Refactors and addressed some issues.

domainatrix.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{domainatrix}
-  s.version = "0.0.9"
+  s.version = "0.0.12"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Paul Dix"]
@@ -19,7 +19,6 @@ Gem::Specification.new do |s|
     "spec/domainatrix_spec.rb",
     "spec/domainatrix/domain_parser_spec.rb",
     "spec/domainatrix/url_spec.rb"]
-  s.has_rdoc = true
   s.homepage = %q{http://github.com/pauldix/domainatrix}
   s.require_paths = ["lib"]
   s.rubygems_version = %q{1.3.5}
@@ -35,4 +34,4 @@ Gem::Specification.new do |s|
     end
   else
   end
-end
+end

lib/domainatrix.rb

@@ -11,4 +11,8 @@ module Domainatrix
   def self.parse(url)
     Url.new(DOMAIN_PARSER.parse(url))
   end
-end
+
+  def self.recognize_tld(tld)
+    DOMAIN_PARSER.add_domain(tld)
+  end
+end

lib/domainatrix/domain_parser.rb

@@ -1,7 +1,7 @@
 module Domainatrix
   class DomainParser
     include Addressable
-    
+
     attr_reader :public_suffixes
 
     def initialize(file_name)
@@ -16,62 +16,71 @@ def read_dat_file(file_name)
       else
         dat_file = File.open(file_name)
       end
-      
+
       dat_file.each_line do |line|
         line = line.strip
-        unless (line =~ /\/\//) || line.empty?
-          parts = line.split(".").reverse
+        add_domain(line) unless line.match(/\/\//)
+      end
+    end
 
-          sub_hash = @public_suffixes
-          parts.each do |part|
-            sub_hash = (sub_hash[part] ||= {})
-          end
-        end
+    def add_domain(name)
+      parts = name.split(".").reverse
+
+      sub_hash = @public_suffixes
+      parts.each do |part|
+        sub_hash = (sub_hash[part] ||= {})
       end
     end
 
     def parse(url)
-      uri = URI.parse(url)
-      if uri.query
-        path = "#{uri.path}?#{uri.query}"
-      else
-        path = uri.path
+      url = "http://" + url unless url.match(/\/\//)
+      begin
+        uri = URI.parse(url)
+      rescue Addressable::URI::InvalidURIError
+        return {}
       end
+      path = uri.path
+      query = uri.query
+      path += "?#{query}" if query
+
       parse_domains_from_host(uri.host).merge({
         :scheme => uri.scheme,
-        :host   => uri.host,
         :path   => path,
         :url    => url
       })
     end
 
     def parse_domains_from_host(host)
+      unless host
+        return {:public_suffix => "", :domain => "", :subdomain => "", :host => ""}
+      end
+      public_suffix = parse_public_suffix_from_host(host)
+      parts = host.gsub(public_suffix, '').split(".")
+      domain = parts.pop
+      subdomains = parts.join(".")
+      {:public_suffix => public_suffix, :domain => domain, :subdomain => subdomains, :host => host}
+    end
+
+    def parse_public_suffix_from_host(host)
       parts = host.split(".").reverse
-      public_suffix = []
-      domain = ""
-      subdomains = []
       sub_hash = @public_suffixes
-      parts.each_index do |i|
-        part = parts[i]
+      public_suffix = []
 
+      parts.each_with_index do |part, index|
         sub_parts = sub_hash[part]
         sub_hash = sub_parts
+        return "" unless sub_parts
+        public_suffix << part
+        next_part = parts[index + 1]
         if sub_parts.has_key? "*"
-          public_suffix << part
-          public_suffix << parts[i+1]
-          domain = parts[i+2]
-          subdomains = parts.slice(i+3, parts.size)
+          public_suffix << next_part
           break
-        elsif sub_parts.empty? || !sub_parts.has_key?(parts[i+1])
-          public_suffix << part
-          domain = parts[i+1]
-          subdomains = parts.slice(i+2, parts.size)
+        elsif sub_parts.empty? || !sub_parts.has_key?(next_part)
           break
-        else
-          public_suffix << part
         end
       end
-      {:public_suffix => public_suffix.reverse.join("."), :domain => domain, :subdomain => subdomains.reverse.join(".")}
+
+      public_suffix.reverse.join(".")
     end
   end
-end
+end

lib/domainatrix/url.rb

@@ -13,6 +13,7 @@ def initialize(attrs = {})
     end
 
     def canonical(options = {})
+      return nil unless valid?
       public_suffix_parts = @public_suffix.split(".")
       url = "#{public_suffix_parts.reverse.join(".")}.#{@domain}"
       if @subdomain && [email protected]?
@@ -29,5 +30,8 @@ def domain_with_public_suffix
     end
     alias domain_with_tld domain_with_public_suffix
 
+    def valid?
+      @public_suffix and !@public_suffix.empty?
+    end
   end
 end