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

data/lib/pubid/iec/parser.rb

@@ -19,6 +19,7 @@ module Pubid
       rule(:identifier) do
         working_programme_with_publisher |
           working_programme |
+          technical_group |
           working_document |
           sheet_supplement_identifier |
           sheet_identifier |
@@ -29,6 +30,42 @@ module Pubid
           identifier_copublishers
       end
 
+      # Technical Group identifier — names a committee rather than a document.
+      # Examples: "IEC TC 1", "ISO/IEC JTC 1", "ISO/IEC JTC 1/SC 7",
+      #           "IEC SC 100A", "CISPR", "CIS/A".
+      rule(:technical_group) do
+        technical_group_with_publisher | technical_group_bare
+      end
+
+      rule(:technical_group_with_publisher) do
+        prefix_sole_publisher >> copublishers.maybe >> space >>
+          tc_body.as(:technical_committee) >>
+          (str("/") >> space? >> sc_body).as(:subcommittee).maybe
+      end
+
+      rule(:technical_group_bare) do
+        # CISPR / CIS-letter forms and bare TYPE NUMBER forms both default
+        # publisher to "IEC" via the Base class default.
+        tc_body.as(:technical_committee) >>
+          (str("/") >> space? >> sc_body).as(:subcommittee).maybe
+      end
+
+      # Committee body forms accepted by the parser. Order matters: longer
+      # or more specific patterns first so they win the ordered choice.
+      rule(:tc_body) do
+        str("CISPR") |
+          (str("CIS") >> str("/") >> match("[A-Z]")) |
+          (tc_type >> space >> digits >> match("[A-Z]").maybe)
+      end
+
+      rule(:sc_body) do
+        str("SC") >> space >> digits >> match("[A-Z]").maybe
+      end
+
+      rule(:tc_type) do
+        str("JTC") | str("JPC") | str("TC") | str("SC") | str("PC")
+      end
+
       # Working Programme format: [PWI|PNW] [TR] number edition
       # Examples: "PWI TR 100-36 ED1", "PWI 100-44 ED1", "PNW 65-915 ED1"
       # TC-style names: "PWI TR SYCSMARTENERGY-1", "PWI SyCLVDC-1 ED1"
@@ -158,12 +195,16 @@ module Pubid
       end
 
       rule(:date) do
-        # :2005-02 or optional for DB without date
+        # :2005-02 or :-- (undated reference — ISO/IEC undated-reference form,
+        # e.g. IEC 60050:--, also accepted as the em-dash variant IEC 60050:—)
         (str(":") | dash) >>
-          (year_digits >>
-            (dash >> month_digits).maybe >>
-            (dash >> day_digits).maybe
-          ).as(:date)
+          (
+            (str("--") | str("—")).as(:undated_marker) |
+              (year_digits >>
+                (dash >> month_digits).maybe >>
+                (dash >> day_digits).maybe
+              ).as(:date)
+          )
       end
 
       rule(:part_and_subpart) do
@@ -251,14 +292,14 @@ module Pubid
       # IEC 60695-2-1/1:1994 (sheet 1 from year 1994)
       # IEC 60695-2-1/1 (sheet 1 without year)
       rule(:sheet_identifier) do
-        identifier_copublishers_no_third.as(:base_identifier) >>
+        identifier_copublishers_no_third.as(:base) >>
           (sheet_notation | sheet_notation_no_year)
       end
 
       # Sheet identifier followed by supplement: base/sheet/supplement
       # IEC 60695-2-1/1:1994/COR1:1995
       rule(:sheet_supplement_identifier) do
-        identifier_copublishers_no_third.as(:base_identifier) >>
+        identifier_copublishers_no_third.as(:base) >>
           sheet_notation >>
           str("/") >> supplement_type_with_stage >>
           space? >> second_part >> third_part
@@ -353,7 +394,7 @@ module Pubid
       # ISO/IEC 17025:2017 handles copublisher in copublishers rule
       # We handle this in identifier_copublishers already
       rule(:joint_identifier) do
-        identifier_copublishers_no_third.as(:base_identifier) >>
+        identifier_copublishers_no_third.as(:base) >>
           space? >> str("|") >> space? >>
           scope { iso_identifier }.as(:joint_identifier)
       end
@@ -381,7 +422,7 @@ module Pubid
 
