pubid 2.0.0.pre.alpha.8 → 2.0.0.pre.alpha.11

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/identifiers/supplement_document.rb

@@ -9,7 +9,7 @@ module Pubid
       # Forward: "BS 1000:Supplement No. 1:1972", "BS 1722-1 Supplement No. 1:1974"
       # Reverse: "Supplement No. 1 (1970) to BS 1831:1969"
       class SupplementDocument < SingleIdentifier
-        attribute :base_identifier, ::Pubid::Identifier, polymorphic: true
+        attribute :base, ::Pubid::Identifier, polymorphic: true
         attribute :supplement_number, :string
         attribute :supplement_year, :integer
         attribute :supplement_type, :string, default: -> {
@@ -21,7 +21,7 @@ module Pubid
         } # Separator before Supplement (":" or " ")
 
         def publisher
-          base_identifier&.publisher
+          base&.publisher
         end
       end
     end

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

@@ -12,24 +12,24 @@ module Pubid
       #   PAS 96:2017 - TC
       #   PP 7722:2006 BOOK
       class ValueAddedPublication < SingleIdentifier
-        attribute :base_identifier, ::Pubid::Identifier, polymorphic: true
+        attribute :base, ::Pubid::Identifier, polymorphic: true
         attribute :format, :string, values: ["PDF", "TC", "BOOK"]
 
-        # Delegate common attributes to base_identifier
+        # Delegate common attributes to base
         def publisher
-          base_identifier&.publisher
+          base&.publisher
         end
 
         def number
-          base_identifier&.number
+          base&.number
         end
 
         def year
-          base_identifier&.year
+          base&.year
         end
 
         def date
-          base_identifier&.date
+          base&.date
         end
       end
     end

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.

data/lib/pubid/bsi/renderer.rb

@@ -94,6 +94,16 @@ module Pubid
 
       private
 
+      # Read a value that may be a component or a bare scalar. BSI's own
+      # `number`/`part`/`subpart` are plain strings, but a bundle member or an
+      # adopted identifier can be a foreign one (ISO, IEC, CEN/CENELEC) whose
+      # equivalents are still a Components::Code.
+      def component_str(value)
+        return "" if value.nil?
+
+        (value.respond_to?(:render) ? value.render(context: @context) : value).to_s
+      end
+
       # ------------------------------------------------------------------
       # SingleIdentifier (base BSI identifier)
       # ------------------------------------------------------------------
@@ -109,7 +119,7 @@ module Pubid
         end
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.second_number
             second_val = id.second_number.render(context: @context)
@@ -118,13 +128,13 @@ module Pubid
 
           space_separated = id.space_separated_part
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             part_str = part_val.to_s.strip
             separator = space_separated ? " " : "-"
             number_str += "#{separator}#{part_str}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             subpart_str = subpart_val.to_s.strip
             number_str += "-#{subpart_str}"
           end
@@ -160,13 +170,13 @@ module Pubid
       # ------------------------------------------------------------------
 
       def render_addendum_document(id, opts)
-        result = id.base_identifier.to_s(**opts)
+        result = id.base.to_s(**opts)
 
-        base_has_year = id.base_identifier.to_s =~ /:(\d{4})$/
+        base_has_year = id.base.to_s =~ /:(\d{4})$/
 
         sep = if id.separator == ":"
                 ":"
-              elsif base_has_year && id.base_identifier.to_s !~ /:\d{4}:/
+              elsif base_has_year && id.base.to_s !~ /:\d{4}:/
                 " "
               else
                 id.separator
@@ -202,7 +212,7 @@ module Pubid
                  end
 
         result = prefix
-        result += " #{id.adopted_identifier}" if id.adopted_identifier
+        result += " #{id.adopted}" if id.adopted
         result += " ED#{id.edition}" if id.edition
 
         result += " (R#{id.reaffirmation_year})" if id.reaffirmation_year
@@ -227,7 +237,7 @@ module Pubid
                  end
 
         result = prefix
-        result += " #{id.adopted_identifier}" if id.adopted_identifier
+        result += " #{id.adopted}" if id.adopted
         result += " ED#{id.edition}" if id.edition
 
         result += " (R#{id.reaffirmation_year})" if id.reaffirmation_year
@@ -249,14 +259,14 @@ module Pubid
         parts << id.prefix if id.prefix
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val}"
           end
 
