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

data/lib/pubid/ashrae/renderer.rb

@@ -56,48 +56,48 @@ module Pubid
 
       # Addendum: "ASHRAE Addendum a to Standard 15-2001"
       def render_addendum(id)
-        return id.base_identifier.to_s unless id.base_identifier
+        return id.base.to_s unless id.base
 
-        base_type = id.base_identifier.type || "Standard"
+        base_type = id.base.type || "Standard"
 
         if id.copublisher
-          "#{id.copublisher} Addendum #{id.addendum_code} to #{id.base_identifier}"
+          "#{id.copublisher} Addendum #{id.addendum_code} to #{id.base}"
         else
-          result = "ASHRAE Addendum #{id.addendum_code} to #{base_type} #{id.base_identifier.code}"
-          result += "-#{id.base_identifier.year}" if id.base_identifier.year
+          result = "ASHRAE Addendum #{id.addendum_code} to #{base_type} #{id.base.code}"
+          result += "-#{id.base.year}" if id.base.year
           result
         end.tap { |r| r << " (#{id.addendum_date})" if id.addendum_date }
       end
 
       # AddendaPackage: "ASHRAE Standard 52.2-1999: Addenda Supplement Package"
       def render_addenda_package(id)
-        return id.base_identifier.to_s unless id.base_identifier
+        return id.base.to_s unless id.base
 
-        result = "ASHRAE #{id.base_identifier.type || 'Standard'} #{id.base_identifier.code}"
-        result += "-#{id.base_identifier.year}" if id.base_identifier.year
+        result = "ASHRAE #{id.base.type || 'Standard'} #{id.base.code}"
+        result += "-#{id.base.year}" if id.base.year
         result += ": Addenda #{id.package_description}" if id.package_description
         result
       end
 
       # CombinedAddenda: "ASHRAE Addenda c and d to Standard 15-1994"
       def render_combined_addenda(id)
-        return id.base_identifier.to_s unless id.base_identifier
+        return id.base.to_s unless id.base
 
-        base_type = id.base_identifier.type || "Standard"
+        base_type = id.base.type || "Standard"
         if id.addendum_codes
-          result = "ASHRAE Addenda #{id.addendum_codes} to #{base_type} #{id.base_identifier.code}"
+          result = "ASHRAE Addenda #{id.addendum_codes} to #{base_type} #{id.base.code}"
         else
-          result = "ASHRAE Addenda to #{base_type} #{id.base_identifier.code}"
+          result = "ASHRAE Addenda to #{base_type} #{id.base.code}"
         end
-        result += "-#{id.base_identifier.year}" if id.base_identifier.year
+        result += "-#{id.base.year}" if id.base.year
         result
       end
 
       # Errata: "ASHRAE Guideline 0-2005 Errata (September 28, 2011)"
       def render_errata(id)
-        return id.base_identifier.to_s unless id.base_identifier
+        return id.base.to_s unless id.base
 
-        result = id.base_identifier.to_s
+        result = id.base.to_s
         result += " Errata"
         result += " (#{id.errata_date})" if id.errata_date
         result
@@ -105,11 +105,11 @@ module Pubid
 
       # Interpretation: "Interpretations for Standard 15.2-2022"
       def render_interpretation(id)
-        return id.base_identifier.to_s unless id.base_identifier
+        return id.base.to_s unless id.base
 
-        base_type = id.base_identifier.type || "Standard"
-        result = "Interpretations for #{base_type} #{id.base_identifier.code}"
-        result += "-#{id.base_identifier.year}" if id.base_identifier.year
+        base_type = id.base.type || "Standard"
+        result = "Interpretations for #{base_type} #{id.base.code}"
+        result += "-#{id.base.year}" if id.base.year
         result
       end
     end

data/lib/pubid/ashrae/single_identifier.rb

@@ -4,7 +4,7 @@ module Pubid
   module Ashrae
     # Base class for single (non-supplement) ASHRAE identifiers
     # Includes: Guideline, Standard
-    class SingleIdentifier < Identifiers::Base
+    class SingleIdentifier < Identifier
     end
   end
 end

data/lib/pubid/ashrae/supplement_identifier.rb

@@ -6,17 +6,17 @@ module Pubid
   module Ashrae
     # Base class for supplement identifiers (addendums, errata, interpretations)
     # Supplements modify or add to a base document
-    class SupplementIdentifier < Identifiers::Base
-      attribute :base_identifier, Identifiers::Base, polymorphic: true
+    class SupplementIdentifier < Identifier
+      attribute :base, Identifier, polymorphic: true
 