       # IEC 60038:2009/Amd 1:2011
       rule(:supplement_identifier_no_third) do
-        identifier_copublishers_no_third.as(:base_identifier) >>
+        identifier_copublishers_no_third.as(:base) >>
           str("/") >> supplement_type_with_stage >>
           space? >> second_part
       end
@@ -392,7 +433,7 @@ module Pubid
 
       # IEC 60038:2009/Amd 1:2011/Cor 1:2012
       rule(:supplement_supplement_identifier) do
-        supplement_identifier_no_third.as(:base_identifier) >>
+        supplement_identifier_no_third.as(:base) >>
           str("/") >> supplement_type_with_stage >>
           space? >> second_part >> third_part
       end

data/lib/pubid/iec/renderer.rb

@@ -37,6 +37,8 @@ module Pubid
           render_fragment(id, opts, context)
         when Identifiers::WorkingDocument
           render_working_document(id, context)
+        when Identifiers::TechnicalGroup
+          render_technical_group(id, context)
         when Identifiers::TestReportForm
           render_test_report_form(id, context)
         when SupplementIdentifier
@@ -75,11 +77,11 @@ module Pubid
       end
 
       def render_supplement(id, opts, context)
-        id.ensure_base_identifier
+        id.ensure_base
 
         if id.synthetic_base?
           render_supplement_standalone(id, context)
-        elsif id.base_identifier
+        elsif id.base
           render_supplement_attached(id, opts, context)
         else
           render_single(id, opts, context)
@@ -97,7 +99,7 @@ module Pubid
         end
 
         abbr = id.typed_stage.abbr.first
-        number_str = id.base_identifier.number.to_s
+        number_str = id.base.number.to_s
         number_str += "-#{id.number}" if id.number
         number_str += "-#{id.subpart}" if id.subpart
         parts << "#{abbr} #{number_str}"
@@ -110,7 +112,7 @@ module Pubid
 
       def render_supplement_attached(id, opts, context)
         parts = []
-        parts << id.base_identifier.to_s(**opts)
+        parts << id.base.to_s(**opts)
 
         abbr = id.typed_stage.abbr.first.upcase
         supp_part = "/#{abbr}#{id.number}"
@@ -155,7 +157,7 @@ module Pubid
       def render_vap(id, opts, context)
         parts = []
 
-        parts << id.base_identifier.to_s(**opts.merge(with_edition: false))
+        parts << id.base.to_s(**opts.merge(with_edition: false))
         parts << " #{id.vap.join('-')}" if id.vap&.any?
         parts << " #{id.edition.render(context:)}" if id.edition&.number
 
@@ -164,7 +166,7 @@ module Pubid
 
       def render_sheet(id, opts, context)
         parts = []
-        parts << id.base_identifier.to_s(**opts)
+        parts << id.base.to_s(**opts)
         parts << "/#{id.sheet_number}"
         parts << ":#{id.year}" if id.year
         parts.join
@@ -172,9 +174,9 @@ module Pubid
 
       def render_fragment(id, opts, context)
         parts = []
-        parts << id.base_identifier.to_s(**opts)
+        parts << id.base.to_s(**opts)
 
-        parts << if id.base_identifier.is_a?(Identifiers::Corrigendum)
+        parts << if id.base.is_a?(Identifiers::Corrigendum)
                    "/FRAGC#{id.fragment_number}"
                  else
                    "/FRAG#{id.fragment_number}"
@@ -185,6 +187,14 @@ module Pubid
         parts.join
       end
 
