whois-parser 0.0.1 → 1.0.0.pre.beta1

data/lib/whois/scanners/scannable.rb

@@ -0,0 +1,65 @@
+require 'strscan'
+
+module Whois
+  module Scanners
+
+    # The Scannable module tries to emulate a super-simple Abstract Syntax Tree structure
+    # including method for accessing ast nodes.
+    #
+    # == Usage
+    #
+    # Include the Scannable module and set the `self.scanner` value.
+    #
+    #   class ParserFoo
+    #     include Scannable
+    #
+    #     self.scanner = ScannerFoo
+    #   end
+    #
+    # Now you can access the AST using the <tt>node</tt> method.
+    #
+    #   node "created_on"
+    #   # => "2009-12-12"
+    #
+    #   node? "created_on"
+    #   # => true
+    #
+    #   node? "created_at"
+    #   # => false
+    #
+    module Scannable
+
+      def self.included(base)
+        base.class_attribute :scanner
+      end
+
+      def node(key)
+        if block_given?
+          value = ast[key]
+          value = yield(value) unless value.nil?
+          value
+        else
+          ast[key]
+        end
+      end
+
+      def node?(key)
+        !ast[key].nil?
+      end
+
+      def parse
+        scanner  = self.scanner.is_a?(Array) ? self.scanner.first : self.scanner
+        settings = self.scanner.is_a?(Array) ? self.scanner.last  : {}
+        scanner.new(settings).parse(content_for_scanner)
+      end
+
+      private
+
+      def ast
+        @ast ||= parse
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/verisign.rb

@@ -0,0 +1,100 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class Verisign < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_response_unavailable,
+          :scan_available,
+          :scan_disclaimer,
+          :scan_notice,
+          :scan_keyvalue_indented,
+          :skip_ianaservice,
+          :skip_lastupdate,
+          :skip_fuffa,
+      ]
+
+
+      tokenizer :scan_response_unavailable do
+        # Check if the string starts with /*
+        # If it does, match until the end of all /* lines
+        # or the end of the file and check for the content.
+        #
+        # Flag the block as visited to force the scanner to ignore this tokenizer
+        # if already used and the content didn't match the unavailable message.
+        if @input.match?(/^\*\n/) && !visited?
+          p = /^[^\*]|\z/
+          y = @input.check_until(p) =~ /^\* Sorry, the Whois database is currently down/
+
+          if y
+            @input.skip_until(p)
+            @ast["response:unavailable"] = true
+          else
+            visited!
+          end
+        end
+      end
+
+      tokenizer :scan_available do
+        if @input.scan(/No match for "(.+?)"\.\n/)
+          @ast["Domain Name"] = @input[1].strip
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^TERMS OF USE:/)
+          @ast["Disclaimer"] = _scan_lines_to_array(/(.+)\n/).join(" ")
+        end
+      end
+
+      tokenizer :scan_notice do
+        if @input.match?(/^NOTICE:/)
+          @ast["Notice"] = _scan_lines_to_array(/(.+)\n/).join(" ")
+        end
+      end
+
+      tokenizer :scan_keyvalue_indented do
+        if @input.scan(/\s+(.+?):(.*?)\n/)
+          key, value = @input[1].strip, @input[2].strip
+          if @ast[key].nil?
+            @ast[key] = value
+          else
+            @ast[key] = [@ast[key]] unless @ast[key].is_a?(Array)
+            @ast[key] << value
+          end
+        end
+      end
+
+      tokenizer :skip_lastupdate do
+        @input.skip(/>>>(.+?)<<<\n/)
+      end
+
+      tokenizer :skip_fuffa do
+        @input.scan(/^\S(.+)(?:\n|\z)/)
+      end
+
+      tokenizer :skip_ianaservice do
+        if @input.match?(/IANA Whois Service/)
+          @ast["IANA"] = true
+          @input.terminate
+        end
+      end
+
+      private
+
+      def visited?
+        !!@tmp["visited:#{@input.pos}"]
+      end
+
+      def visited!
+        @tmp["visited:#{@input.pos}"] = true
+        nil
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.ati.tn.rb

@@ -0,0 +1,32 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.ati.tn record.
+    class WhoisAtiTn < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_disclaimer,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^Domain (.+) not found/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^All rights reserved/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/(.+)\n/).join("\n")
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.audns.net.au.rb