-      # Delegate publisher to base_identifier
+      # Delegate publisher to base
       def publisher
-        base_identifier&.publisher
+        base&.publisher
       end
 
-      # Delegate code to base_identifier
+      # Delegate code to base
       def code
-        base_identifier&.code
+        base&.code
       end
     end
   end

data/lib/pubid/ashrae.rb

@@ -9,7 +9,7 @@ module Pubid
     PREFIXES = ["ASHRAE"].freeze
 
     autoload :Builder, "#{__dir__}/ashrae/builder"
-    autoload :Identifier, "#{__dir__}/ashrae/identifier"
+    autoload :Identifier, "#{__dir__}/ashrae/identifiers/base"
     autoload :Identifiers, "#{__dir__}/ashrae/identifiers"
     autoload :Parser, "#{__dir__}/ashrae/parser"
     autoload :Renderer, "#{__dir__}/ashrae/renderer"
@@ -68,5 +68,5 @@ end
 Pubid::Registry.register(:ashrae, Pubid::Ashrae)
 
 # Per-flavor format registry: inherits global formats, overrides :human
-Pubid::Ashrae::Identifiers::Base.format_registry = Pubid::FormatRegistry.new(parent: Pubid::Identifier.format_registry)
-Pubid::Ashrae::Identifiers::Base.format_registry.register(:human, renderer: Pubid::Ashrae::Renderer)
+Pubid::Ashrae::Identifier.format_registry = Pubid::FormatRegistry.new(parent: Pubid::Identifier.format_registry)
+Pubid::Ashrae::Identifier.format_registry.register(:human, renderer: Pubid::Ashrae::Renderer)

data/lib/pubid/bipm/builder.rb

@@ -0,0 +1,107 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    # Turns the parse tree into the matching concrete identifier. Each root
+    # branch tags its subtree with a distinct key, so dispatch is a lookup on
+    # which key the parser produced.
+    class Builder
+      def self.build(parsed_data)
+        new.build(parsed_data)
+      end
+
+      def build(data)
+        if data[:committee_short]
+          build_committee(data[:committee_short], form: "short")
+        elsif data[:committee_long_en]
+          build_committee(data[:committee_long_en], form: "long", language: "E")
+        elsif data[:committee_bare]
+          # Bare MRA form: no type word, so build_committee yields nil type.
+          build_committee(data[:committee_bare], form: "short")
+        elsif data[:committee_long_fr]
+          build_committee(data[:committee_long_fr], form: "long", language: "F")
+        elsif data[:meeting_en]
+          build_meeting(data[:meeting_en])
+        elsif data[:meeting_fr]
+          build_meeting(data[:meeting_fr], language: "F")
+        elsif data.key?(:metrologia)
+          build_metrologia(data[:metrologia])
+        elsif data[:si_brochure]
+          build_si_brochure(data[:si_brochure])
+        elsif data[:si_brochure_variant]
+          build_si_brochure_variant(data[:si_brochure_variant])
+        elsif data[:mep]
+          build_mep(data[:mep])
+        elsif data[:guide]
+          build_guide(data[:guide])
+        else
+          raise "Unrecognized BIPM parse tree: #{data.inspect}"
+        end
+      end
+
+      private
+
+      def build_committee(node, form:, language: nil)
+        Identifiers::CommitteeDocument.new(
+          group: node[:group].to_s,
+          type_code: Identifier::TYPE_WORD_TO_CODE[node[:type_word].to_s],
+          number: node[:number]&.to_s,
+          year: node[:year]&.to_s&.to_i,
+          language: language || node[:language]&.to_s,
+          form: form,
+        )
+      end
+
+      def build_meeting(node, language: nil)
+        Identifiers::Meeting.new(
+          group: node[:group].to_s,
+          number: node[:number].to_s,
+          year: node[:year]&.to_s&.to_i,
+          language: language || node[:language]&.to_s,
+        )
+      end
+
+      def build_metrologia(node)
+        node = {} unless node.is_a?(Hash)
+        Identifiers::MetrologiaArticle.new(
+          volume: node[:volume]&.to_s&.to_i,
+          issue: node[:issue]&.to_s,
+          article: node[:article]&.to_s,
+        )
+      end
+
+      def build_si_brochure(node)
+        Identifiers::SiBrochure.new(
+          edition: node[:edition].to_s,
+          version: node[:version].to_s,
+          years: node[:years].to_s,
+          language: node[:language]&.to_s,
+        )
+      end
+
+      def build_si_brochure_variant(node)
+        Identifiers::SiBrochure.new(variant: node[:variant].to_s)
+      end
+
+      def build_mep(node)
+        Identifiers::Mep.new(
+          mep_code: node[:mep_code]&.to_s,
+          report_code: node[:report_code]&.to_s,
+          appendix: node[:appendix]&.to_s,
+          annex: node[:annex]&.to_s,
+          part: node[:part]&.to_s,
+        )
+      end
+
+      def build_guide(node)
+        Identifiers::Guide.new(
+          group: node[:group].to_s,
+          guide_kind: node[:guide_kind].to_s,
+          number: node[:number].to_s,
+          appendix: node[:appendix]&.to_s,
+          part: node[:part]&.to_s,
+        )
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifier.rb