+      def render_technical_group(id, _context)
+        parts = []
+        parts << id.publisher_portion
+        parts << " #{id.technical_committee}"
+        parts << id.subcommittee if id.subcommittee
+        parts.join
+      end
+
       def render_working_document(id, context)
         if id.wp_stage
           parts = []

data/lib/pubid/iec/single_identifier.rb

@@ -48,7 +48,7 @@ module Pubid
           (part ? "-#{part}" : ""),
           (subpart ? "-#{subpart}" : ""),
           (stage_iteration ? ".#{stage_iteration}" : ""),
-          (date ? ":#{date.year}" : ""),
+          (date ? ":#{date.render}" : ""),
         ].join
       end
 

data/lib/pubid/iec/supplement_identifier.rb

@@ -4,7 +4,7 @@ module Pubid
   module Iec
     # Identifier that represents a supplement to a base identifier.
     class SupplementIdentifier < SingleIdentifier
-      attribute :base_identifier, Identifier, polymorphic: true
+      attribute :base, Identifier, polymorphic: true
 
       # Serialize the supplemented document under "base". The supplement's own
       # number/date/stage come from the inherited common maps; route the nested
@@ -15,10 +15,10 @@ module Pubid
       end
 
       def base_to_kv(model, doc)
-        base = model.base_identifier
+        base = model.base
         return unless base
 
-        # `base_identifier` synthesises a base for a supplement that has none
+        # `base` synthesises a base for a supplement that has none
         # (e.g. a bare "+AMD1:2008" nested in a consolidation): a fake doc whose
         # number equals the supplement's own. It renders nothing and the getter
         # recreates it identically on load, so don't persist it. A standalone
@@ -34,7 +34,7 @@ module Pubid
         return unless value
 
         base = ::Pubid::Iec::Identifier.from_hash(value)
-        model.base_identifier = base
+        model.base = base
 
         # A base of the exact SingleIdentifier class is the synthetic self-base
         # of a standalone supplement ("IEC/FDAM 60038-1"); a real attached base
@@ -55,18 +55,18 @@ module Pubid
         @supplement_number = @number
       end
 
-      # Override base_identifier getter to ensure it's always created for standalone supplements
-      def base_identifier
-        @base_identifier || ensure_base_identifier
+      # Override base getter to ensure it's always created for standalone supplements
+      def base
+        @base || ensure_base
       end
 
       # Override number getter to trigger swap for standalone supplements
       def number
-        # Guard against recursion in ensure_base_identifier
+        # Guard against recursion in ensure_base
         return @number if @ensuring_base
 
         # For standalone supplements, trigger base creation first
-        ensure_base_identifier if @synthetic_base.nil? && publisher && part
+        ensure_base if @synthetic_base.nil? && publisher && part
 
         if @synthetic_base
           @supplement_number || @number
@@ -85,7 +85,7 @@ module Pubid
         end
       end
 
-      # Delegate publisher and copublishers to base_identifier whenever there is
+      # Delegate publisher and copublishers to base whenever there is
       # a base (a supplement carries the publisher of the document it
       # supplements). Checking base first — rather than `@publisher || base` — is
       # required now that the publisher attribute has an IEC default, which would
@@ -93,20 +93,20 @@ module Pubid
       def publisher
         return @publisher if @ensuring_base
 
-        base_identifier ? base_identifier.publisher : @publisher
+        base ? base.publisher : @publisher
       end
 
       def copublishers
         return @copublishers if @ensuring_base
 
-        base_identifier ? base_identifier.copublishers : @copublishers
+        base ? base.copublishers : @copublishers
       end
 
-      # Ensure we have a base_identifier for standalone supplements
+      # Ensure we have a base for standalone supplements
       # Creates a synthetic base identifier from supplement attributes if needed
-      # Returns the base_identifier
-      def ensure_base_identifier
-        return @base_identifier if @base_identifier
+      # Returns the base
+      def ensure_base
+        return @base if @base
 
         # Guard flag to prevent recursion in number getter
         @ensuring_base = true
@@ -140,14 +140,14 @@ module Pubid
           stage_code: :undated,
         )
 
-        @base_identifier = base
+        @base = base
 
         # Store the supplement number for getter override
-        # The original number is the base number (moved to base_identifier)
+        # The original number is the base number (moved to base)
         # The original part is the supplement number (what we want to return)
         @supplement_number = supplement_number
 
