whois-parser-next 0.3.2 → 0.4.0

data/lib/whois/parsers/whois.nic.mn.rb

@@ -0,0 +1,51 @@
+#--
+# Ruby Whois
+#
+# An intelligent pure Ruby WHOIS client and parser.
+#++
+
+require_relative 'base_icann_compliant'
+
+module Whois
+  class Parsers
+    # Parser for the IANA-listed .mn WHOIS host, whois.nic.mn.
+    class WhoisNicMn < BaseIcannCompliant
+      self.scanner = Scanners::BaseIcannCompliant, {
+        pattern_available: /^[ \t]*Domain not found\.[ \t\r]*$/i,
+      }
+
+      property_supported :status do
+        classify_status
+      end
+
+      property_supported :available? do
+        classify_status == :available
+      end
+
+      property_supported :registered? do
+        classify_status == :registered
+      end
+
+      private
+
+      def classify_status
+        cached_properties_fetch(:classified_status) do
+          missing = content_for_scanner.match?(/\A[ \t]*Domain not found\.[ \t\r]*(?:\n|\z)/i)
+          denial = content_for_scanner.match?(/^[ \t]*(?:[%#][ \t]*)?(?:access denied|permission denied|request denied|forbidden)\b/i)
+          domains = content_for_scanner.scan(/^[ \t]*Domain Name:[ \t]*((?:[a-z0-9-]+\.)+mn)[ \t\r]*$/i).flatten.uniq
+          has_registrar = content_for_scanner.match?(/^[ \t]*Registrar:[ \t]*\S/i)
+          has_record_detail = content_for_scanner.match?(/^[ \t]*(?:Creation Date|Updated Date|Registry Expiry Date|Name Server):[ \t]*\S/i)
+          has_record = domains.one? && has_registrar && has_record_detail
+
+          if missing && domains.empty? && !denial && !response_throttled?
+            :available
+          elsif has_record && !missing && !denial && !response_throttled?
+            :registered
+          else
+            :unknown
+          end
+        end
+      end
+    end
+  end
+end

data/lib/whois/parsers/whois.nic.mx.rb

@@ -8,6 +8,7 @@
 
 
 require_relative 'base'
+require_relative 'registry_response_safety'
 
 
 module Whois
@@ -25,21 +26,22 @@ module Whois
     # and examples.
     #
     class WhoisNicMx < Base
+      include RegistryResponseSafety
+
+      DOMAIN_PATTERN = /\A(?:[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?\.)+mx\z/i
+      NO_RECORD_MARKER = %r{^[ \t]*No_Se_Encontro_El_Objeto/Object_Not_Found[ \t]*$}i
+      RECORD_EVIDENCE = /^[ \t]*(?:Created On|Expiration Date|Registrar|Name Servers|DNS):/i
 
       property_supported :status do
-        if available?
-          :available
-        else
-          :registered
-        end
+        classify_status
       end
 
       property_supported :available? do
-        !!(content_for_scanner =~ /Object_Not_Found/)
+        classify_status == :available
       end
 
       property_supported :registered? do
-        !available?
+        classify_status == :registered
       end
 
 
@@ -73,6 +75,44 @@ module Whois
         end
       end
 
+      def response_unavailable?
+        super || content_for_scanner.match?(
+          /^\s*(?:access denied|permission denied|request denied|forbidden)\b/i
+        )
+      end
+
+      private
+
+      def classify_status
+        cached_properties_fetch(:classified_status) do
+          response = content_for_scanner
+          absent = response.lines.one? { |line| line.match?(NO_RECORD_MARKER) }
+          domain_rows = response.scan(/^[ \t]*Domain Name:[ \t]*(.*?)[ \t\r]*$/i).flatten
+          domains = domain_rows.grep(DOMAIN_PATTERN)
+          has_created = response.match?(/^[ \t]*Created On:[ \t]*\S/i)
+          has_expires = response.match?(/^[ \t]*Expiration Date:[ \t]*\S/i)
+          has_registrar = response.match?(/^[ \t]*Registrar:[ \t]*\S/i)
+          registered = [
+            domain_rows.one?,
+            domains.one?,
+            has_created,
+            has_expires,
+            has_registrar,
+            !response_unavailable?,
+            !response_throttled?,
+          ].all?
+          has_record_evidence = response.match?(RECORD_EVIDENCE)
+
+          if absent && domain_rows.empty? && !has_record_evidence && !response_unavailable? && !response_throttled?
+            :available
+          elsif registered && !absent
+            :registered
+          else
+            :unknown
+          end
+        end
+      end
+
     end
 
