pubid 2.0.0.pre.alpha.9 → 2.0.0.pre.alpha.12

data/lib/pubid/bipm/urn_parser.rb

@@ -35,18 +35,26 @@ module Pubid
         end
       end
 
+      # `number` is the relaton index key and is derived, not printed, so it has
+      # to be set here exactly as `Builder` sets it — a URN-reconstructed
+      # identifier must agree with a parsed one attribute for attribute.
       def build_metrologia(parts)
         Identifiers::MetrologiaArticle.new(
-          volume: parts[1]&.to_i, issue: parts[2], article: parts[3],
+          number: presence(parts[1]), issue: parts[2], article: parts[3],
         )
       end
 
       def build_si_brochure(parts)
         Identifiers::SiBrochure.new(
-          language: parts[1].to_s.upcase, edition: parts[2],
-          version: parts[3], years: parts[4]
+          number: presence(parts[2]), language: parts[1].to_s.upcase,
+          edition: parts[2], version: parts[3], years: parts[4]
         )
       end
+
+      # A URN segment that was never filled in is an empty string, not nil.
+      def presence(segment)
+        segment unless segment.nil? || segment.empty?
+      end
     end
   end
 end

data/lib/pubid/bipm.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+require "pubid"
 module Pubid
   module Bipm
     extend Pubid::PrefixesSupport
@@ -27,8 +28,13 @@ module Pubid
     # @param identifier [String] the printed identifier
     # @return [Identifier] the parsed identifier
     def self.parse(identifier)
+      unless identifier.is_a?(String)
+        raise Pubid::Errors::InvalidInputError,
+              Pubid::INPUT_NOT_A_STRING_MESSAGE
+      end
+
       if identifier.length > Pubid::MAX_INPUT_LENGTH
-        raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
+        raise Pubid::Errors::InvalidInputError, Pubid::INPUT_TOO_LONG_MESSAGE
       end
 
       Identifier.parse(identifier)

data/lib/pubid/bsi/builder.rb

@@ -164,7 +164,7 @@ module Pubid
 
         # Build attributes hash (conditional arguments must be handled separately)
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
           issue_number: issue_number,
           index_format: format,
         }
@@ -181,7 +181,7 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
         }
         attrs[:date] = Components::Date.new(year: year_val) if year_val
 
@@ -206,9 +206,9 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
         }
-        attrs[:part] = Components::Code.new(value: part_val) if part_val
+        attrs[:part] = part_val if part_val
         attrs[:date] = Components::Date.new(year: year_val) if year_val
 
         Identifiers::ExplanatorySupplement.new(attrs)
@@ -248,13 +248,13 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
           method_code: method_code,
           method_to: method_to,
           method_and: method_and,
           is_plural: is_plural,
         }
-        attrs[:part] = Components::Code.new(value: part_val) if part_val
+        attrs[:part] = part_val if part_val
         attrs[:date] = Components::Date.new(year: year_val) if year_val
 
         Identifiers::Method.new(attrs)
@@ -280,7 +280,7 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
           test_series: test_series,
           test_id: test_id,
         }
@@ -313,7 +313,7 @@ module Pubid
 
         # Build attributes hash (conditional arguments must be handled separately)
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
           section_id: section_id,
           section_format: format,
         }
@@ -343,7 +343,7 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
         }
         attrs[:spec_code] = Components::Code.new(value: spec_code) if spec_code
         if publisher_val
@@ -373,9 +373,9 @@ module Pubid
 
         # Build attributes hash
         attrs = {
-          number: Components::Code.new(value: number_val),
+          number: number_val,
         }
-        attrs[:part] = Components::Code.new(value: part_val) if part_val
+        attrs[:part] = part_val if part_val
         attrs[:publisher] = Components::Publisher.new(body: "DISC")
         attrs[:date] = Components::Date.new(year: year_val) if year_val
 
@@ -432,9 +432,9 @@ module Pubid
         # Build attributes hash
         attrs = {
           prefix: prefix_val,
-          number: Components::Code.new(value: number_val),
+          number: number_val,
         }
-        attrs[:part] = Components::Code.new(value: part_val) if part_val
+        attrs[:part] = part_val if part_val
         attrs[:iteration] = iteration_val if iteration_val
         attrs[:edition] = final_edition if final_edition
         attrs[:publisher] = Components::Publisher.new(body: "BS")
@@ -456,20 +456,20 @@ module Pubid
           # Build base identifier
           base_id = SingleIdentifier.new(
             publisher: Components::Publisher.new(body: "BS"),
-            number: Components::Code.new(value: base_number),
+            number: base_number,
           )
 