-        @base_identifier
+        @base
       ensure
         @ensuring_base = false
       end
@@ -160,6 +160,19 @@ module Pubid
       def to_s(**opts)
         render(format: :human, **opts)
       end
+
+      # MR supplement suffix: `/{type}.{number}.{year}` (e.g. "/amd.1.2016").
+      # The MrString renderer recurses into `base` and appends this so the
+      # full supplement chain round-trips losslessly (issue #142).
+      def mr_supplement_suffix
+        segments = []
+        segments << mr_type if mr_type
+        segments << number&.to_s if number
+        segments << date&.year&.to_s if date&.year
+        return nil if segments.empty?
+
+        segments.join(".")
+      end
     end
   end
 end

data/lib/pubid/iec.rb

@@ -23,12 +23,7 @@ module Pubid
     autoload :Renderer, "#{__dir__}/iec/renderer"
 
     def self.parse(identifier_string)
-      if identifier_string.length > Pubid::MAX_INPUT_LENGTH
-        raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
-      end
-
-      parsed = Parser.new.parse(identifier_string)
-      Builder.new.build(parsed)
+      Identifier.parse(identifier_string)
     end
 
     def self.parse_urn(urn)

data/lib/pubid/ieee/aiee/builder.rb

@@ -23,7 +23,7 @@ module Pubid
                               end
 
           # Extract code
-          code_str = extract_value(parsed[:number])
+          code_str = extract_number(parsed[:number])
           if code_str
             attributes[:code] = Components::Code.parse(code_str)
           end
@@ -36,9 +36,11 @@ module Pubid
           month_str = extract_value(parsed[:month])
           attributes[:month] = month_str if month_str
 
-          # Extract separator (for rendering distinction)
+          # Extract the long-form date separator ("," or ".") for rendering.
+          # Named `date_separator` on the identifier to avoid colliding with the
+          # CodeNumber code-part `separator`.
           separator_str = extract_value(parsed[:separator])
-          attributes[:separator] = separator_str if separator_str
+          attributes[:date_separator] = separator_str if separator_str
 
           # Set original format based on parsed data
           attributes[:original_format] = if separator_str || month_str
@@ -53,6 +55,18 @@ module Pubid
 
         private
 
+        # Extract the number, flattening a "431 (105)" main-number/parenthetical
+        # subtree into a plain string. Without this, a Hash node leaks its raw
+        # Parslet inspect form (e.g. `{main_number: "431"@8, ...}`) into `to_s`.
+        def extract_number(value)
+          return extract_value(value) unless value.is_a?(Hash)
+
+          main = extract_value(value[:main_number])
+          paren = value[:parenthetical]
+          alt = paren.is_a?(Hash) ? extract_value(paren[:alt_number]) : nil
+          alt ? "#{main} (#{alt})" : main
+        end
+
         def extract_value(value)
           return nil if value.nil?
           return nil if value.is_a?(Array) && value.empty?

data/lib/pubid/ieee/aiee/identifier.rb

@@ -1,88 +1,73 @@
 # frozen_string_literal: true
 
-require "lutaml/model"
-
 module Pubid
   module Ieee
     module Aiee
-      # Base class for AIEE identifiers
-      # AIEE (American Institute of Electrical Engineers): 1884-1963
-      class Identifier < Lutaml::Model::Serializable
-        attribute :publisher, :string, default: -> { "AIEE" }
-        attribute :type, :string # "No.", "No", "Standard", "Trans."
-        attribute :code, Components::Code
-        attribute :year, :string
-        attribute :month, :string
-        attribute :separator, :string # "," or "." for long form dates
-        attribute :original_format, :string # "short" or "long" (preserves parsed format)
-        attribute :relationships, Pubid::Ieee::Components::Relationship,
-                  collection: true
+      # AIEE identifiers (American Institute of Electrical Engineers: 1884-1963).
+      #
+      # Reparented onto Pubid::Ieee::Identifier + the CodeNumber mixin so AIEE
+      # shares the flat split-column serialization (number/prefix/parts/
+      # separator), a polymorphic `_type`, `#root`, and from_hash routing with
+      # every other IEEE type. Previously it descended from bare
+      # Lutaml::Model::Serializable and serialized a nested `code` object with no
+      # `_type`, so Pubid::Ieee::Identifier.from_hash could not route back to
+      # this class and dropped the code entirely — every AIEE row failed the
+      # relaton index-v2 round-trip gate and had a nil `root.number`.
+      #
+      # The date long-form separator is named `date_separator` (not `separator`)
+      # to avoid colliding with the CodeNumber code-part `separator`.
+      class Identifier < Pubid::Ieee::Identifier
+        include Pubid::Ieee::Identifiers::CodeNumber
 