   end

data/lib/whois/parsers/whois.nic.net.sa.rb

@@ -65,6 +65,12 @@ module Whois
       private
 
       def classify_status
+        cached_properties_fetch(:classified_status) do
+          classify_status_from_content
+        end
+      end
+
+      def classify_status_from_content
         registered = registered_domain_names
         absent = absence_domain_names
 

data/lib/whois/parsers/whois.nic.news.rb

@@ -2,6 +2,8 @@ require_relative 'base_unsupported_registry'
 
 module Whois
   class Parsers
+    # IANA currently lists RDAP only for .news. Keep this adapter only so a
+    # legacy upstream WHOIS route fails explicitly instead of claiming status.
     class WhoisNicNews < BaseUnsupportedRegistry
     end
   end

data/lib/whois/parsers/whois.nic.one.rb

@@ -8,7 +8,9 @@ module Whois
     # footer. Keep that exact evidence separate from an unparseable response.
     class WhoisNicOne < BaseIcannCompliant
       property_supported :status do
-        if no_data_found?
+        if contradictory_registration_and_absence?
+          :unknown
+        elsif no_data_found?
           :available
         elsif registered_response?
           :registered
@@ -31,6 +33,10 @@ module Whois
         content_for_scanner.lines.first.to_s.strip.casecmp?("No Data Found")
       end
 
+      def contradictory_registration_and_absence?
+        no_data_found? && registered_response?
+      end
+
       def registered_response?
         content_for_scanner.match?(/^Domain Name:\s+\S+/) &&
           content_for_scanner.match?(/^Domain Status:\s+\S+/)

data/lib/whois/parsers/whois.nic.online.rb

@@ -12,8 +12,25 @@ module Whois
 
       self.scanner = Scanners::BaseIcannCompliant
 
+      property_supported :status do
+        if contradictory_registration_and_absence?
+          :unknown
+        elsif available?
+          :available
+        elsif node('Domain Name').to_s.strip.empty? ||
+              [node('Registry Domain ID'), node('Creation Date'), node('Registrar')].all? { |value| value.to_s.strip.empty? }
+          :unknown
+        else
+          :registered
+        end
+      end
+
       property_supported :available? do
-        !!(content_for_scanner =~ /^>>> Domain \S+ is available for registration\s*$/i)
+        !contradictory_registration_and_absence? && !!(content_for_scanner =~ /^>>> Domain \S+ is available for registration\s*$/i)
+      end
+
+      property_supported :registered? do
+        status == :registered
       end
 
       # Radix can answer with a limit or client-denial notice that contains no
@@ -33,6 +50,13 @@ module Whois
         node("Registry Expiry Date") { |value| parse_time(value) }
       end
 
+      private
+
+      def contradictory_registration_and_absence?
+        !node('Domain Name').to_s.strip.empty? &&
+          content_for_scanner.match?(/^>>> Domain \S+ is available for registration\s*$/i)
+      end
+
     end
 