           # Build identifiers for each part
           id1 = SingleIdentifier.new(
             publisher: Components::Publisher.new(body: "BS"),
-            number: Components::Code.new(value: base_number),
-            part: Components::Code.new(value: part1),
+            number: base_number,
+            part: part1,
           )
 
           id2 = SingleIdentifier.new(
             publisher: Components::Publisher.new(body: "BS"),
-            number: Components::Code.new(value: base_number),
-            part: Components::Code.new(value: part2),
+            number: base_number,
+            part: part2,
           )
 
           Identifiers::BundledIdentifier.new(
@@ -636,7 +636,7 @@ module Pubid
           publisher: Components::Publisher.new(body: publisher_val),
         )
         id.prefix = prefix_val if prefix_val && !prefix_val.empty?
-        id.number = Components::Code.new(value: number_val) if number_val
+        id.number = number_val if number_val
 
         id.explicit_prefix = has_explicit_publisher || has_explicit_prefix
         id.explicit_publisher = has_explicit_publisher
@@ -646,11 +646,11 @@ module Pubid
           parts_array = parts_val[:parts]
           if parts_array.any?
             part_str = parts_array.first[:part].to_s
-            id.part = Components::Code.new(value: part_str)
+            id.part = part_str
           end
         # Handle space-separated part (direct part attribute)
         elsif space_separated_part_val
-          id.part = Components::Code.new(value: space_separated_part_val.to_s)
+          id.part = space_separated_part_val.to_s
           # Mark this part as space-separated for rendering
           id.space_separated_part = true
         end
@@ -790,8 +790,10 @@ module Pubid
         corrigendum = !amd_data[:corrigendum].nil?
         parenthesized = !data[:parenthesized_amd].nil?
 
+        # NOTE `amd_data[:amendment_number]` above is a PARSE-TREE key and keeps
+        # its name; only the attribute moved to `number`.
         attrs = {
-          amendment_number: Components::Code.new(value: amendment_number),
+          number: Components::Code.new(value: amendment_number),
           corrigendum: corrigendum,
           parenthesized: parenthesized,
         }
@@ -809,8 +811,10 @@ module Pubid
         # Could also use a more sophisticated algorithm for year conversion
         full_year = year_val ? "20#{year_val}" : nil
 
+        # NOTE `data[:document_number]` above is a PARSE-TREE key and keeps its
+        # name; only the attribute moved to `number`.
         attrs = {
-          document_number: document_number,
+          number: Components::Code.new(value: document_number),
         }
         attrs[:date] = Components::Date.new(year: full_year.to_i) if full_year
 
@@ -969,7 +973,7 @@ module Pubid
           value&.to_s
 
         when :number
-          Components::Code.new(value: value.to_s)
+          value.to_s
 
         when :parts
           # Extract parts - handle both formats:
@@ -983,26 +987,25 @@ module Pubid
             # Check if subpart is already separated (new format from part_with_subpart rule)
             if first_part.key?(:subpart)
               {
-                part: Components::Code.new(value: part_str),
-                subpart: Components::Code.new(value: first_part[:subpart].to_s),
+                part: part_str,
+                subpart: first_part[:subpart].to_s,
               }
             else
               # Old format - split on dash to get part and subpart
               part_components = part_str.split("-")
-              result = { part: Components::Code.new(value: part_components.first) }
+              result = { part: part_components.first }
               if part_components.length > 1
-                result[:subpart] =
-                  Components::Code.new(value: part_components[1])
+                result[:subpart] = part_components[1]
               end
               result
             end
           end
 
         when :part
-          Components::Code.new(value: value[:part].to_s) if value.is_a?(Hash)
+          value[:part].to_s if value.is_a?(Hash)
 
         when :subpart
-          Components::Code.new(value: value.to_s)
+          value.to_s
 
         when :year, :date
           # Only create date if value is present
@@ -1323,8 +1326,15 @@ module Pubid
 
