pubid 2.0.0.pre.alpha.2 → 2.0.0.pre.alpha.4

data/lib/pubid/iec/builder.rb

@@ -1,25 +1,16 @@
-require_relative "../components/publisher"
 # frozen_string_literal: true
-require_relative "../components/code"
-require_relative "../components/date"
 
 module Pubid
   module Iec
     class Builder < Pubid::Builder::Base
-      def initialize(scheme)
-        @scheme = scheme
-      end
-
       def locate_identifier_klass(parsed_hash)
         # Check for working programme
         if parsed_hash[:wp_stage]
-          require_relative "identifiers/working_document"
           return Identifiers::WorkingDocument
         end
 
         # Check for working document
         if parsed_hash[:technical_committee] && parsed_hash[:wd_number]
-          require_relative "identifiers/working_document"
           return Identifiers::WorkingDocument
         end
 
@@ -27,7 +18,8 @@ module Pubid
         # :type_with_stage will be nil if it is an IS.
         typed_stage = locate_typed_stage(parsed_hash[:type_with_stage])
 
-        @scheme.locate_identifier_klass_by_type_code(typed_stage.type_code)
+        Pubid::Iec.locate_type(typed_stage.type_code) ||
+          raise(ArgumentError, "Unknown type code: #{typed_stage.type_code}")
       end
 
       def build(parsed_hash)
@@ -55,7 +47,8 @@ module Pubid
 
           # Locate the supplement identifier class
           typed_stage = locate_typed_stage(supplement_type)
-          identifier_class = @scheme.locate_identifier_klass_by_type_code(typed_stage.type_code)
+          identifier_class = Pubid::Iec.locate_type(typed_stage.type_code) ||
+            raise(ArgumentError, "Unknown type code: #{typed_stage.type_code}")
 
           # Create the supplement identifier
           supplement = identifier_class.new
@@ -96,14 +89,11 @@ module Pubid
 
         if is_fragment
           # For FRAG typed stages, look up the fragment's own typed stage
-          begin
-            fragment_typed_stage = @scheme.locate_typed_stage_by_abbr(
-              parsed_hash[:type_with_stage].to_s,
-            )
-          rescue ArgumentError
-            # FRAGC is a rendering notation, not a typed stage abbreviation
-            fragment_typed_stage = @scheme.locate_typed_stage_by_abbr("FRAG")
-          end
+          fragment_typed_stage = Pubid::Iec.locate_stage(parsed_hash[:type_with_stage])
+
+          # FRAGC is a rendering notation, not a typed stage abbreviation;
+          # fall back to the canonical "FRAG" typed stage
+          fragment_typed_stage ||= Pubid::Iec.locate_stage("FRAG")
 
           # For FRAG/FRAGC, we need to build the base Amendment/Corrigendum
           # then wrap it with FragmentIdentifier
@@ -158,7 +148,6 @@ module Pubid
 
         # Detect rendering style from parsed abbreviation
         if identifier.class.attributes.key?(:rendering_style) && identifier.typed_stage
-          require_relative "rendering_style"
           ts = identifier.typed_stage
 
           # Detect IEC format from parsed abbreviation
@@ -179,7 +168,7 @@ module Pubid
           with_language_code = if langs&.any?
                                  # Check if original_code was single-char (E, F, R, A, S, D)
                                  first_lang = langs.first
-                                 if first_lang.is_a?(Components::Language) && first_lang.original_code && first_lang.original_code.length == 1
+                                 if first_lang.is_a?(::Pubid::Components::Language) && first_lang.original_code && first_lang.original_code.length == 1
                                    :single
                                  else
                                    :iso # 2-char codes (en, fr, ru, ar, es, de)
@@ -202,7 +191,7 @@ module Pubid
 
         # After building base identifier, apply wrappers
         if fragment_type_data && fragment_number_data
-          frag_ts = @scheme.locate_typed_stage_by_abbr("FRAG")
+          frag_ts = Pubid::Iec.locate_stage("FRAG")
           identifier = wrap_with_fragment(identifier, fragment_number_data.to_s,
                                           nil, frag_ts)
         end
@@ -229,11 +218,25 @@ module Pubid
         end
       end
 