   end

data/lib/whois/parsers/whois.nic.open.rb

@@ -0,0 +1,95 @@
+require_relative 'base_icann_compliant'
+
+module Whois
+  class Parsers
+    # Conservative parser for American Express's restricted .OPEN registry.
+    # The registry's exact "No Data Found" response is not public availability:
+    # the published policy restricts registrations to the operator, qualified
+    # affiliates, and trademark licensees.
+    class WhoisNicOpen < BaseIcannCompliant
+      DOMAIN_PATTERN = /\A(?:[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?\.)+open\z/i
+      DOMAIN_ID_PATTERN = /\AD\d+-OPEN\z/i
+      NO_DATA_MARKER = /^[ \t]*No Data Found[ \t]*$/i
+      RECORD_FIELD = /^[ \t]*(?:Domain Name|Registry Domain ID|Updated Date|Creation Date|Registry Expiry Date|Registrar|Registrar IANA ID|Domain Status|Name Server|DNSSEC):[ \t]*\S/i
+      REGISTRAR_NAME = /\AAmerican Express Travel Related Services, Inc\.\z/i
+      NAMESERVER_PATTERN = /\A(?:[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?\.?\z/i
+
+      property_supported :domain do
+        values = field_values('Domain Name')
+        values.one? && values.first.match?(DOMAIN_PATTERN) ? values.first.downcase : nil
+      end
+
+      property_supported :domain_id do
+        values = field_values('Registry Domain ID')
+        values.one? && values.first.match?(DOMAIN_ID_PATTERN) ? values.first : nil
+      end
+
+      property_supported :status do
+        classify_status
+      end
+
+      property_supported :available? do
+        classify_status == :available
+      end
+
+      property_supported :registered? do
+        classify_status == :registered
+      end
+
+      property_supported :expires_on do
+        values = field_values('Registry Expiry Date')
+        parse_time(values.first) if values.one?
+      end
+
+      def response_unavailable?
+        body = content_for_scanner
+        marker_count = body.lines.count { |line| line.match?(NO_DATA_MARKER) }
+
+        super || (marker_count.positive? && (marker_count != 1 || body.match?(RECORD_FIELD)))
+      end
+
+      private
+
+      def classify_status
+        cached_properties_fetch(:classified_status) do
+          complete_registered_record? ? :registered : :unknown
+        end
+      end
+
+      def complete_registered_record?
+        domains = field_values('Domain Name')
+        ids = field_values('Registry Domain ID')
+        registrars = field_values('Registrar')
+        registrar_ids = field_values('Registrar IANA ID')
+
+        domains.one? && domains.first.match?(DOMAIN_PATTERN) &&
+          ids.one? && ids.first.match?(DOMAIN_ID_PATTERN) &&
+          registrars.one? && registrars.first.match?(REGISTRAR_NAME) &&
+          registrar_ids == ['9999'] &&
+          valid_date_field?('Creation Date') &&
+          valid_date_field?('Updated Date') &&
+          valid_date_field?('Registry Expiry Date') &&
+          valid_nameservers? && field_values('Domain Status').any?
+      end
+
+      def valid_date_field?(name)
+        values = field_values(name)
+        return false unless values.one?
+
+        Time.iso8601(values.first)
+        true
+      rescue ArgumentError
+        false
+      end
+
+      def valid_nameservers?
+        values = field_values('Name Server')
+        values.any? && values.all? { |value| value.match?(NAMESERVER_PATTERN) }
+      end
+
+      def field_values(name)
+        content_for_scanner.scan(/^[ \t]*#{Regexp.escape(name)}:[ \t]*(.*?)[ \t\r]*$/i).flatten
+      end
+    end
+  end
+end

data/lib/whois/parsers/whois.nic.quest.rb

@@ -3,6 +3,23 @@ require_relative 'whois.centralnic.com'
 module Whois
   class Parsers
     class WhoisNicQuest < WhoisCentralnicCom
+      property_supported :available? do
+        !contradictory_registration_and_absence? && !!node('status:available')
+      end
+
+      property_supported :registered? do
+        !contradictory_registration_and_absence? &&
+          !available? && !Array.wrap(node('Status') || node('Domain Status')).empty? &&
+          !node('Domain Name').to_s.strip.empty? &&
+          Array.wrap(node('Status') || node('Domain Status')).none? { |value| value.to_s.strip.casecmp('unknown').zero? }
+      end
+
+      private
+
+      def contradictory_registration_and_absence?
+        !node('Domain Name').to_s.strip.empty? &&
+          content_for_scanner.match?(/^(?:DOMAIN NOT FOUND|The queried object does not exist: DOMAIN NOT FOUND|>>> Domain \S+ is available for registration)\s*$/i)
+      end
     end
   end
 end