         adopted_id = nil
 
-        # Check for EN ISO or EN IEC patterns (triple-level)
-        if adopted_str_clean.match?(/EN\s+(ISO\/IEC|IEC|ISO)/)
+        # Check for EN ISO or EN IEC patterns (triple-level).
+        # The test must be ANCHORED, because the repair below is: an unanchored
+        # match accepts "CEN ISO/TS 12180-1:2007" (the "EN ISO" is inside
+        # "CEN ISO"), but `sub(/^EN\s+/, "")` then changes nothing, so
+        # adopted_id stayed nil and the whole parse returned nil for 49
+        # "DD CEN ISO/..." identifiers in BSI's own pass fixtures. Anchored,
+        # a "CEN ..." string falls through to the CEN branch below, which is
+        # where it belongs.
+        if adopted_str_clean.match?(/\AEN\s+(ISO\/IEC|IEC|ISO)/)
           # Parse the ISO/IEC part
           iso_iec_str = adopted_str_clean.sub(/^EN\s+/, "")
 
@@ -1336,11 +1346,11 @@ module Pubid
             adopted_id = Pubid::Iec.parse(iso_iec_str)
           end
 
-          # Wrap ISO/IEC identifier in EN adoption
+          # Wrap ISO/IEC identifier in EN adoption (the publisher is the
+          # "EN" default)
           if adopted_id
             adopted_id = Pubid::CenCenelec::Identifiers::AdoptedEuropeanNorm.new(
-              publisher: ["EN"],
-              adopted_identifier: adopted_id,
+              adopted: adopted_id,
             )
           end
 
@@ -1373,7 +1383,7 @@ module Pubid
           identifier = if adopted_id.class.name.start_with?("Pubid::CenCenelec::")
                          Identifiers::AdoptedEuropeanNorm.new(
                            publisher: Components::Publisher.new(body: bsi_prefix),
-                           adopted_identifier: adopted_id,
+                           adopted: adopted_id,
                            edition: final_edition&.to_s,
                            translation_lang: data[:translation_lang]&.to_s,
                            translation_upper: data[:translation_upper]&.to_s,
@@ -1386,7 +1396,7 @@ module Pubid
                          # Otherwise it's ISO/IEC, use AdoptedInternationalStandard
                          Identifiers::AdoptedInternationalStandard.new(
                            publisher: Components::Publisher.new(body: bsi_prefix),
-                           adopted_identifier: adopted_id,
+                           adopted: adopted_id,
                            edition: final_edition&.to_s,
                            translation_lang: data[:translation_lang]&.to_s,
                            translation_upper: data[:translation_upper]&.to_s,

data/lib/pubid/bsi/components/date.rb

@@ -1,11 +1,17 @@
-# frozen_string_literal: true
+# frozen_string: true
 
 module Pubid
   module Bsi
     module Components
-      # Date component for BSI flavor
-      class Date < Pubid::Components::Date
-      end
+      # BSI has no date semantics of its own, so this is the shared
+      # component, not a subclass: an empty `class Date <
+      # Pubid::Components::Date` rejected the shared instances a nested
+      # cross-flavor adopted identifier carries (pubid#379 — the adoption
+      # wrappers delegate `date` to the foreign object), and made `==`
+      # compare the parsed instance against the deserialized one by class.
+      # Aliased so the builder's `Components::Date.new` call sites construct
+      # the shared class.
+      Date = ::Pubid::Components::Date
     end
   end
 end

data/lib/pubid/bsi/identifiers/adopted_european_norm.rb

@@ -7,7 +7,7 @@ module Pubid
       # Example: "BS EN 10077-1:2006" where EN 10077-1:2006 is a CEN identifier object
       # Example: "BS EN ISO 8601:2019" where EN ISO 8601:2019 is a CEN AdoptedEuropeanNorm wrapping ISO
       class AdoptedEuropeanNorm < BritishStandard
-        attribute :adopted_identifier, ::Pubid::Identifier, polymorphic: true # CEN object
+        attribute :adopted, ::Pubid::Identifier, polymorphic: true # CEN object
         attribute :edition, :string
         attribute :translation_lang, :string
         attribute :translation_upper, :string
@@ -25,29 +25,58 @@ module Pubid
           nil
         end
 