@@ -0,0 +1,160 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    # Base class for every BIPM identifier AND the flavor's parse/create entry
+    # point — mirrors Pubid::Jis::Identifier. Concrete identifiers under
+    # Pubid::Bipm::Identifiers descend from this class.
+    #
+    # BIPM has several unrelated document families (committee documents,
+    # meetings, the Metrologia journal, the SI Brochure and its appendices,
+    # mises en pratique, and Consultative-Committee guides) with no shared
+    # numbering, so every family-specific attribute lives here as a flat,
+    # nil-defaulted attribute and is used only by the classes that need it. The
+    # canonical `to_hash` drops any attribute at its default/empty value, so an
+    # unused attribute never appears in a family's serialized hash.
+    class Identifier < ::Pubid::Identifier
+      # Committee acronyms BIPM owns (JCGM excluded — see Pubid::Jcgm).
+      GROUPS = %w[
+        CIPM CGPM JCRB CCTF CCQM CCT CCL CCAUV CCU CCM CCEM CCPR CCRI
+      ].freeze
+
+      # Committee document type codes and their printed long-form names. Note
+      # DECL prints as "Statement" in BOTH languages in the source data.
+      TYPE_CODES = %w[REC RES DECN ACT DECL].freeze
+      TYPE_NAME_EN = {
+        "REC" => "Recommendation", "RES" => "Resolution",
+        "DECN" => "Decision", "ACT" => "Action", "DECL" => "Statement"
+      }.freeze
+      TYPE_NAME_FR = {
+        "REC" => "Recommandation", "RES" => "Résolution",
+        "DECN" => "Décision", "ACT" => "Action", "DECL" => "Statement"
+      }.freeze
+      # Every printed type word (abbrev + English + French) → canonical code.
+      TYPE_WORD_TO_CODE = TYPE_CODES.each_with_object({}) do |code, map|
+        map[code] = code
+        map[TYPE_NAME_EN[code]] = code
+        map[TYPE_NAME_FR[code]] = code
+      end.freeze
+
+      # Shared attributes (flat). Booleans/strings carry no default so they stay
+      # nil and drop from the serialized hash unless set.
+      attribute :group, :string
+      attribute :type_code, :string # REC/RES/DECN/ACT/DECL (avoids base `type`)
+      attribute :number, :string    # may be hyphenated ("10-1") or nil
+      attribute :year, :integer
+      attribute :language, :string  # "E"/"F"; nil = language-neutral
+      # Surface style of committee documents: "short" (abbreviated key form) vs
+      # "long" (full type-word name). Defaults to "short" so it drops from the
+      # canonical hash for the indexed primary form.
+      attribute :form, :string, default: "short"
+      # Metrologia journal fields.
+      attribute :volume, :integer
+      attribute :issue, :string # may be alphanumeric ("1A")
+      attribute :article, :string
+      # SI Brochure fields.
+      attribute :edition, :string   # "9e"
+      attribute :version, :string   # "v3.01"
+      attribute :years, :string     # "2019/2024"
+      # SI Brochure appendix/derived variant: "Appendix 3", "Concise", "FAQ".
+      attribute :variant, :string
+      # Mise en pratique (MEP) fields.
+      attribute :mep_code, :string    # "S1", "KUPRTM"
+      attribute :report_code, :string # MEP report variant: "BIPM-2019/05"
+      # Consultative-Committee guide kind ("MeP"/"RSI"); group + number reused.
+      attribute :guide_kind, :string
+      # Full-content "Appendix N [Annex N] Part N.M" tail shared by MEP + guide
+      # primary-docidentifier forms (nil for the bare short docnumber spelling).
+      attribute :appendix, :string
+      attribute :annex, :string
+      attribute :part, :string
+
+      # Polymorphic type map for lutaml::Model key_value (de)serialization:
+      # maps each subclass's polymorphic_name to its class name so a stored hash
+      # rebuilds the correct identifier type via from_hash.
+      BIPM_TYPE_MAP = {
+        "pubid:bipm:committee-document" =>
+          "Pubid::Bipm::Identifiers::CommitteeDocument",
+        "pubid:bipm:meeting" => "Pubid::Bipm::Identifiers::Meeting",
+        "pubid:bipm:metrologia-article" =>
+          "Pubid::Bipm::Identifiers::MetrologiaArticle",
+        "pubid:bipm:si-brochure" => "Pubid::Bipm::Identifiers::SiBrochure",
+        "pubid:bipm:mep" => "Pubid::Bipm::Identifiers::Mep",
+        "pubid:bipm:guide" => "Pubid::Bipm::Identifiers::Guide",
+      }.freeze
+
+      key_value do
+        map "_type", to: :_type, polymorphic_map: BIPM_TYPE_MAP
+        map "group", to: :group
+        map "type_code", to: :type_code
+        map "number", to: :number
+        map "year", to: :year
+        map "language", to: :language
+        map "form", to: :form
+        map "volume", to: :volume
+        map "issue", to: :issue
+        map "article", to: :article
+        map "edition", to: :edition
+        map "version", to: :version
+        map "years", to: :years
+        map "variant", to: :variant
+        map "mep_code", to: :mep_code
+        map "report_code", to: :report_code
+        map "guide_kind", to: :guide_kind
+        map "appendix", to: :appendix
+        map "annex", to: :annex
+        map "part", to: :part
+      end
+
+      # Publisher is always "BIPM". A plain constant (not a `publisher` method)
+      # so it doesn't shadow the inherited lutaml `publisher` attribute, which
+      # would otherwise fail serialization type validation.
+      PUBLISHER = "BIPM"
+
+      # French connective before a committee acronym: "de la" for the (feminine)
+      # CGPM, "du" for every other (masculine) committee.
+      def self.french_connective(group)
+        group == "CGPM" ? "de la" : "du"
+      end
+
+      def long?
