whois-parser-next 0.1.0

data/lib/whois/scanners/base_afilias.rb

@@ -0,0 +1,51 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the Afilias-based records.
+    class BaseAfilias < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_reserved,
+          :scan_throttled,
+          :scan_disclaimer,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.scan(/^(?:NOT FOUND|Domain not found\.)\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_reserved do
+        if settings[:pattern_reserved] && @input.scan(settings[:pattern_reserved])
+          @ast["status:reserved"] = true
+        end
+      end
+
+      # .ORG
+      tokenizer :scan_throttled do
+        if @input.match?(/^WHOIS LIMIT EXCEEDED/)
+          @ast["response:throttled"] = true
+          @input.skip(/^.+\n/)
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if settings[:pattern_disclaimer]
+          if @input.match?(settings[:pattern_disclaimer])
+            @ast["field:disclaimer"] = _scan_lines_to_array(/^(.+)\n/).join(" ")
+          end
+        elsif @input.pos == 0 && @input.match?(/^(.+\n){3,}\n/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/^(.+)\n/).join(" ")
+        end
+      end
+    end
+
+  end
+end

data/lib/whois/scanners/base_cocca2.rb

@@ -0,0 +1,58 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class BaseCocca2 < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_disclaimer,
+          :skip_lastupdate,
+          :skip_token_additionalsection,
+          :skip_comment,
+          :skip_no_object,
+          :skip_footer,
+          :scan_keyvalue,
+      ]
+
+
+      DISCLAIMER_MATCHES = [
+        "TERMS OF USE:", # global
+        "Terminos de Uso:", # whois.nic.hn
+        "The data in the WHOIS database of Meridian", # whois.meridiantld.net
+        "This information is provided", # whois.gg
+      ]
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^#{DISCLAIMER_MATCHES.join('|')}/)
+          @ast["field:disclaimer"] = @input.scan_until(/>>>/) ||
+                                     # special handler for whois.nic.cx exception
+                                     @input.scan_until(/\Z/)
+        end
+      end
+
+      tokenizer :skip_lastupdate do
+        @input.skip(/>>>(.+?)<<<\n/)
+      end
+
+      tokenizer :skip_token_additionalsection do
+        @input.skip(/Additional Section\n/)
+      end
+
+      tokenizer :skip_comment do
+        @input.skip(/^%.*\n/)
+      end
+
+      tokenizer :skip_no_object do
+        @input.skip(/^The queried object does not exist: No Object Found\n/i)
+      end
+
+      tokenizer :skip_footer do
+        @input.skip(/^For more information on domain status codes,.*\n/i)
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_icann_compliant.rb

@@ -0,0 +1,45 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the Icann Compliant-based records.
+    class BaseIcannCompliant < Base
+
+      self.tokenizers += [
+          :skip_head,
+          :scan_available,
+          :scan_throttled,
+          :skip_empty_line,
+          :skip_blank_line,
+          :scan_keyvalue,
+          :skip_end,
+      ]
+
+      tokenizer :scan_available do
+        if settings[:pattern_available] && @input.skip_until(settings[:pattern_available])
+          @ast['status:available'] = true
+        end
+      end
+
+      tokenizer :scan_throttled do
+        if settings[:pattern_throttled] && @input.skip_until(settings[:pattern_throttled])
+          @ast['response:throttled'] = true
+        end
+      end
+
+      tokenizer :skip_head do
+        if @input.skip_until(/Domain Name:/)
+          @input.scan(/\s?(.+)\n/)
+          @ast["Domain Name"] = @input[1].strip
+        end
+      end
+
+      tokenizer :skip_end do
+        @input.terminate
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_iisse.rb

@@ -0,0 +1,36 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for the IIS.se company.
+    class BaseIisse < 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?(/# Copyright/)
+          lines = []
+          while @input.scan(/#(.*)\n\n?/) # rubocop:disable Style/WhileUntilModifier
+            lines << @input[1].strip if @input[1].strip != ""
+          end
+          @ast["field:disclaimer"] = lines.join(" ")
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_shared1.rb

@@ -0,0 +1,31 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class BaseShared1 < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_available,
+          :scan_reserved,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^No Data Found\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_reserved do
+        if settings[:pattern_reserved] && @input.scan(settings[:pattern_reserved])
+          @ast["status:reserved"] = true
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_shared2.rb

@@ -0,0 +1,35 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class BaseShared2 < Base
+
+      self.tokenizers += [
+          :skip_blank_line,
+          :scan_available,
+          :scan_keyvalue,
+          :skip_lastupdate,
+          :skip_fuffa,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.scan(/^Not found: (.+)\n/)
+          @ast["Domain Name"] = @input[1]
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :skip_lastupdate do
+        @input.skip(/>>>(.+?)<<<\n/)
+      end
+
+      tokenizer :skip_fuffa do
+        @input.skip(/^\S(.+)\n/)
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_shared3.rb