+        # Walk to the adopted document for the relaton-index key.
+        #
+        # The `#number` delegation below is not enough on its own: a
+        # "DD ENV ISO 11079:1999" adopts a CenCenelec EuropeanPrestandard,
+        # which is ITSELF a wrapper around the ISO standard, so the delegation
+        # returned that wrapper's own (nil) number and the chain died one level
+        # short. `#root` recurses, so it reaches the ISO standard however many
+        # adoption layers sit in between.
+        def root
+          adopted ? adopted.root : self
+        end
+
         # Delegate common methods to adopted identifier
         def number
-          adopted_identifier&.number
+          delegate_target&.number
         end
 
         def year
-          adopted_identifier&.year if adopted_identifier&.methods&.include?(:year)
+          delegate_target&.year if delegate_target&.methods&.include?(:year)
         end
 
         def date
-          adopted_identifier&.date if adopted_identifier&.methods&.include?(:date)
+          delegate_target&.date if delegate_target&.methods&.include?(:date)
         end
 
         def parts
-          adopted_identifier&.parts if adopted_identifier&.methods&.include?(:parts)
+          delegate_target&.parts if delegate_target&.methods&.include?(:parts)
         end
 
         def part
-          adopted_identifier&.part if adopted_identifier&.methods&.include?(:part)
+          delegate_target&.part if delegate_target&.methods&.include?(:part)
         end
 
         def subpart
-          adopted_identifier&.subpart if adopted_identifier&.methods&.include?(:subpart)
+          adopted&.subpart if adopted&.methods&.include?(:subpart)
+        end
+
+        private
+
+        # The identifier that holds the number, part and date. For
+        # "BS EN ISO 8848:2021" the adopted document is a CEN adoption of an
+        # ISO standard, and the CEN adoption keeps no number of its own, so
+        # read through it to the ISO standard. The CEN class used to do this
+        # with delegating readers, which broke its serialization. `subpart`
+        # keeps reading the CEN object, as it did before.
+        def delegate_target
+          target = adopted
+          if target.is_a?(::Pubid::CenCenelec::Identifiers::AdoptedEuropeanNorm)
+            target.adopted
+          else
+            target
+          end
         end
       end
     end

data/lib/pubid/bsi/identifiers/adopted_international_standard.rb

@@ -7,7 +7,8 @@ module Pubid
       # Example: "BS ISO 8601:2019" where ISO 8601:2019 is an ISO identifier object
       # Example: "BS IEC 62600:2020" where IEC 62600:2020 is an IEC identifier object
       class AdoptedInternationalStandard < BritishStandard
-        attribute :adopted_identifier, ::Pubid::Identifier, polymorphic: true # ISO/IEC object
+        # The adopted ISO/IEC identifier object
+        attribute :adopted, ::Pubid::Identifier, polymorphic: true
         attribute :edition, :string
         attribute :translation_lang, :string
         attribute :translation_upper, :string
@@ -27,23 +28,23 @@ module Pubid
 
         # Delegate common methods to adopted identifier
         def number
-          adopted_identifier&.number
+          adopted&.number
         end
 
         def year
-          adopted_identifier&.year if adopted_identifier&.methods&.include?(:year)
+          adopted&.year if adopted&.methods&.include?(:year)
         end
 
         def date
-          adopted_identifier&.date if adopted_identifier&.methods&.include?(:date)
+          adopted&.date if adopted&.methods&.include?(:date)
         end
 
         def parts
-          adopted_identifier&.parts if adopted_identifier&.methods&.include?(:parts)
+          adopted&.parts if adopted&.methods&.include?(:parts)
         end
 
         def part
-          adopted_identifier&.part if adopted_identifier&.methods&.include?(:part)
+          adopted&.part if adopted&.methods&.include?(:part)
         end
       end
     end

data/lib/pubid/bsi/identifiers/british_industrial_practice.rb

@@ -6,8 +6,7 @@ module Pubid
       # British Industrial Practice (BIP)
       # Examples: "BIP 2225:2022", "BIP 0142:2014", "BIP 0009:2020"
       class BritishIndustrialPractice < SingleIdentifier
-        attribute :number, Bsi::Components::Code
-        attribute :date, Bsi::Components::Date
+
 
         def self.type
           {

data/lib/pubid/bsi/identifiers/bundled_identifier.rb

@@ -24,6 +24,16 @@ module Pubid
           # Compare first identifier
           identifiers.first <=> other.identifiers.first
         end
+
+        # A bundle carries no number of its own — each member holds one — so
+        # `root.number`, the key relaton-index sorts and bsearches on, was "".
+        # Walk to the first member, the shape ConsolidatedIdentifier already
+        # uses. Members of one bundle are consecutive numbers of a single
+        # series, so the bundle files under the first, and `<=>` above orders
+        # bundles by that same member.
+        def root
+          identifiers&.first ? identifiers.first.root : self
+        end
       end
     end
   end

