pubid 2.0.0.pre.alpha.8 → 2.0.0.pre.alpha.9

data/lib/pubid/jcgm/urn_generator.rb

@@ -4,7 +4,9 @@ module Pubid
   module Jcgm
     class UrnGenerator < Pubid::UrnGenerator::Base
       def generate
-        if identifier.is_a?(SupplementIdentifier)
+        if identifier.is_a?(Identifiers::Meeting)
+          generate_meeting_urn
+        elsif identifier.is_a?(SupplementIdentifier)
           generate_supplement_urn
         else
           generate_base_urn
@@ -13,16 +15,21 @@ module Pubid
 
       protected
 
+      # urn:jcgm:meeting:<number>:<year>, e.g. urn:jcgm:meeting:17:2012
+      def generate_meeting_urn
+        parts = ["urn", "jcgm", "meeting"]
+        parts << identifier.number.value if identifier.number
+        parts << identifier.date.year if identifier.date
+        parts.join(":")
+      end
+
       def generate_base_urn
         parts = ["urn", "jcgm"]
 
-        gum_number = maybe(:gum_number)
-        if gum_number
-          number = gum_number.to_s
-          parts << "gum.#{number}"
+        if identifier.is_a?(Identifiers::GumGuide) && identifier.number
+          parts << "gum.#{identifier.number}"
         elsif identifier.number
-          number = identifier.number.to_s
-          parts << number
+          parts << identifier.number.to_s
         end
 
         part = maybe(:part)
@@ -55,9 +62,11 @@ module Pubid
           parts << "iter.#{iter}"
         end
 
+        # Guide and GUM-guide are already implied by the number form (bare
+        # number vs "gum.N"), so their type_code is not appended.
         if identifier.typed_stage
           type_code = identifier.typed_stage.type_code
-          if type_code && type_code.to_s != "guide"
+          if type_code && !%w[guide gum_guide].include?(type_code.to_s)
             parts << type_code.to_s
           end
         end
@@ -73,7 +82,7 @@ module Pubid
       def generate_supplement_urn
         parts = ["urn", "jcgm"]
 
-        base_id = maybe(:base_identifier)
+        base_id = maybe(:base)
         if base_id
           base_gen = self.class.new(base_id)
           base_urn = base_gen.generate_base_urn
@@ -88,10 +97,9 @@ module Pubid
             parts << supp_type if supp_type
           end
 
-          iteration = maybe(:iteration)
-          if iteration
-            iter = iteration.to_s
-            parts << iter
+          # The supplement's own number (the "1" in "/Cor 1") — NOT iteration.
+          if identifier.number
+            parts << identifier.number.to_s
           end
 
           if identifier.date&.is_a?(::Pubid::Components::Date) && identifier.date.present?

data/lib/pubid/jcgm/urn_parser.rb

@@ -4,20 +4,32 @@ module Pubid
   module Jcgm
     # Parses JCGM URNs back into identifiers.
     #
-    # UrnGenerator emits: `urn:jcgm:{number}:{year}`.
+    # UrnGenerator emits `urn:jcgm:{number}:{year}` for guides and
+    # `urn:jcgm:meeting:{number}:{year}` for committee/meeting records.
     #
     # Examples:
     # - urn:jcgm:200:2012 → JCGM 200:2012
     # - urn:jcgm:100:2008 → JCGM 100:2008
+    # - urn:jcgm:meeting:17:2012 → JCGM 17th Meeting (2012)
     class UrnParser < Pubid::UrnParser::Base
       def parse_urn(urn)
-        body = strip_namespace(urn)
-        parts = split_parts(body)
+        parts = split_parts(strip_namespace(urn))
+        return parse_meeting_urn(parts) if parts.first == "meeting"
+
         number, year = parts
         text = "JCGM #{number}"
         text += ":#{year}" if year
         flavor_parse(text)
       end
+
+      private
+
+      # urn:jcgm:meeting:<number>:<year> -> "JCGM <ordinal> Meeting (<year>)"
+      def parse_meeting_urn(parts)
+        _, number, year = parts
+        ordinal = Identifiers::Meeting.ordinal(number)
+        flavor_parse("JCGM #{ordinal} Meeting (#{year})")
+      end
     end
   end
 end

data/lib/pubid/jis/identifier.rb

@@ -117,6 +117,58 @@ module Pubid
         render(format: :human, **opts)
       end
 