+        form == "long"
+      end
+
+      # BIPM keeps the publication year in its own `year` integer attribute
+      # rather than the base `date` component, so the base #exclude's
+      # :year->:date remap would nil the (unused) inherited `date` and leave
+      # `year` intact. Clear `year` directly when either alias is excluded so
+      # `matches?(row, ignore: [:year])` treats a partial (date-less) reference
+      # as a year wildcard.
+      def exclude(*args)
+        result = super
+        result.year = nil if args.include?(:year) || args.include?(:date)
+        result
+      end
+
+      # Basic string representation. Delegates to renderer.
+      def to_s(**opts)
+        render(format: :human, **opts)
+      end
+
+      # Parse a BIPM identifier string into an identifier object.
+      # @param identifier [String] the BIPM identifier string
+      # @return [Identifier] the appropriate identifier object
+      # @raise [RuntimeError] if parsing fails
+      def self.parse(identifier)
+        if identifier.length > Pubid::MAX_INPUT_LENGTH
+          raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
+        end
+
+        # Normalize legacy/docnumber-style spellings before parsing
+        # (data/bipm/update_codes.yaml), mirroring Pubid::Ccsds / Pubid::Plateau.
+        normalized = Core::UpdateCodes.apply(identifier, :bipm)
+        parsed = Parser.parse(normalized)
+        Builder.build(parsed)
+      rescue Parslet::ParseFailed => e
+        raise "Failed to parse BIPM identifier '#{identifier}': #{e.message}"
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/committee_document.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # A committee document: a Recommendation (REC), Resolution (RES),
+      # Decision (DECN), Action (ACT) or Declaration (DECL) issued by one of the
+      # BIPM committees.
+      #
+      # Printed forms (all round-trip):
+      #   "CCTF REC 2 (2012)"              short, language-neutral
+      #   "CCTF REC 2 (2012, E)"           short, with language
+      #   "CGPM DECL (1889)"               number-less
+      #   "CCL Recommendation 1 (2001)"    full English name  (form: "long")
+      #   "Recommandation 1 du CCL (2001)" full French name   (form: "long")
+      class CommitteeDocument < Identifier
+        def self.type
+          { key: :committee_document, web: :committee_document,
+            title: "Committee Document", short: "committee-document" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/guide.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # A Consultative-Committee guideline document (realization / calibration
+      # guide). Printed from its short docnumber form, the relaton index key.
+      #
+      # Printed forms (all round-trip):
+      #   "CCL-GD-MeP-1"    metre realization guide
+      #   "CCEM-GD-RSI-1"   ampere realization guide
+      #   "CCM-GD-RSI-2"    kilogram dissemination guide
+      #
+      # Shape: <committee>-GD-<kind>-<number>, where <kind> is "MeP" (mise en
+      # pratique guide) or "RSI" (réalisation du SI guide). The committee reuses
+      # the shared `group` attribute; the trailing sequence reuses `number`.
+      class Guide < Identifier
+        def self.type
+          { key: :guide, web: :guide, title: "Guide", short: "guide" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/meeting.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # A committee meeting (proceedings). Printed with a naive ordinal derived
+      # from the first number component's last digit (as the source data
+      # generates it: "11st", "12nd", "13rd"), and ranges keep that first
+      # component's ordinal ("100-1th").
+      #
+      # Printed forms (all round-trip):
+      #   "CGPM 17th Meeting (1983)"             English      (language nil)
+      #   "CCAUV 11st Meeting (2017)"            naive ordinal
+      #   "CIPM 100-1th Meeting (2011)"          range
+      #   "CCAUV 10<sup>e</sup> réunion (2015)"  French       (language "F")
+      class Meeting < Identifier
+        def self.type
+          { key: :meeting, web: :meeting, title: "Meeting", short: "meeting" }
+        end
+
+        # English ordinal suffix from the first number component's last digit.
+        # Naive (matches the source data): 1→st, 2→nd, 3→rd, else→th — no
+        # 11/12/13 exception, so "11" → "11st", and a range "22-23" → "22-23nd".
+        def self.ordinal_suffix(number)
+          first = number.to_s.split("-").first.to_s
+          case first[-1]
+          when "1" then "st"
+          when "2" then "nd"
+          when "3" then "rd"
+          else "th"
+          end
+        end
+
+        # "15" → "15th", "100-1" → "100-1th".
+        def ordinal_en
+          "#{number}#{self.class.ordinal_suffix(number)}"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/mep.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # A mise en pratique (MEP) of an SI base-unit definition, and the related
+      # BIPM report variant. Printed from the short docnumber form that relaton
+      # uses as the index key — NOT the long "Appendix 2 Part x" content string.
+      #
+      # Printed forms (all round-trip):
+      #   "SI MEP S1"              standard MEP code (unit letter + number)
+      #   "SI MEP KUPRTM"          alphabetic MEP code
+      #   "Rapport BIPM-2019/05"   report variant (mep_code absent, report_code set)
+      class Mep < Identifier
+        def self.type
+          { key: :mep, web: :mep, title: "Mise en pratique", short: "mep" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/metrologia_article.rb