@@ -289,7 +299,7 @@ module Pubid
       end
 
       def render_amendment(id)
-        base = id.base_identifier ? id.base_identifier.to_s : ""
+        base = id.base ? id.base.to_s : ""
 
         if id.amd_suffix_form
           # Trailing " AMD5" / " AMD AA" suffix form
@@ -321,8 +331,10 @@ module Pubid
           parts_list = id.identifiers[1..].map do |idd|
             part_val = idd.class.attributes.key?(:part) ? idd.part : nil
             if part_val
-              pv = part_val.render(context: @context)
-              pv.to_s
+              # A bundle member may be a foreign identifier (ISO, CEN/CENELEC),
+              # whose `part` is still a Components::Code, so read through
+              # whichever shape it holds.
+              component_str(part_val)
             else
               idd.to_s
             end
@@ -350,23 +362,23 @@ module Pubid
               abbrev_str = prefix.to_s
             end
             if idd.number
-              number_val = idd.number.render(context: @context)
+              number_val = idd.number
               abbrev_str += " " if !abbrev_str.empty?
               abbrev_str += number_val.to_s
             end
             if idd.part
-              part_val = idd.part.render(context: @context)
+              part_val = idd.part
               abbrev_str += "-#{part_val}"
             end
             parts << abbrev_str
           else
             abbrev_str = ""
             if idd.number
-              number_val = idd.number.render(context: @context)
+              number_val = idd.number
               abbrev_str = number_val.to_s
             end
             if idd.part
-              part_val = idd.part.render(context: @context)
+              part_val = idd.part
               abbrev_str += "-#{part_val}"
             end
             parts << abbrev_str
@@ -394,7 +406,7 @@ module Pubid
                      "00"
                    end
 
-        "#{year_str}/#{id.document_number} DC"
+        "#{year_str}/#{id.number} DC"
       end
 
       def render_consolidated_identifier(id, opts)
@@ -471,7 +483,7 @@ module Pubid
       end
 
       def render_corrigendum(id)
-        result = id.base_identifier ? id.base_identifier.to_s : ""
+        result = id.base ? id.base.to_s : ""
         result += "#{id.separator}C"
         result += id.corrigendum_number.to_s if id.corrigendum_number
         result += ":#{id.corrigendum_year}" if id.corrigendum_year
@@ -483,7 +495,7 @@ module Pubid
         parts << "BS"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           parts << number_str
         end
 
@@ -507,9 +519,9 @@ module Pubid
         parts << "DISC"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val.to_s.strip}"
           end
           parts << "PD #{number_str}"
@@ -529,14 +541,14 @@ module Pubid
         parts << type_abbr
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val.to_s.strip}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val.to_s.strip}"
           end
 
@@ -562,14 +574,14 @@ module Pubid
         parts << "EP"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val.to_s.strip}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val.to_s.strip}"
           end
 
@@ -589,7 +601,7 @@ module Pubid
       end
 
       def render_expert_commentary(id)
-        base_str = id.base_identifier.to_s
+        base_str = id.base.to_s
         base_str = base_str.sub(/ (Expert Commentary|ExComm(\s*\(.*\))?)$/, "")
 
         case id.format
@@ -607,14 +619,14 @@ module Pubid
         parts << "BS"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val}"
           end
 
@@ -637,13 +649,13 @@ module Pubid
         parts << "BSI Flex"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val}"
           end
           parts << number_str
@@ -665,9 +677,9 @@ module Pubid
       def render_handbook(id)
         abbr = id.original_abbr || "Handbook"
 
-        number_str = id.number.render(context: @context)
+        number_str = id.number
         if id.part
-          part_val = id.part.render(context: @context)
+          part_val = id.part
           number_str += "-#{part_val.to_s.strip}"
         end
 
@@ -681,7 +693,7 @@ module Pubid
         parts << id.publisher.to_s if id.publisher
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           parts << number_str
         end
 
@@ -708,9 +720,9 @@ module Pubid
         parts << id.publisher.to_s if id.publisher
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           parts << number_str
@@ -771,7 +783,7 @@ module Pubid
         parts << type_abbr
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.second_number
             second_val = id.second_number.render(context: @context)
@@ -779,11 +791,11 @@ module Pubid
           end
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val.to_s.strip}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val.to_s.strip}"
           end
 