+      # JIS exposes its publisher as the `PUBLISHER` constant rather than the
+      # inherited `publisher` attribute, so the generic mr_publisher returns
+      # nil. Without this override every JIS id would lose its "JIS." prefix.
+      # Lowercased to match the all-lowercase MR convention.
+      def mr_publisher
+        PUBLISHER.downcase
+      end
+
+      # JIS stores its identity across `series` (division letter A–Z), `number`
+      # (string, may have leading zeros), and `parts` (multi-level). The
+      # generic mr_number_with_part only knows about `number`/`part`/`subpart`,
+      # so without these overrides every JIS id would collapse onto its
+      # document number and drop the division letter (issue #142).
+      # Lowercased to match the all-lowercase MR convention.
+      def mr_type
+        return nil unless respond_to?(:type_prefix)
+
+        type_prefix&.downcase
+      end
+
+      def mr_number_with_part
+        segments = []
+        segments << series.to_s.downcase if series
+        segments << number.to_s if number
+        parts&.each { |p| segments << p.to_s }
+        return nil if segments.empty?
+
+        segments.join("-")
+      end
+
+      # JIS stores the year as a bare integer (not a Components::Date), so the
+      # inherited mr_year (which reads `date`) returns nil. Emit it directly,
+      # including the reaffirmation marker so `JIS X:2019` and `JIS X:2019R`
+      # stay distinct in MR.
+      def mr_year
+        return nil unless year
+
+        year_with_reaffirmation.to_s
+      end
+
+      def mr_languages
+        return nil unless language
+
+        "(#{language})"
+      end
+
+      def mr_all_parts
+        return nil unless all_parts?
+
+        "all-parts"
+      end
+
       # from_hash is the shared polymorphic dispatch on Pubid::Identifier.
       # JIS_TYPE_MAP remains as the key_value polymorphic_map.
 

data/lib/pubid/nist/builder.rb

@@ -22,7 +22,7 @@ module Pubid
 
       # Build an identifier object from parsed data
       # @param parsed [Hash, Array] the parsed identifier data
-      # @return [Identifiers::Base] the constructed identifier object
+      # @return [Identifier] the constructed identifier object
       def build(parsed)
         # Parslet can return array of hashes - merge them
         parsed_hash = parsed.is_a?(Array) ? flatten_array(parsed) : parsed
@@ -287,8 +287,6 @@ module Pubid
 
               identifier.number = Components::Code.new(value: "#{first_num.value}-#{number_part}")
               identifier.edition = edition_obj
-              identifier.edition_component = edition_obj
-              identifier.revision = "r#{edition_id}"
             # CS Emergency pattern: e104-43 -> number=104, edition_year=1943
             # Logic: e104-43 means "emergency 104 from 1943" (43 = 1943)
             elsif first_num.value.to_s.match?(/^e(\d{3})$/) &&
@@ -304,7 +302,6 @@ module Pubid
 
               identifier.number = Components::Code.new(value: number_part)
               identifier.edition = edition_obj
-              identifier.edition_component = edition_obj
             elsif first_num.value.to_s.match?(/^(\d+)e(\d+)$/) &&
                 second_num.value.to_s.match?(/^\d{2,4}$/)
               # Pattern: "11e2-1915" OR "123e2-50" parsed as first="11e2"|"123e2", second="1915"|"50"
@@ -326,7 +323,6 @@ module Pubid
               edition_obj = Components::Edition.new(type: "e",
                                                     id: edition_id, additional_text: year_part)
               identifier.edition = edition_obj
-              identifier.edition_component = edition_obj
             elsif first_num.value.to_s.match?(/^(\d+)supp?$/) &&
                 second_num.value.to_s.match?(/^\d{4}$/)
               # Pattern: "25supp-1924" parsed as first="25supp", second="1924"
@@ -351,7 +347,6 @@ module Pubid
               identifier.number = first_num
               edition_obj = Components::Edition.new(type: "e",
                                                     id: second_num.value.to_s)
-              identifier.edition_component = edition_obj
               identifier.edition = edition_obj
               identifier.edition_year = second_num.value.to_s
             elsif part_num && series.part_num_as_component?
@@ -379,11 +374,34 @@ module Pubid
                                                        id: extracted_revision.to_s)
         end
 