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # A Metrologia journal reference, at volume, volume+issue, or
+      # volume+issue+article granularity. Issue may be alphanumeric ("1A") and
+      # the article keeps its printed form ("06007", "S138").
+      #
+      # Printed forms (all round-trip):
+      #   "Metrologia 51"          volume only
+      #   "Metrologia 1 1"         volume + issue
+      #   "Metrologia 51 1 128"    volume + issue + article
+      class MetrologiaArticle < Identifier
+        GROUP = "Metrologia"
+
+        def self.type
+          { key: :metrologia_article, web: :metrologia_article,
+            title: "Metrologia Article", short: "metrologia-article" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers/si_brochure.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    module Identifiers
+      # The SI Brochure — a single bespoke record. The English form drops the
+      # "sur le SI" phrase the French form carries.
+      #
+      # Printed forms (both round-trip):
+      #   "BIPM SI Brochure 9e v3.01 (2019/2024, E)"
+      #   "BIPM SI Brochure sur le SI 9e v3.01 (2019/2024, F)"
+      class SiBrochure < Identifier
+        def self.type
+          { key: :si_brochure, web: :si_brochure,
+            title: "SI Brochure", short: "si-brochure" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/bipm/identifiers.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Bipm
+    # Namespace for concrete BIPM identifier types.
+    module Identifiers
+      autoload :CommitteeDocument, "#{__dir__}/identifiers/committee_document"
+      autoload :Guide, "#{__dir__}/identifiers/guide"
+      autoload :Meeting, "#{__dir__}/identifiers/meeting"
+      autoload :Mep, "#{__dir__}/identifiers/mep"
+      autoload :MetrologiaArticle, "#{__dir__}/identifiers/metrologia_article"
+      autoload :SiBrochure, "#{__dir__}/identifiers/si_brochure"
+    end
+  end
+end