+      # The IEC builder's own `build` method drives identifier construction
+      # (it locates the right class per parsed hash), so the parent class's
+      # `@identifier_class` slot is never read. Provide a placeholder so
+      # `Builder.new` doesn't fail with NotImplementedError.
+      def default_identifier_class
+        Identifiers::InternationalStandard
+      end
+
+      # Look up a typed stage by abbreviation via the self-describing
+      # `Pubid::Iec` module instead of an injected scheme instance.
+      def locate_typed_stage(typed_stage_string)
+        typed_stage_string = "" if typed_stage_string.nil?
+        Pubid::Iec.locate_stage(typed_stage_string) ||
+          raise(ArgumentError, "Unknown type abbreviation: '#{typed_stage_string}'")
+      end
+
       # Wrap identifier with FragmentIdentifier for /FRAGN notation
       def wrap_with_fragment(base_identifier, fragment_number,
-edition_data = nil, typed_stage = nil)
-        require_relative "identifiers/fragment_identifier"
-
+      edition_data = nil, typed_stage = nil)
         fragment = Identifiers::FragmentIdentifier.new(
           base_identifier: base_identifier,
           fragment_number: fragment_number,
@@ -252,8 +255,6 @@ edition_data = nil, typed_stage = nil)
 
       # Wrap identifier with SheetIdentifier for /N:YEAR notation
       def wrap_with_sheet(base_identifier, sheet_number, sheet_year)
-        require_relative "identifiers/sheet_identifier"
-
         # Only pass year if it's present
         year_value = sheet_year&.to_s
 
@@ -272,10 +273,6 @@ edition_data = nil, typed_stage = nil)
 
       # Wrap identifier with ConsolidatedIdentifier for +AMD chains
       def wrap_with_consolidated(base_identifier, supplements_data)
-        require_relative "identifiers/consolidated_identifier"
-        require_relative "identifiers/amendment"
-        require_relative "identifiers/corrigendum"
-
         supplements = supplements_data.filter_map do |supp|
           type = supp[:supplement_type].to_s
           number = supp[:supplement_number].to_s
@@ -302,13 +299,10 @@ edition_data = nil, typed_stage = nil)
         )
       end
 
-      # Wrap identifier with VapIdentifier for CSV/CMV/RLV/SER suffix
+      # Wrap identifier with VapIdentifier for one or more VAP codes
+      # (e.g. "CSV" or "EXV-CMV").
       def wrap_with_vap(base_identifier, vap_suffix_data)
-        require_relative "identifiers/vap_identifier"
-        require_relative "identifiers/consolidated_identifier"
-        require_relative "components/vap_suffix"
-
-        vap_suffix = Components::VapSuffix.new(code: vap_suffix_data.to_s)
+        vap = vap_suffix_data.to_s.split("-")
 
         # Extract edition - need to go deep for ConsolidatedIdentifier
         edition = nil
@@ -326,7 +320,7 @@ edition_data = nil, typed_stage = nil)
 
         Identifiers::VapIdentifier.new(
           base_identifier: base_identifier,
-          vap_suffix: vap_suffix,
+          vap: vap,
           edition: edition,
         )
       end
@@ -337,6 +331,12 @@ edition_data = nil, typed_stage = nil)
         super
       end
 
+      # IEC normalizes em/en dashes to hyphens but preserves slashes and
+      # spaces, which can be semantically meaningful in IEC identifiers.
+      def normalize_number_with_part(value)
+        value.to_s.tr(Parser::DASH_CHARS.join, "-")
+      end
+
       def cast(type, value)
         case type
         when :base_identifier
@@ -374,28 +374,7 @@ edition_data = nil, typed_stage = nil)
           # or "60038-1" ('1' is part)
           # or "60038-1-2" ('1' is part, '2' is subpart)
           # or "29110-5-1-1" ('5' is part, '1-1' is subpart)
-
-          # Split the number into parts
-          normalized_value = value.to_s.tr(Parser::DASH_CHARS.join, "-")
-
-          parts = normalized_value.split("-")
-          number = parts.shift # The first part is always the number
-          part = parts.shift # The second part is the part, if present
-          subpart = parts.any? ? parts.join("-") : nil # The remaining parts form the subpart, if present
-
-          part = convert_roman_to_integer(part)
-
-          code_hash = { number: Components::Code.new(value: number) }
-
-          if part
-            code_hash[:part] = Components::Code.new(value: part)
-          end
-
-          if subpart
-            code_hash[:subpart] = Components::Code.new(value: subpart)
-          end
-
-          code_hash
+          parse_number_with_part(value, code_class: Components::Code)
 