data/lib/whois/parsers/whois.nic.sbs.rb

@@ -3,6 +3,34 @@ module Whois
   class Parsers
     class WhoisNicSbs < BaseIdentityDigital
       self.scanner = Scanners::BaseIcannCompliant, { pattern_available: /^The queried object does not exist: DOMAIN NOT FOUND\s*$/i }
+
+      property_supported :status do
+        if contradictory_registration_and_absence?
+          :unknown
+        elsif available?
+          :available
+        elsif node('Domain Name').to_s.strip.empty? ||
+              [node('Registry Domain ID'), node('Creation Date'), node('Registrar')].all? { |value| value.to_s.strip.empty? }
+          :unknown
+        else
+          :registered
+        end
+      end
+
+      property_supported :available? do
+        !contradictory_registration_and_absence? && !!node('status:available')
+      end
+
+      property_supported :registered? do
+        status == :registered
+      end
+
+      private
+
+      def contradictory_registration_and_absence?
+        !node('Domain Name').to_s.strip.empty? &&
+          content_for_scanner.match?(/^The queried object does not exist: DOMAIN NOT FOUND\s*$/i)
+      end
     end
   end
 end

data/lib/whois/parsers/whois.nic.store.rb

@@ -10,6 +10,34 @@ module Whois
       self.scanner = Scanners::BaseIcannCompliant, {
         pattern_available: /^>>> Domain \S+ is available for registration\s*$/i,
       }
+
+      property_supported :status do
+        if contradictory_registration_and_absence?
+          :unknown
+        elsif available?
+          :available
+        elsif node('Domain Name').to_s.strip.empty? ||
+              [node('Registry Domain ID'), node('Creation Date'), node('Registrar')].all? { |value| value.to_s.strip.empty? }
+          :unknown
+        else
+          :registered
+        end
+      end
+
+      property_supported :available? do
+        !contradictory_registration_and_absence? && !!node('status:available')
+      end
+
+      property_supported :registered? do
+        status == :registered
+      end
+
+      private
+
+      def contradictory_registration_and_absence?
+        !node('Domain Name').to_s.strip.empty? &&
+          content_for_scanner.match?(/^>>> Domain \S+ is available for registration\s*$/i)
+      end
     end
   end
 end

data/lib/whois/parsers/whois.nic.top.rb

@@ -5,10 +5,32 @@ module Whois
     # The .top registry uses the shared ICANN record shape and returns the full
     # queried .top name in its exact no-record response.
     class WhoisNicTop < BaseTop1000Icann
+      RECORD_EVIDENCE = /^(?:Domain Name|Registry Domain ID|Creation Date|Registrar|Domain Status):[ \t]*\S/i
+      NO_RECORD_MARKER = /
+        ^The\ queried\ object\ does\ not\ exist:\s+
+        (?:DOMAIN\ NOT\ FOUND|no\ matching\ objects\ found|
+          (?:[A-Za-z0-9](?:[A-Za-z0-9-]*[A-Za-z0-9])?\.)*
+          [A-Za-z0-9](?:[A-Za-z0-9-]*[A-Za-z0-9])?\.top)
+        \s*$
+      /ix
+
       property_supported :available? do
-        super() || content_for_scanner.match?(
-          /^The queried object does not exist:\s+(?:[A-Za-z0-9](?:[A-Za-z0-9-]*[A-Za-z0-9])?\.)*[A-Za-z0-9](?:[A-Za-z0-9-]*[A-Za-z0-9])?\.top\s*$/i
-        )
+        raise ResponseIsThrottled if response_throttled?
+        raise ResponseIsUnavailable if response_unavailable?
+
+        super() || content_for_scanner.match?(NO_RECORD_MARKER)
+      end
+
+      def response_unavailable?
+        body = content_for_scanner
+        ambiguous_absence = body.match?(NO_RECORD_MARKER) && body.match?(RECORD_EVIDENCE)
+
+        super || body.match?(/^\s*access denied\b/i) ||
+          ambiguous_absence
+      end
+
+      def response_throttled?
+        super || content_for_scanner.match?(/^\s*too many requests\b/i)
       end
     end
   end

data/lib/whois/parsers/whois.nic.tv.rb

@@ -7,9 +7,30 @@ module Whois
     class WhoisNicTv < BaseIcannCompliant
       include RegistryResponseSafety
 