+        # Expose the SP subseries (Appendix A.2 of the NIST PubID Syntax) as
+        # a queryable attribute. For "NIST SP 800-53" the subseries is "800"
+        # (first_num) and the sequence is "53" (already folded into number).
+        # Pure metadata extraction — number is left intact so existing
+        # rendering and indexing stay byte-identical.
+        if identifier.is_a?(Identifiers::SpecialPublication) &&
+            first_num && identifier.respond_to?(:subseries=)
+          first_str = first_num.value.to_s
+          subseries_code =
+            if Identifiers::SpecialPublication::SP_SUBSERIES.include?(first_str)
+              first_str
+            elsif (m = first_str.match(/\A(\d+GB)\b/)) &&
+                Identifiers::SpecialPublication::SP_SUBSERIES.include?(m[1])
+              # Parser emits "1190GB-12" as a single first_number token; strip
+              # the suffix to surface the bare "1190GB" subseries code.
+              m[1]
+            end
+          if subseries_code
+            identifier.subseries =
+              Components::Code.new(value: subseries_code)
+          end
+        end
+
         # Series-specific post-processing (e.g., IR reverses the "84e2946"
         # form that preprocessing produced back to "84-2946").
         series.finalize_identifier(identifier, parsed_hash)
 
-        # publisher_was_parsed defaults to true (see Identifiers::Base), so only
+        # publisher_was_parsed defaults to true (see Identifier), so only
         # the prefix-less case needs recording: assign false when no publisher
         # was parsed/extracted, and leave it unset (default true, omitted from
         # to_hash) when one was. Keeps the common publisher-bearing id lean.
@@ -437,11 +455,13 @@ module Pubid
           component = Components::Supplement.new
           # Split the fused "Jun1925"/"Jun1926" strings into isolated start/end
           # month+year nodes (start reuses :month/:year, end uses *_end).
-          if range_start && (m = range_start.match(/\A([A-Za-z]{3,9})(\d{4})\z/))
+          # Each end is either a fused "Jun1925" (month+year) or a bare "1925"
+          # (year-only range, no month).
+          if range_start && (m = range_start.match(/\A([A-Za-z]{3,9})?(\d{4})\z/))
             component.month = m[1]
             component.year = m[2]
           end
-          if range_end && (m = range_end.match(/\A([A-Za-z]{3,9})(\d{4})\z/))
+          if range_end && (m = range_end.match(/\A([A-Za-z]{3,9})?(\d{4})\z/))
             component.month_end = m[1]
             component.year_end = m[2]
           end

data/lib/pubid/nist/caster.rb

@@ -32,7 +32,7 @@ module Pubid
         when :publisher
           return nil if value.nil? || value.to_s.strip.empty?
 
-          # publisher is a plain string attribute (see Identifiers::Base).
+          # publisher is a plain string attribute (see Identifier).
           value.to_s
 
         when :dated_date
@@ -48,6 +48,11 @@ module Pubid
 
           { dated_seq: value.to_s }
 
+        when :year
+          # Bare 4-digit year (e.g. from "NIST Research Library (2022)").
+          # The Identifier#year attribute is typed as :integer.
+          value&.to_s&.to_i
+
         when :series
           return nil if value.nil? || value.to_s.strip.empty?
 
@@ -196,9 +201,6 @@ module Pubid
                                            value: data[:letter_suffix].to_s.upcase),
                 edition: Components::Edition.new(type: "r",
                                                  id: data[:revision_id].to_s),
-                edition_component: Components::Edition.new(type: "r",
-                                                            id: data[:revision_id].to_s),
-                revision: "r#{data[:revision_id]}",
               }
             end
 