@@ -0,0 +1,25 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.audns.net.au record.
+    class WhoisAudnsNetAu < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^No Data Found\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.cctld.by.rb

@@ -0,0 +1,35 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.cctld.by server.
+    #
+    # @author Aliaksei Kliuchnikau <aliaksei.kliuchnikau@gmail.com>
+    class WhoisCctldBy < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :skip_dash_line,
+          :scan_available,
+          :scan_keyvalue,
+          :skip_provider_signature,
+      ]
+
+      tokenizer :scan_available do
+        if @input.skip(/^Object does not exist/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :skip_dash_line do
+        @input.skip(/^-+\n/)
+      end
+
+      tokenizer :skip_provider_signature do
+        @input.scan(/^(.+)\n/)
+      end
+
+    end
+  end
+end

data/lib/whois/scanners/whois.centralnic.com.rb

@@ -0,0 +1,32 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.centralnic.com record.
+    class WhoisCentralnicCom < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_disclaimer,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^DOMAIN NOT FOUND\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^\S([^:]+)\n/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/(.+)\n/).join(" ")
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.cira.ca.rb

@@ -0,0 +1,72 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.cira.ca record.
+    class WhoisCiraCa < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_disclaimer,
+          :skip_comment,
+          :scan_header,
+          :scan_keyvalue,
+          :scan_nameserver,
+      ]
+
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^% Use of CIRA/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/^%(.*)\n/).join("\n")
+        end
+      end
+
+      tokenizer :scan_header do
+        if @input.scan(/^(.+?):\n/)
+          @tmp["group"] = @input[1]
+        end
+      end
+
+      tokenizer :scan_keyvalue do
+        if @input.scan(/^(.+?):(.*?)\n/)
+          start = @input[1]
+          key, value = start.strip, @input[2].strip
+
+          # This is a nested key
+          target = if start.index(" ") == 0
+            error!("Expected group.") if @tmp["group"].nil?
+            @ast[@tmp["group"]] ||= {}
+            @ast[@tmp["group"]]
+          else
+            @tmp.delete("group")
+            @ast
+          end
+
+          more  = _scan_lines_to_array(/^\s{#{start.size}}(.+)\n/)
+          value = more.unshift(value).join("\n") unless more.empty?
+
+          if target[key].nil?
+            target[key] = value
+          else
+            target[key] = Array.wrap(target[key])
+            target[key] << value
+          end
+        end
+      end
+
+      tokenizer :scan_nameserver do
+        if @input.scan(/^\s+(.+?)\n/) && @tmp["group"] == "Name servers"
+          @ast["field:nameservers"] ||= []
+          @ast["field:nameservers"] <<  @input[1].strip
+        end
+      end
+
+      tokenizer :skip_comment do
+        @input.skip(/^%.*\n/)
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.cnnic.cn.rb

@@ -0,0 +1,42 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.cnnic.cn record.
+    class WhoisCnnicCn < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_reserved,
+          :scan_reserved_list,
+          :scan_available,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.match?(/^no matching record/)
+          @ast["status:available"] = true
+          @input.scan_until(/\n/)
+        end
+      end
+
+      tokenizer :scan_reserved do
+        if @input.match?(/^The domain you requested is prohibited/)
+          @ast["status:reserved"] = true
+          @input.scan_until(/\n/)
+        end
+      end
+
+      tokenizer :scan_reserved_list do
+        if @input.scan(/^Sorry, The domain you requested is in the reserved list/)
+          @ast["status:reserved"] = true
+          @input.scan_until(/\n/)
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.denic.de.rb