         when :type_with_stage
           # "WD"
@@ -416,13 +395,17 @@ edition_data = nil, typed_stage = nil)
           }
         when :stage_iteration
           # "1" or "2"
-          Components::Code.new(value: value.to_s)
+          Pubid::Components::Iteration.new(number: value.to_s)
 
         when :date
           parse_date(value)
 
         when :edition
-          Pubid::Components::Edition.new(number: value)
+          # Coerce to String: the parsed value may be a Parslet::Slice, which
+          # Edition.number (Type::Value) would store verbatim and serialize as a
+          # !ruby/object:Parslet::Slice node — unloadable by the index reader's
+          # YAML.safe_load(permitted_classes: [Symbol]).
+          Pubid::Components::Edition.new(number: value&.to_s)
 
         when :sheet_number
           value.to_s
@@ -445,8 +428,7 @@ edition_data = nil, typed_stage = nil)
         when :joint_identifier
           case value[:publisher]
           when "ISO"
-            require_relative "../iso/builder"
-            Iso::Builder.new(Iso::Scheme).build(value)
+            Pubid::Iso.build_from_parse(value)
           end
 
         else

data/lib/pubid/iec/components/code.rb

@@ -1,58 +1,34 @@
-require "lutaml/model"
-require_relative "../../components/code"
 # frozen_string_literal: true
 
 module Pubid
   module Iec
     module Components
+      # Code component for IEC identifiers.
+      #
+      # Inherits +value+ and +prefix+ from the shared Code; overrides
+      # rendering to place the prefix (e.g. "TR", "TS") before the value.
       class Code < ::Pubid::Components::Code
-        attribute :prefix, :string, default: -> {}
-        attribute :number, :string
-        attribute :part, :string, default: -> {}
-
-        def initialize(value: nil, number: nil, prefix: nil, part: nil)
-          # Support both 'value' (for convenience) and 'number' (explicit)
-          @number = value || number
-          @prefix = prefix
-          @part = part
-        end
-
         def to_s
           result = ""
           result += "#{prefix} " if prefix
-          result += number
-          result += "-#{part}" if part
+          result += value.to_s
           result
         end
 
-        def full_code
-          to_s
-        end
+        alias full_code to_s
 
         # Parse IEC code formats:
         # - "60034" (just number)
-        # - "60034-1" (number with part)
         # - "TR 61000" (with prefix)
         # - "TS 62443" (with prefix)
         def self.parse(string)
           parts = string.strip.split(" ", 2)
 
           if parts.size == 2
-            # Has prefix: "TR 61000-1-2"
-            prefix = parts[0]
-            number_part = parts[1]
+            new(prefix: parts[0], value: parts[1])
           else
-            # No prefix: "60034-1"
-            prefix = nil
-            number_part = parts[0]
+            new(value: parts[0])
           end
-
-          # Split number and part
-          number_parts = number_part.split("-", 2)
-          number = number_parts[0]
-          part = number_parts[1]
-
-          new(prefix: prefix, number: number, part: part)
         end
       end
     end

data/lib/pubid/iec/components/publisher.rb

@@ -1,5 +1,4 @@
 require "lutaml/model"
-require_relative "../../components/publisher"
 # frozen_string_literal: true
 
 module Pubid

data/lib/pubid/iec/components.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iec
+    module Components
+      autoload :Code, "#{__dir__}/components/code"
+      autoload :ConsolidatedAmendment, "#{__dir__}/components/consolidated_amendment"
+      autoload :Publisher, "#{__dir__}/components/publisher"
+      autoload :Sheet, "#{__dir__}/components/sheet"
+      autoload :TrfInfo, "#{__dir__}/components/trf_info"
+      autoload :VapSuffix, "#{__dir__}/components/vap_suffix"
+    end
+  end
+end