@@ -0,0 +1,37 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class BaseShared3 < Base
+
+      self.tokenizers += [
+          :scan_disclaimer,
+          :skip_empty_line,
+          :skip_comment,
+          :scan_available,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :skip_comment do
+        @input.skip(/^;.*\n/)
+      end
+
+      tokenizer :scan_available do
+        if @input.skip(/^not found.+\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :scan_disclaimer do
+        if @input.pos == 0 && @input.match?(/^;.*/)
+          @ast["field:disclaimer"] = _scan_lines_to_array(/^;(.*)\n/).join(" ")
+          @ast["field:disclaimer"].gsub!("  ", "\n")
+        end
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/base_whoisd.rb

@@ -0,0 +1,44 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    # Scanner for Whoisd-based records.
+    class BaseWhoisd < Base
+
+      self.tokenizers += [
+          :scan_available,
+          :skip_comment,
+          :scan_section,
+          :scan_keyvalue,
+          :skip_empty_line,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^%ERROR:101: no entries found/)
+          @ast['status:available'] = true
+        end
+      end
+
+      tokenizer :scan_section do
+        if @input.scan(/\n(contact|nsset|keyset):\s+(.+)\n/)
+          @tmp["_section"] = "node:#{@input[1]}/#{@input[2].strip}"
+          while scan_keyvalue
+          end
+          @tmp.delete("_section")
+        end
+      end
+
+      tokenizer :todo_content do
+        @input.scan(/(.*)\n/)
+      end
+
+      tokenizer :skip_comment do
+        @input.skip(/^%.*\n/)
+      end
+
+    end
+
+  end
+end

data/lib/whois/scanners/iana.rb

@@ -0,0 +1,89 @@
+require_relative 'base'
+
+module Whois
+  module Scanners
+
+    class Iana < Base
+
+      self.tokenizers += [
+          :skip_empty_line,
+          :scan_disclaimer,
+          :scan_section,
+      ]
+
+
+      tokenizer :scan_disclaimer do
+        if @input.match?(/^%(.*?)\n/)
+          @ast["Disclaimer"] = _scan_lines_to_array(/%(.*?)\n/).select { |line| line =~ /\w+/ }.join(" ")
+        end
+      end
+
+      tokenizer :scan_section do
+        if @input.scan(/^(.+):(.+)\n/)
+
+          # Adapt the section's name depending on the first line
+          section = case @input[1].strip
+                    when 'contact'
+                      @input[2].strip # use the contact type name as identifier
+                    when 'created', 'changed'
+                      'dates'
+                    when 'nserver'
+                      'nameservers'
+                    else
+                      @input[1].strip # Default name is the first label found
+                    end
+
+          content = parse_section_pairs
+          @input.match?(/\n+/) || error("Unexpected end of section")
+          @ast[section] = content
+        end
+      end
+
+
+      private
+
+      def parse_section_pairs
+        # Sets by default the firsts values found in the section parsing bellow
+        section_name, section_value = @input[1].strip, @input[2].strip
+        # contents = {section_name =>  section_value}
+
+        contents = {}
+
+        while content = parse_section_pair
+          contents.merge!(content)
+        end
+
+        if contents.key? section_name
+          contents[section_name].insert(0, "#{section_value}\n")
+        else
+          contents[section_name] = section_value
+        end
+
+        if contents.empty?
+          false
+        else
+          contents
+        end
+      end
+
+      def parse_section_pair
+        if @input.scan(/^(.+):\s*(.+)\n/)
+          key     =  @input[1].strip
+          values  = [@input[2].strip]
+
+          while value = parse_section_pair_newlinevalue(key)
+            values << value
+          end
+          { key => values.join("\n") }
+        end
+      end
+
+      def parse_section_pair_newlinevalue(key)
+        if @input.scan(/^#{key}:\s*(.+)\n/)
+          @input[1].strip
+        end
+      end
+
+    end
+  end
+end

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 +1,32 @@
+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,
+          :skip_lastupdate,
+          :scan_keyvalue,
+      ]
+
+
+      tokenizer :scan_available do
+        if @input.skip(/^(No Data Found|Domain not found\.)\n/)
+          @ast["status:available"] = true
+        end
+      end
+
+      tokenizer :skip_lastupdate do
+        if @input.skip(/^>>> Last update of WHOIS database:.*\n/)
+          @input.terminate
+        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/i)
+          @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