@@ -811,7 +823,7 @@ module Pubid
         parts << type_abbr
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.second_number
             second_val = id.second_number.render(context: @context)
@@ -819,11 +831,11 @@ module Pubid
           end
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val.to_s.strip}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val.to_s.strip}"
           end
 
@@ -850,7 +862,7 @@ module Pubid
         parts << id.publisher.to_s if id.publisher
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           parts << number_str
         end
 
@@ -886,16 +898,16 @@ module Pubid
 
       def render_standalone_amendment(id)
         base = if id.corrigendum
-                 "AMD Corrigendum #{id.amendment_number}"
+                 "AMD Corrigendum #{id.number}"
                else
-                 "AMD #{id.amendment_number}"
+                 "AMD #{id.number}"
                end
 
         id.parenthesized ? "(#{base})" : base
       end
 
       def render_supplement_document(id)
-        base_str = id.base_identifier.to_s
+        base_str = id.base.to_s
 
         if id.reverse_format
           supplement_str = if id.supplement_type == "No."
@@ -919,14 +931,14 @@ module Pubid
         parts << "BS"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val}"
           end
 
@@ -949,14 +961,14 @@ module Pubid
         parts << "TS"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
 
           if id.part
-            part_val = id.part.render(context: @context)
+            part_val = id.part
             number_str += "-#{part_val}"
           end
           if id.subpart
-            subpart_val = id.subpart.render(context: @context)
+            subpart_val = id.subpart
             number_str += "-#{subpart_val}"
           end
 
@@ -979,7 +991,7 @@ module Pubid
         parts << "BS"
 
         if id.number
-          number_str = id.number.render(context: @context)
+          number_str = id.number
           parts << number_str
         end
 
@@ -998,7 +1010,7 @@ module Pubid
       end
 
       def render_value_added_publication(id, opts)
-        base_str = id.base_identifier.to_s(**opts)
+        base_str = id.base.to_s(**opts)
 
         case id.format
         when "TC"

data/lib/pubid/bsi/single_identifier.rb

@@ -9,8 +9,13 @@ module Pubid
     # `from_hash` (no facade needed).
     class Identifier < ::Pubid::Identifier
       def self.parse(string)
+        unless string.is_a?(String)
+          raise Pubid::Errors::InvalidInputError,
+                Pubid::INPUT_NOT_A_STRING_MESSAGE
+        end
+
         if string.length > Pubid::MAX_INPUT_LENGTH
-          raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
+          raise Pubid::Errors::InvalidInputError, Pubid::INPUT_TOO_LONG_MESSAGE
         end
 
         # Delegate to IEC for bare IEC identifiers. This handles IEC-specific
@@ -27,8 +32,6 @@ module Pubid
 
         parsed = parser.parse(string)
         Builder.build(parsed)
-      rescue Parslet::ParseFailed => e
-        raise StandardError, "Failed to parse '#{string}': #{e.message}"
       end
 
       attribute :publisher, Bsi::Components::Publisher, default: -> {
@@ -36,10 +39,16 @@ module Pubid
       }
       attribute :prefix, :string # Specialized prefix (A, AU, C, M, 2A, etc.)
       attribute :flex_prefix, :string # Flex type prefix (CECC, E9111, M, etc.)
-      attribute :number, Bsi::Components::Code
+      # `number`/`part`/`subpart` are plain strings, not Bsi::Components::Code:
+      # BSI never populates a field of that component beyond `value`. This class
+      # body is the sole home of Pubid::Bsi::Identifier (bsi/identifier.rb is a
+      # comment-only load stub), so declaring here is safe - every subclass body
+      # opens after it has run. `second_number` keeps the component type: it is
+      # not one of the three attributes the shared base declares.
+      attribute :number, :string
       attribute :iteration, :string # For bracket notation like 1000[9]
-      attribute :part, Bsi::Components::Code
-      attribute :subpart, Bsi::Components::Code
+      attribute :part, :string
+      attribute :subpart, :string
       attribute :second_number, Bsi::Components::Code # For collections like PAS 2035/2030
       attribute :date, Bsi::Components::Date
       attribute :stage, Pubid::Components::Stage
@@ -61,7 +70,7 @@ module Pubid
         return num_cmp unless num_cmp.zero?
 