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

@@ -11,7 +11,16 @@ module Pubid
       #   21/30445138 DC
       #   24/30300822 DC
       class CommitteeDocument < SingleIdentifier
-        attribute :document_number, :string # The 8-digit document number
+        # The 8-digit document number lives in the `number` inherited from
+        # SingleIdentifier (a Bsi::Components::Code). It used to be a separate
+        # `document_number` :string, which left `number` nil — so `root.number`,
+        # the key relaton-index sorts and bsearches on, was "".
+        #
+        # The duplicate is DELETED rather than mirrored into `number`: keeping
+        # both would emit the same digits twice in `to_hash`, and a derived
+        # `#number` method is not available here because `number` is a lutaml
+        # attribute, whose generated accessor such a method would collide with.
+        # Nothing is retyped — the inherited attribute is already a Code.
 
         # TYPED_STAGES for committee documents (draft by default)
         TYPED_STAGES = [

data/lib/pubid/bsi/identifiers/handbook.rb

@@ -6,9 +6,7 @@ module Pubid
       # BSI Handbook
       # Examples: "Handbook 17:1963", "HB 10146:1998"
       class Handbook < SingleIdentifier
-        attribute :number, Bsi::Components::Code
-        attribute :part, Bsi::Components::Code
-        attribute :date, Bsi::Components::Date
+
         attribute :original_abbr, :string # Preserve "Handbook" or "HB"
 
         def self.type

data/lib/pubid/bsi/identifiers/practice_guide.rb

@@ -6,8 +6,7 @@ module Pubid
       # BSI Practice Guide (PP - Published Practice)
       # Examples: "PP 888:1982", "PP 7307:1986", "PP 7722:2006"
       class PracticeGuide < SingleIdentifier
-        attribute :number, Bsi::Components::Code
-        attribute :date, Bsi::Components::Date
+
 
         def self.type
           {

data/lib/pubid/bsi/identifiers/set.rb

@@ -23,6 +23,17 @@ module Pubid
           # Compare first identifier
           identifiers.first <=> other.identifiers.first
         end
+
+        # As for BundledIdentifier: a set holds no number of its own, so walk
+        # to the first member for the relaton-index key.
+        #
+        # NOTE the members are frequently a FOREIGN flavor (BS ISO ... holds
+        # Pubid::Iso identifiers), which `to_hash` still rejects — see the
+        # hand-off bsi-set-cross-flavor-type. `#root` is unaffected by that:
+        # it only reads the object, never serializes it.
+        def root
+          identifiers&.first ? identifiers.first.root : self
+        end
       end
     end
   end

data/lib/pubid/bsi/identifiers/standalone_amendment.rb

@@ -6,7 +6,16 @@ module Pubid
       # Standalone Amendment identifier (no base reference)
       # Examples: "AMD 11015", "(AMD 10971)", "AMD Corrigendum 14716"
       class StandaloneAmendment < SingleIdentifier
-        attribute :amendment_number, Components::Code
+        # The amendment number lives in the `number` inherited from
+        # SingleIdentifier — already a Bsi::Components::Code, exactly what the
+        # deleted `amendment_number` was, so nothing is retyped. It used to sit
+        # under the private name, leaving `number` nil and `root.number` "".
+        #
+        # A standalone amendment is its OWN document (there is no base standard
+        # to walk to — that is what "standalone" means), so it needs a real
+        # number rather than a #root override. `Identifiers::Amendment`, the
+        # attached kind, keeps its own `amendment_number` and is untouched:
+        # there the ordinal is a supplement ordinal, not the document number.
         attribute :corrigendum, :boolean, default: false
         attribute :parenthesized, :boolean, default: false
 

data/lib/pubid/bsi/parser.rb

@@ -4,7 +4,7 @@ require "parslet"
 
 module Pubid
   module Bsi
-    class Parser < Parslet::Parser
+    class Parser < ::Pubid::Parser::Grammar
       # Case-insensitive string match (Parslet 2.0 has no built-in `stri`).
       # Each character becomes a two-case character class; other characters
       # (spaces, punctuation) match verbatim.