@@ -633,7 +635,7 @@ module Pubid
 
         when :parsed_format
           # Format detection result from parser. :short is the render default
-          # (a nil parsed_format renders short — see Identifiers::Base#to_s), so
+          # (a nil parsed_format renders short — see Identifier#to_s), so
           # store only non-default formats (e.g. "mr"); "short" stays unset and
           # is omitted from to_hash. detect_format only emits :mr or :short.
           v = value&.to_s
@@ -877,8 +879,6 @@ module Pubid
           # Store as id directly - renders as "e199206"
           {
             edition: Components::Edition.new(type: "e", id: edition_date),
-            edition_component: Components::Edition.new(type: "e",
-                                                       id: edition_date),
           }
 
         when :edition_e
@@ -889,8 +889,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "e", id: edition_id),
-            edition_component: Components::Edition.new(type: "e",
-                                                       id: edition_id),
           }
 
         when :edition_r
@@ -901,9 +899,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "r", id: edition_id),
-            edition_component: Components::Edition.new(type: "r",
-                                                       id: edition_id),
-            revision: "r#{edition_id}", # Also set revision string attribute for compatibility
           }
 
         when :edition_r_no_space
@@ -915,9 +910,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "r", id: edition_id),
-            edition_component: Components::Edition.new(type: "r",
-                                                       id: edition_id),
-            revision: "r#{edition_id}", # Also set revision string attribute for compatibility
           }
 
         when :edition_rev
@@ -928,9 +920,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "r", id: edition_id),
-            edition_component: Components::Edition.new(type: "r",
-                                                       id: edition_id),
-            revision: "r#{edition_id}", # Also set revision string attribute for compatibility
           }
 
         when :edition_r_letter
@@ -943,10 +932,6 @@ module Pubid
           {
             edition: Components::Edition.new(type: "r", id: edition_id,
                                              additional_text: edition_letter),
-            edition_component: Components::Edition.new(type: "r",
-                                                       id: edition_id,
-                                                       additional_text: edition_letter),
-            revision: "r#{edition_id}#{edition_letter}", # Also set revision string attribute for compatibility
           }
 
         when :edition_r_letter_only
@@ -957,9 +942,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "r", id: edition_letter),
-            edition_component: Components::Edition.new(type: "r",
-                                                       id: edition_letter),
-            revision: "r#{edition_letter}", # Also set revision string attribute for compatibility
           }
 
         when :edition_historical
@@ -970,8 +952,6 @@ module Pubid
 
           {
             edition: Components::Edition.new(type: "-", id: edition_id),
-            edition_component: Components::Edition.new(type: "-",
-                                                       id: edition_id),
           }
 
         when :edition_r_with_space_letter
@@ -991,10 +971,6 @@ module Pubid
             {
               edition: Components::Edition.new(type: "r", id: edition_id,
                                                additional_text: edition_letter),
-              edition_component: Components::Edition.new(type: "r",
-                                                         id: edition_id,
-                                                         additional_text: edition_letter),
-              revision: "r#{edition_id}#{edition_letter}",
             }
           else
             # Space was in original input - preserve format
@@ -1002,11 +978,6 @@ module Pubid
               edition: Components::Edition.new(type: "r", id: edition_id,
                                                additional_text: edition_letter,
                                                original_prefix: " r"),
-              edition_component: Components::Edition.new(type: "r",
-                                                         id: edition_id,
-                                                         additional_text: edition_letter,
-                                                         original_prefix: " r"),
-              revision: "r#{edition_id}#{edition_letter}",
             }
           end
 
@@ -1027,19 +998,12 @@ module Pubid
             # No original_prefix - space was added by preprocessing
             {
               edition: Components::Edition.new(type: "r", id: edition_id),
-              edition_component: Components::Edition.new(type: "r",
-                                                         id: edition_id),
-              revision: "r#{edition_id}",
             }
           else
             # Space was in original input - preserve format
             {
               edition: Components::Edition.new(type: "r", id: edition_id,
                                                original_prefix: " r"),
-              edition_component: Components::Edition.new(type: "r",
-                                                         id: edition_id,
-                                                         original_prefix: " r"),
-              revision: "r#{edition_id}",
             }
           end
 
