK-RFC1918-IANA-RESERVED
-+ ORGNAME: INTERNET ASSIGNED NUMBERS AUTHORITY
-+ ADDRESS: 4676 ADMIRALTY WAY,SUITE 330
-+ CITY: MARINA DEL REY
-+ STATEPROV: CA
-+ POSTALCODE: 90292-6695
-+ COUNTRY: US
-+ ------------------------------------------------------
-+ US - ARIN
-+ ------------------------------------------------------
-+ NETNAME: NET74
-+ ORGNAME: AMERICAN REGISTRY FOR INTERNET NUMBERS
-+ ADDRESS: 3635 CONCORDE PARKWAY
-+ ADDRESS: SUITE 200
-+ CITY: CHANTILLY
-+ STATEPROV: VA
-+ POSTALCODE: 20151
-+ COUNTRY: US

data/log/system.log.20110820

@@ -1,20 +0,0 @@
-# Logfile created on 2011-06-19 21:07:36 -0400 by logger.rb/25413
-I, [2011-06-19T21:07:36.051026 #1168]  INFO -- : starting ... >>> 2011-06-19 21:07:36 -0400
-I, [2011-06-19T21:08:02.305872 #3676]  INFO -- : starting ... >>> 2011-06-19 21:08:02 -0400
-I, [2011-06-19T21:08:03.772274 #3676]  INFO -- : Record count is 4
-I, [2011-06-19T21:08:03.772274 #3676]  INFO -- : ending ... >>> 2011-06-19 21:08:03 -0400
-I, [2011-06-19T21:14:04.313707 #3816]  INFO -- : starting ... >>> 2011-06-19 21:14:04 -0400
-I, [2011-06-19T21:14:10.335317 #3816]  INFO -- : Record count is 4
-I, [2011-06-19T21:14:10.335317 #3816]  INFO -- : ending ... >>> 2011-06-19 21:14:10 -0400
-I, [2011-06-19T21:36:31.696191 #3208]  INFO -- : Starting IP address test file ... >>> 2011-06-19 21:36:31 -0400
-I, [2011-06-19T21:36:34.379396 #3208]  INFO -- : Record count is 4
-I, [2011-06-19T21:36:34.379396 #3208]  INFO -- : Ending ... >>> 2011-06-19 21:36:34 -0400
-I, [2011-06-19T22:43:07.127192 #2507]  INFO -- : Starting IP address test file ... >>> Sun Jun 19 22:43:07 -0400 2011
-I, [2011-06-19T22:43:16.796865 #2507]  INFO -- : Record count is 10
-I, [2011-06-19T22:43:16.797023 #2507]  INFO -- : Ending ... >>> Sun Jun 19 22:43:16 -0400 2011
-I, [2011-06-19T22:55:59.322407 #2549]  INFO -- : Starting IP address test file ... >>> Sun Jun 19 22:55:59 -0400 2011
-I, [2011-06-19T22:56:07.553043 #2549]  INFO -- : Record count is 10
-I, [2011-06-19T22:56:07.553205 #2549]  INFO -- : Ending ... >>> Sun Jun 19 22:56:07 -0400 2011
-I, [2011-06-19T22:57:05.825931 #2555]  INFO -- : Starting IP address test file ... >>> Sun Jun 19 22:57:05 -0400 2011
-I, [2011-06-19T22:57:10.268695 #2555]  INFO -- : Record count is 4
-I, [2011-06-19T22:57:10.268870 #2555]  INFO -- : Ending ... >>> Sun Jun 19 22:57:10 -0400 2011

data/log/system.log.20110903

@@ -1,7 +0,0 @@
-# Logfile created on Tue Aug 23 21:55:21 -0400 2011 by logger.rb/22285
-I, [2011-08-23T21:55:21.285906 #2337]  INFO -- : Starting IP address test file ... >>> Tue Aug 23 21:55:21 -0400 2011
-I, [2011-08-23T21:55:22.845128 #2337]  INFO -- : Record count is 5
-I, [2011-08-23T21:55:22.845300 #2337]  INFO -- : Ending ... >>> Tue Aug 23 21:55:22 -0400 2011
-I, [2011-08-23T22:00:28.868131 #2361]  INFO -- : Starting IP address test file ... >>> Tue Aug 23 22:00:28 -0400 2011
-I, [2011-08-23T22:00:30.466053 #2361]  INFO -- : Record count is 5
-I, [2011-08-23T22:00:30.466226 #2361]  INFO -- : Ending ... >>> Tue Aug 23 22:00:30 -0400 2011