parse source details

lib/termium.rb

@@ -23,6 +23,7 @@ class Error < StandardError; end
 require_relative "termium/language_module"
 require_relative "termium/parameter"
 require_relative "termium/source"
+require_relative "termium/source_details"
 require_relative "termium/source_ref"
 require_relative "termium/subject"
 require_relative "termium/textual_support"

lib/termium/source.rb

@@ -1,26 +1,30 @@
 # frozen_string_literal: true
 
+require_relative "source_details"
+
 module Termium
   # For <source>
   class Source < Lutaml::Model::Serializable
     ISO_BIB_REGEX = /\AISO-([\d-]+)\s+\*\s+(\d{4})\s+.*/.freeze
     ISOIEC_BIB_REGEX = /\AISO-IEC-([\d-]+)\s+\*\s+(\d{4})\s+.*/.freeze
 
     attribute :order, :integer
-    attribute :details, :string
+    attribute :details, SourceDetails
+
     xml do
       root "source"
       map_attribute "order", to: :order
       map_attribute "details", to: :details
     end
 
     def content
-      if (matches = details.match(ISOIEC_BIB_REGEX))
+      presentable_details = details.to_s
+      if (matches = presentable_details.match(ISOIEC_BIB_REGEX))
         "ISO/IEC #{matches[1]}:#{matches[2]}"
-      elsif (matches = details.match(ISO_BIB_REGEX))
+      elsif (matches = presentable_details.match(ISO_BIB_REGEX))
         "ISO #{matches[1]}:#{matches[2]}"
       else
-        details
+        presentable_details
       end
     end
 

lib/termium/source_details.rb

@@ -0,0 +1,61 @@
+# frozen_string_literal: true
+
+module Termium
+  # Parses the asterisk-delimited source details string
+  class SourceDetails
+    attr_accessor :raw, :standard, :author_name, :year, :organization, :department, :office, :division, :role
+
+    class << self
+      def cast(value)
+        return value if value.is_a?(SourceDetails)
+        return nil if value.nil? || value.to_s.strip.empty?
+
+        new(
+          value
+        )
+      end
+
+      def serialize(value)
+        value&.raw
+      end
+
+      alias from_xml cast
+    end
+
+    def initialize(value)
+      return if value.nil? || value.to_s.strip.empty?
+
+      columns = value.to_s.split("*").map(&:strip)
+
+      @raw = value.to_s
+      @year = presence(columns[1])
+      @organization = presence(columns[2])
+      @department = presence(columns[3])
+      @office = presence(columns[4])
+      @division = presence(columns[5])
+      @role = presence(columns[6])
+
+      first_column = presence(columns[0])
+      if standard_identifier?(first_column)
+        @standard = first_column
+      else
+        @author_name = first_column
+      end
+    end
+
+    def presence(value)
+      value unless value.nil? || value.empty?
+    end
+
+    # Matches ISO or ISO-IEC standard identifiers with a document number
+    # e.g., "ISO-9000", "ISO-IEC-2382-16", "ISO-IEC-27001"
+    def standard_identifier?(value)
+      return false if value.nil?
+
+      value.match?(/\AISO(-IEC)?-\d+/i)
+    end
+
+    alias to_xml raw
+    alias to_s raw
+  end
+end