+      RECORD_EVIDENCE = /^(?:Domain Name|Registry Domain ID|Creation Date|Registrar|Domain Status):[ \t]*\S/i
+
       self.scanner = Scanners::BaseIcannCompliant, {
         pattern_available: /^No Data Found\n/,
       }
+
+      property_supported :available? do
+        raise ResponseIsThrottled if response_throttled?
+        raise ResponseIsUnavailable if response_unavailable?
+
+        super()
+      end
+
+      def response_unavailable?
+        body = content_for_scanner
+        ambiguous_absence = body.match?(/^No Data Found\s*$/i) && body.match?(RECORD_EVIDENCE)
+
+        super || body.match?(/^\s*access denied\b/i) ||
+          ambiguous_absence
+      end
+
+      def response_throttled?
+        super || content_for_scanner.match?(/^\s*too many requests\b/i)
+      end
     end
   end
 end

data/lib/whois/parsers/whois.nic.uk.rb

@@ -8,6 +8,7 @@
 
 
 require_relative 'base'
+require_relative 'registry_response_safety'
 
 
 module Whois
@@ -22,6 +23,22 @@ module Whois
     # @see http://www.nominet.org.uk/other/whois/detailedinstruct/
     #
     class WhoisNicUk < Base
+      include RegistryResponseSafety
+
+      DOMAIN_NAME_LINE = /^[ \t]*Domain name:[ \t]*(.*?)[ \t]*\r?$/i
+      NO_MATCH_LINE = /^[ \t]*No match for "([^"]+)"\.[ \t]*\r?$/i
+      NOT_REGISTERED_LINE = /^[ \t]*This domain name has not been registered\.[ \t]*\r?$/i
+      REGISTRATION_STATUS_LINE = /\A[ \t]*Registration status:[ \t]*(.*?)[ \t]*\r?\n?\z/i
+      REGISTRATION_SUPPORT_LINE = /^[ \t]*(?:Registered on|Expiry date|Last updated|Registrar):/i
+      REGISTRATION_EVIDENCE_LINE = /^[ \t]*(?:Domain name|Registration status|Registered on|Expiry date|Last updated|Registrar|Name servers):/i
+      REGISTRATION_STATUSES = {
+        'registered until expiry date.' => :registered,
+        'registration request being processed.' => :registered,
+        'renewal request being processed.' => :registered,
+        'no longer required' => :registered,
+        'renewal required.' => :registered,
+        'no registration status listed.' => :reserved,
+      }.freeze
 
       # == Values for Status
       #
@@ -29,37 +46,37 @@ module Whois
       # @see http://www.nominet.org.uk/registrants/maintain/renew/status/
       #
       property_supported :status do
-        if content_for_scanner =~ /\s+Registration status:\s+(.+?)\n/
-          case ::Regexp.last_match(1).downcase
-          when "registered until expiry date."
-            :registered
-          when "registration request being processed."
-            :registered
-          when "renewal request being processed."
-            :registered
-          when "no longer required"
-            :registered
-          when "no registration status listed."
-            :reserved
-          # NEWSTATUS redemption (https://github.com/weppos/whois/issues/5)
-          when "renewal required."
-            :registered
-          else
-            Whois::Parser.bug!(ParserError, "Unknown status `#{::Regexp.last_match(1)}'.")
-          end
-        elsif invalid?
+        body = content_for_scanner
+        absence_markers = body.lines.count do |line|
+          line.match?(NO_MATCH_LINE) || line.match?(NOT_REGISTERED_LINE)
+        end
+
+        if absence_markers.positive?
+          authoritative_absence?(body) ? :available : :unknown
+        elsif invalid? && !registration_evidence?(body)
           :invalid
         else
-          :available
+          domains = domain_name_values(body)
+          statuses = registration_status_values(body)
+          if domains.length != 1 || !domains.first.downcase.end_with?('.uk') || statuses.length != 1
+            :unknown
+          else
+            status = REGISTRATION_STATUSES[statuses.first.downcase]
+            if status == :registered && !body.match?(REGISTRATION_SUPPORT_LINE)
+              :unknown
+            else
+              status || :unknown
+            end
+          end
         end
       end
 
       property_supported :available? do
-        !!(content_for_scanner =~ /This domain name has not been registered/)
+        status == :available
       end
 
       property_supported :registered? do
-        !available?
+        status == :registered
       end
 