@@ -1089,7 +1053,6 @@ module Pubid
                 )
                 return {
                   edition: edition_obj,
-                  edition_component: edition_obj,
                   edition_year: edition_id.to_s,
                 }
               else
@@ -1113,7 +1076,6 @@ module Pubid
           # Return as hash to set edition and edition_year
           {
             edition: edition_obj, # Main attribute for tests
-            edition_component: edition_obj, # V2 component
             edition_year: value.to_s, # Keep string for render logic
           }
 

data/lib/pubid/nist/circular_supplement_builder.rb

@@ -25,7 +25,7 @@ module Pubid
         @builder = builder
       end
 
-      # Build CircularSupplement with base_identifier wrapping
+      # Build CircularSupplement with base wrapping
       # @param parsed_hash [Hash] the parsed supplement data
       # Build a CIRC/LCIRC supplement. Most forms collapse onto the base
       # identifier's normal class (Circular / LetterCircular) with isolated
@@ -56,10 +56,12 @@ module Pubid
           ys = range[:supp_year_start]&.to_s
           me = range[:supp_month_end]&.to_s
           ye = range[:supp_year_end]&.to_s
+          # Month is optional: a year-only range ("sup1925-1926") passes just the
+          # bare year, which supplement_from parses into year/year_end.
           identifier.supplement = @builder.supplement_from(
             value: nil, has_revision: false,
-            range_start: (ms && ys ? "#{ms}#{ys}" : nil),
-            range_end: (me && ye ? "#{me}#{ye}" : nil)
+            range_start: (ys ? "#{ms}#{ys}" : nil),
+            range_end: (ye ? "#{me}#{ye}" : nil)
           )
           return identifier
         end
@@ -216,7 +218,7 @@ module Pubid
 
           # Recursively build base identifier
           # This will go through normal build() process which extracts edition from "101e2"
-          supplement.base_identifier = @builder.build(base_hash)
+          supplement.base = @builder.build(base_hash)
 
           # NEW: Handle revision + implicit supplement pattern (e.g., "145r11/1925")
           # Create update format: "Upd1-{year}{revision_number}"
@@ -246,7 +248,7 @@ module Pubid
           }
 
           # Recursively build base identifier
-          supplement.base_identifier = @builder.build(base_hash)
+          supplement.base = @builder.build(base_hash)
         end
 
         # Build supplement edition from captured data

data/lib/pubid/nist/components/supplement.rb

@@ -139,8 +139,9 @@ module Pubid
         end
 
         # A range is marked by the presence of an END; START reuses month/year.
+        # Months are optional, so a year-only range ("1925-1926") also counts.
         def date_range?
-          month && year && month_end && year_end
+          year && year_end
         end
 
         def build_date_range_format
@@ -164,7 +165,9 @@ module Pubid
         end
 
         def build_long_date_range_format
-          "Supplement #{month} #{year}-#{month_end} #{year_end}"
+          start = [month, year].compact.join(" ")
+          finish = [month_end, year_end].compact.join(" ")
+          "Supplement #{start}-#{finish}"
         end
 
         def build_long_month_year_format

data/lib/pubid/nist/configuration.rb

@@ -58,7 +58,13 @@ module Pubid
       private
 
       def default_series_path
-        File.join(__dir__, "../../../gems/pubid-nist/series.yaml")
+        # Prefer data/nist/series.yaml (the canonical home in the monorepo);
+        # fall back to archived-gems/pubid-nist/series.yaml for backward
+        # compat with the v1 layout.
+        monorepo = File.join(__dir__, "../../../data/nist/series.yaml")
+        return monorepo if File.exist?(monorepo)
+
+        File.join(__dir__, "../../../archived-gems/pubid-nist/series.yaml")
       end
 
       def load_series

data/lib/pubid/nist/identifier.rb

@@ -3,8 +3,6 @@
 # The NIST flavor base class is Pubid::Nist::Identifier — a real
 # Pubid::Identifier subclass that every concrete NIST identifier descends from,
 # so `id.is_a?(Pubid::Nist::Identifier)` and the shared polymorphic `from_hash`
-# work natively (no facade needed). Its full body, `.parse`, and the
-# Identifiers::Base back-compat alias are defined in identifiers/base.rb to keep
+# work natively (no facade needed). Its full body and `.parse` are defined in identifiers/base.rb to keep
 # that large body in one place; this file just ensures it is loaded when a
 # consumer references Pubid::Nist::Identifier directly.
-require_relative "identifiers/base"