         # Then by part
-        part_cmp = (part || Components::Code.new(value: "0")).to_s <=> (other.part || Components::Code.new(value: "0")).to_s
+        part_cmp = (part || "0").to_s <=> (other.part || "0").to_s
         return part_cmp unless part_cmp.zero?
 
         # Then by date
@@ -75,6 +84,54 @@ module Pubid
           0
         end
       end
+
+      # Annex F series groupings per "Rules for the structure and drafting
+      # of UK standards" (BSI, 2017), F.1.3-F.1.5. Each maps a `prefix`
+      # token to the broader series category.
+      SERIES_CATEGORIES = {
+        # F.1.3 Automobile
+        "AU" => :automotive,
+        # F.1.4 Aerospace
+        "A" => :aerospace, "B" => :aerospace, "C" => :aerospace,
+        "F" => :aerospace, "G" => :aerospace, "HC" => :aerospace,
+        "L" => :aerospace, "S" => :aerospace, "SP" => :aerospace,
+        "X" => :aerospace, "M" => :aerospace, "TA" => :aerospace,
+        "2A" => :aerospace, "2B" => :aerospace, "2C" => :aerospace,
+        "2F" => :aerospace, "2G" => :aerospace, "2HC" => :aerospace,
+        "2HR" => :aerospace, "2L" => :aerospace, "2M" => :aerospace,
+        "2S" => :aerospace, "2SP" => :aerospace, "2TA" => :aerospace,
+        "2X" => :aerospace, "3A" => :aerospace, "3B" => :aerospace,
+        "3F" => :aerospace, "3G" => :aerospace, "3HR" => :aerospace,
+        "3J" => :aerospace, "3L" => :aerospace, "3S" => :aerospace,
+        "3TA" => :aerospace, "4F" => :aerospace, "4L" => :aerospace,
+        "4S" => :aerospace, "5S" => :aerospace, "7S" => :aerospace,
+        # F.1.5 Marine
+        "MA" => :marine, "PL" => :marine, "QC" => :marine,
+      }.freeze
+
+      # Series accessor — semantically named alias for `prefix` per
+      # issue #252. Returns the prefix token (e.g., "AU", "MA", "2A").
+      def series
+        prefix
+      end
+
+      # Series category — one of :automotive, :aerospace, :marine, or nil
+      # when the identifier doesn't carry a recognized Annex F prefix.
+      def series_category
+        SERIES_CATEGORIES[prefix.to_s] if prefix
+      end
+
+      def automotive?
+        series_category == :automotive
+      end
+
+      def aerospace?
+        series_category == :aerospace
+      end
+
+      def marine?
+        series_category == :marine
+      end
     end
 
     # Backward-compatible alias: BSI's base class is Pubid::Bsi::Identifier.

data/lib/pubid/bsi/urn_generator.rb

@@ -21,8 +21,17 @@ module Pubid
           parts << identifier.flex_prefix.to_s.downcase
         end
 
-        if identifier.number
-          number = identifier.number.to_s
+        # Fall back to the wrapped document's number. A wrapper — an adopted
+        # European norm, a bundle, a set — carries no number of its own, and
+        # `AdoptedEuropeanNorm#number` only delegates ONE level, so a
+        # "DD ENV ISO 11079:1999" (which adopts a CEN prestandard that is
+        # itself a wrapper around the ISO standard) emitted the identity-free
+        # `urn:bsi:dd` — the same URN as every other DD adoption. `#root`
+        # recurses to the origin document, and for a non-wrapper it is `self`,
+        # so this changes nothing for an ordinary identifier.
+        urn_number = identifier.number || identifier.root.number
+        if urn_number
+          number = urn_number.to_s
           if identifier.iteration && !identifier.iteration.empty?
             number += "[#{identifier.iteration}]"
           end

data/lib/pubid/bsi.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+require "pubid"
 module Pubid
   module Bsi
     extend Pubid::PrefixesSupport
@@ -221,8 +222,13 @@ module Pubid
     ].freeze
 
     def self.parse(string)
+      unless string.is_a?(String)
+        raise Pubid::Errors::InvalidInputError,
+              Pubid::INPUT_NOT_A_STRING_MESSAGE
+      end
+
       if string.length > Pubid::MAX_INPUT_LENGTH
-        raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
+        raise Pubid::Errors::InvalidInputError, Pubid::INPUT_TOO_LONG_MESSAGE
       end
 
       Identifier.parse(string)