 
@@ -149,7 +166,51 @@ module Whois
       #   and will be replenished in 50 seconds.
       #
       def response_throttled?
-        !!(content_for_scanner =~ /The WHOIS query quota for .+ has been exceeded/)
+        content_for_scanner.match?(/^[ \t]*The WHOIS query quota for .+ has been exceeded\b/i) ||
+          RegistryResponseSafety.instance_method(:response_throttled?).bind_call(self)
+      end
+
+      def registration_status_values(body)
+        lines = body.lines
+        lines.each_with_index.filter_map do |line, index|
+          match = REGISTRATION_STATUS_LINE.match(line)
+          next unless match
+
+          value = match[1].strip
+          if value.empty?
+            next_index = index + 1
+            next_index += 1 while next_index < lines.length && lines[next_index].strip.empty?
+            value = lines[next_index].to_s.strip
+          end
+          value unless value.empty?
+        end
+      end
+
+      def domain_name_values(body)
+        lines = body.lines
+        lines.each_with_index.filter_map do |line, index|
+          match = DOMAIN_NAME_LINE.match(line)
+          next unless match
+
+          value = match[1].strip
+          if value.empty?
+            next_index = index + 1
+            next_index += 1 while next_index < lines.length && lines[next_index].strip.empty?
+            value = lines[next_index].to_s.strip.split(/\s+/, 2).first.to_s
+          end
+          value unless value.empty?
+        end
+      end
+
+      def authoritative_absence?(body)
+        no_match = body.lines.filter_map { |line| NO_MATCH_LINE.match(line)&.captures&.first }
+        not_registered = body.lines.count { |line| line.match?(NOT_REGISTERED_LINE) }
+        no_match.length == 1 && no_match.first.downcase.end_with?('.uk') &&
+          not_registered == 1 && !registration_evidence?(body) && !invalid?
+      end
+
+      def registration_evidence?(body)
+        body.match?(REGISTRATION_EVIDENCE_LINE)
       end
 
 

data/lib/whois/parsers/whois.nic.vip.rb

@@ -5,9 +5,30 @@ module Whois
     # Identity Digital's current .vip endpoint uses the ICANN key/value
     # layout and reports an absent name as `No Data Found`.
     class WhoisNicVip < BaseIcannCompliant
+      RECORD_EVIDENCE = /^(?:Domain Name|Registry Domain ID|Creation Date|Registrar|Domain Status):[ \t]*\S/i
+
       self.scanner = Scanners::BaseIcannCompliant, {
         pattern_available: /^No Data Found\s*$/,
       }
+
+      property_supported :available? do
+        raise ResponseIsThrottled if response_throttled?
+        raise ResponseIsUnavailable if response_unavailable?
+
+        super()
+      end
+
+      def response_unavailable?
+        body = content_for_scanner
+        ambiguous_absence = body.match?(/^No Data Found\s*$/i) && body.match?(RECORD_EVIDENCE)
+
+        super || body.match?(/^\s*access denied\b/i) ||
+          ambiguous_absence
+      end
+
+      def response_throttled?
+        super || content_for_scanner.match?(/^\s*too many requests\b/i)
+      end
     end
   end
 end

data/lib/whois/parsers/whois.nic.wiki.rb

@@ -2,7 +2,28 @@ require_relative 'base_identity_digital'
 module Whois
   class Parsers
     class WhoisNicWiki < BaseIdentityDigital
+      RECORD_EVIDENCE = /^(?:Domain Name|Registry Domain ID|Creation Date|Registrar|Domain Status):[ \t]*\S/i
+
       self.scanner = Scanners::BaseIcannCompliant, { pattern_available: /^No Data Found\s*$/i }
+
+      property_supported :available? do
+        raise ResponseIsThrottled if response_throttled?
+        raise ResponseIsUnavailable if response_unavailable?
+
+        super()
+      end
+
+      def response_unavailable?
+        body = content_for_scanner
+        ambiguous_absence = body.match?(/^No Data Found\s*$/i) && body.match?(RECORD_EVIDENCE)
+
+        super || body.match?(/^\s*access denied\b/i) ||
+          ambiguous_absence
+      end
+
+      def response_throttled?
+        super || content_for_scanner.match?(/^\s*too many requests\b/i)
+      end
     end
   end
 end