spec/termium/source_details_spec.rb

@@ -0,0 +1,108 @@
+# frozen_string_literal: true
+
+RSpec.describe Termium::SourceDetails do
+  describe ".cast" do
+    context "with ISO-IEC standard identifier" do
+      let(:input) { "ISO-IEC-2382-16 * 1996 *  *  * " }
+      subject(:details) { described_class.cast(input) }
+
+      it "sets standard from first column" do
+        expect(details.standard).to eq("ISO-IEC-2382-16")
+      end
+
+      it "sets year" do
+        expect(details.year).to eq("1996")
+      end
+
+      it "does not set author_name" do
+        expect(details.author_name).to be_nil
+      end
+
+      it "does not set organization" do
+        expect(details.organization).to be_nil
+      end
+    end
+
+    context "with organization name (not a standard identifier)" do
+      let(:input) { "ISO/IEC JTC 1 * 2011" }
+      subject(:details) { described_class.cast(input) }
+
+      it "sets author_name from first column" do
+        expect(details.author_name).to eq("ISO/IEC JTC 1")
+      end
+
+      it "sets year" do
+        expect(details.year).to eq("2011")
+      end
+
+      it "does not set standard" do
+        expect(details.standard).to be_nil
+      end
+    end
+
+    context "with personal author (has organization)" do
+      let(:input) { "Ranger, Natalie * 2006 * Bureau de la traduction * Services linguistiques" }
+      subject(:details) { described_class.cast(input) }
+
+      it "sets author_name from first column" do
+        expect(details.author_name).to eq("Ranger, Natalie")
+      end
+
+      it "sets year" do
+        expect(details.year).to eq("2006")
+      end
+
+      it "sets organization" do
+        expect(details.organization).to eq("Bureau de la traduction")
+      end
+
+      it "sets department" do
+        expect(details.department).to eq("Services linguistiques")
+      end
+
+      it "does not set standard" do
+        expect(details.standard).to be_nil
+      end
+    end
+
+    context "when value is already SourceDetails" do
+      let(:original) { described_class.cast("ISO-IEC-2382-16 * 1996 *  *  * ") }
+
+      it "returns the same object" do
+        expect(described_class.cast(original)).to be(original)
+      end
+    end
+
+    context "with nil input" do
+      it "returns nil" do
+        expect(described_class.cast(nil)).to be_nil
+      end
+    end
+
+    context "with empty string" do
+      it "returns nil" do
+        expect(described_class.cast("")).to be_nil
+      end
+    end
+  end
+
+  describe ".serialize" do
+    let(:details) { described_class.cast("ISO-IEC-2382-16 * 1996 *  *  * ") }
+
+    it "returns the raw string" do
+      expect(described_class.serialize(details)).to eq("ISO-IEC-2382-16 * 1996 *  *  * ")
+    end
+
+    it "returns nil for nil input" do
+      expect(described_class.serialize(nil)).to be_nil
+    end
+  end
+
+  describe "#to_s" do
+    let(:details) { described_class.cast("Ranger, Natalie * 2006 * Bureau de la traduction") }
+
+    it "returns the raw string" do
+      expect(details.to_s).to eq("Ranger, Natalie * 2006 * Bureau de la traduction")
+    end
+  end
+end

spec/termium/source_spec.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+RSpec.describe Termium::Source do
+  describe "XML round-trip" do
+    let(:xml_input) { '<source order="1" details="ISO-IEC-2382-16 * 1996 *  *  * "/>' }
+
+    it "preserves details attribute" do
+      source = described_class.from_xml(xml_input)
+      output = source.to_xml
+
+      expect(output).to include('details="ISO-IEC-2382-16 * 1996 *  *  * "')
+    end
+
+    it "parses details into SourceDetails object" do
+      source = described_class.from_xml(xml_input)
+
+      expect(source.details).to be_a(Termium::SourceDetails)
+      expect(source.details.standard).to eq("ISO-IEC-2382-16")
+    end
+  end
+
+  describe "#content" do
+    context "with ISO-IEC standard" do
+      let(:source) { described_class.from_xml('<source order="1" details="ISO-IEC-2382-16 * 1996 *  *  * "/>') }
+
+      it "formats as ISO/IEC reference" do
+        expect(source.content).to eq("ISO/IEC 2382-16:1996")
+      end
+    end
+
+    context "with ISO standard" do
+      let(:source) { described_class.from_xml('<source order="1" details="ISO-2382-12 * 1988 *  *  * "/>') }
+
+      it "formats as ISO reference" do
+        expect(source.content).to eq("ISO 2382-12:1988")
+      end
+    end
+
+    context "with non-standard source" do
+      let(:source) { described_class.from_xml('<source order="1" details="Ranger, Natalie * 2006 * Bureau de la traduction"/>') }
+
+      it "returns raw details" do
+        expect(source.content).to eq("Ranger, Natalie * 2006 * Bureau de la traduction")
+      end
+    end
+  end
+end