-        def initialize(**args)
-          super()
-
-          # Handle number/code as string or Code object
-          # Builder passes :code, but also support :number for backward compatibility
-          code_value = args[:code] || args[:number]
-          if code_value.is_a?(String)
-            self.code = Components::Code.parse(code_value)
-          elsif code_value
-            self.code = code_value
-          end
-
-          # Set other attributes
-          attrs = self.class.attributes
-          args.each do |key, value|
-            next if %i[code number].include?(key)
-
-            setter = :"#{key}="
-            public_send(setter, value) if attrs.key?(key)
-          end
-        end
+        attribute :publisher, :string, default: -> { "AIEE" }
+        # AIEE carries no implicit type keyword, so override the base's "Std"
+        # default (else "AIEE 11-1937" would render as "AIEE Std 11-1937").
+        attribute :type, :string, default: -> {}
+        attribute :date_separator, :string  # "," or "." for long-form dates
+        attribute :original_format, :string # "short"/"long" (preserves parsed format)
 
-        # Provide number accessor for backward compatibility
-        # Returns the string representation of the code
-        def number
-          code&.to_s
+        # A distinct polymorphic name so from_hash routes to AIEE. The derived
+        # default (last name segment "Identifier") would be
+        # "pubid:ieee:identifier", colliding with the base class.
+        def self.polymorphic_name
+          "pubid:ieee:aiee"
         end
 
         # Parse AIEE identifier string
         def self.parse(input)
           parsed = Parser.new.parse(input)
-          builder = Builder.new
-          builder.build(parsed)
+          Builder.new.build(parsed)
         end
 
-        def to_s(date_format: nil)
+        def to_s(date_format: nil, trademark: false)
           result = [publisher]
           result << type if type
           result << code.to_s if code
 
           base = result.join(" ")
 
-          # Determine which format to use
-          # Priority: explicit parameter > original_format > default (long if month/separator, short otherwise)
-          format = date_format&.to_s || original_format
-
-          if !format
-            # Auto-detect from parsed attributes
-            format = month || separator ? "long" : "short"
+          # IEEE attaches the mark to the number, before the date suffix
+          # ("AIEE No 13™-1930").
+          if trademark
+            base += Pubid::Ieee.trademark_symbol_for(code_obj&.number,
+                                                     code_obj&.prefix,
+                                                     publishers: [publisher])
           end
 
-          # Date formatting based on format parameter
+          # Priority: explicit parameter > original_format > format auto-detect
+          format = date_format&.to_s || original_format
+          format ||= month || date_separator ? "long" : "short"
+
           if year
             case format
             when "short", :short
-              # Short form: dash + year
               base += "-#{year}"
             when "long", :long
-              # Long form: separator + optional month + year
-              sep = separator || "," # Default to comma for long form
+              sep = date_separator || "," # Default to comma for long form
               base += "#{sep} #{"#{month} " if month}#{year}"
             else
-              # Preserve original format (backward compatibility)
-              base += if separator
-                        "#{separator} #{"#{month} " if month}#{year}"
+              base += if date_separator
+                        "#{date_separator} #{"#{month} " if month}#{year}"
                       elsif month
                         ", #{month} #{year}"
                       else
@@ -91,11 +76,10 @@ module Pubid
             end
           end
 
-          # Add relationships if present
-          if relationships && !relationships.empty?
-            relationship_str = relationships.join(" / ")
-            base += " (#{relationship_str})"
-          end
+          # The descriptive relationships narrative is deliberately kept out of
+          # to_s (an identifier string must stay bounded — it is used as a
+          # document number/filename downstream). It stays reachable on
+          # `relationships`.
 
           base
         end