@@ -0,0 +1,96 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class WhoisDenicDe < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_response_throttled,
+          :scan_response_error,
+          :scan_disclaimer,
+          :scan_pair,
+          :scan_contact,
+          :skip_db_time,
+      ]
+
+
+      tokenizer :scan_response_throttled do
+        if @input.match?(/^% Error: 55000000002/)
+          @ast["response:throttled"] = true
+          @input.skip(/^.+\n/)
+        end
+      end
+
+      tokenizer :scan_response_error do
+        if @input.match?(/^% Error: 55000000010/)
+          @ast["response:error"] = true
+          @input.skip(/^.+\n/)
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/% Copyright \(c\) *\d{4} by DENIC\n/)
+          @input.scan_until(/% Terms and Conditions of Use\n/)
+          lines = []
+          while @input.match?(/%/) && @input.scan(/%(.*)\n/)
+            lines << @input[1].strip unless @input[1].strip == ""
+          end
+          @ast["Disclaimer"] = lines.join(" ")
+        end
+      end
+
+      tokenizer :scan_pair do
+        parse_pair(@ast)
+      end
+
+      tokenizer :scan_contact do
+        if @input.scan(/\[(.*)\]\n/)
+          contact_name = @input[1]
+          contact = {}
+          while parse_pair(contact)
+          end
+          @ast[contact_name] = {
+            "id" => nil,
+            "name" => contact['Name'],
+            "organization" => contact['Organisation'],
+            "address" => contact['Address'],
+            "city" => contact['City'],
+            "zip" => contact['PostalCode'],
+            "state" => nil,
+            "country" => contact['Country'],
+            "country_code" => contact['CountryCode'],
+            "phone" => contact['Phone'],
+            "fax" => contact['Fax'],
+            "email" => contact['Email'],
+            "created_on" => nil,
+            "updated_on" => contact['Changed']
+          }
+        end
+      end
+
+      tokenizer :skip_db_time do
+        @input.skip(/^% DB time is (.+)\n/)
+      end
+
+
+      private
+
+      def parse_pair(store)
+        if @input.scan(/([^  \[]*):(.*)\n/)
+          key, value = @input[1].strip, @input[2].strip
+          if store[key].nil?
+            store[key] = value
+          else
+            store[key].is_a?(Array) || store[key] = [store[key]]
+            store[key] << value
+          end
+          store
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.dns.hr.rb

@@ -0,0 +1,25 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.dns.hr record.
+    class WhoisDnsHr < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^%ERROR: no entries found\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.domainregistry.ie.rb

@@ -0,0 +1,46 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.domainregistry.ie server.
+    class WhoisDomainregistryIe < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_disclaimer,
+          :scan_contact,
+          :scan_keyvalue,
+          :scan_available,
+          :skip_application_pending,
+      ]
+
+      tokenizer :scan_available do
+        if @input.skip(/^% Not Registered - .+\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^% Rights restricted by copyright/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/^%(.+)\n/).join("\n")
+        end
+      end
+
+      tokenizer :scan_contact do
+        if @input.match?(/^person:/)
+          lines = _scan_lines_to_hash(/(.+?):(.*?)\n/)
+          @ast["field:#{lines['nic-hdl']}"] = lines
+        end
+      end
+
+      tokenizer :skip_application_pending do
+        if @input.match?(/^% Application Pending/)
+          _scan_lines_to_array(/^%(.+)\n/)
+          @ast["status:pending"] = true
+        end
+      end
+
+    end
+  end
+end

data/lib/whois/scanners/whois.fi.rb

@@ -0,0 +1,39 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.fi record.
+    class WhoisFi < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_disclaimer,
+          :scan_keyvalue,
+          :scan_reserved
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^Domain not found/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_reserved do
+        if @input.skip(/^Domain not available/)
+          @ast["status:reserved"] = true
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^More information/)
+          @ast["field:disclaimer"] = @input.scan_until(/(.*)\n\n/).strip
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.nc.rb

@@ -0,0 +1,32 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the whois.nc record.
+    class WhoisNc < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :skip_more,
+          :scan_available,
+          :scan_keyvalue,
+      ]
+
+
+      MORES = ['Whois \.NC', 'more details on']
+
+      tokenizer :scan_available do
+        if @input.skip(/^No entries found .+\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :skip_more do
+        MORES.any? { |more| @input.skip(/^#{more}.*\n/) }
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/whois.nic.cz.rb

@@ -0,0 +1,22 @@
+require_relative 'base_whoisd'
+
+module Whois
+  module Scanners
+
+    class WhoisNicCz < BaseWhoisd
+
+      self.tokenizers += [
+          :scan_response_throttled,
+      ]
+
+      tokenizer :scan_response_throttled do
+        if @input.match?(/Your connection limit exceeded\. Please slow down and try again later/)
+          @ast["response:throttled"] = true
+          @input.skip(/^.+\n/)
+        end
+      end
+
+    end
+
+  end
+end