glossarist 2.6.4 → 2.6.6

data/config.yml

@@ -26,23 +26,90 @@ designation:
   base:
     normative_status:
       - preferred
-      - deprecated
       - admitted
+      - deprecated
+      - superseded
       - <símbolo> # in iev-data => '<symbol>'
       - 티에스 # in iev-data => translates to 'TS' I think it is a synonym and not status.
       - prąd startowy # in iev-data => 'starting current' I think it is a synonym and not status.
+  relationship_type:
+    # Term-level (designation-to-designation within the same concept entry)
+    - abbreviated_form_for
+    - short_form_for
 
 related_concept:
   type:
+    # Lifecycle (ISO 10241-1)
     - deprecates
     - supersedes
     - superseded_by
+    # Hierarchical (ISO 10241-1 / ISO 25964)
     - narrower
     - broader
+    # Hierarchical sub-types — ISO 25964 generic (BTG/NTG)
+    - broader_generic
+    - narrower_generic
+    # Hierarchical sub-types — ISO 25964 partitive (BTP/NTP)
+    - broader_partitive
+    - narrower_partitive
+    # Hierarchical sub-types — ISO 25964 instantial (BTI/NTI)
+    - broader_instantial
+    - narrower_instantial
+    # Equivalence (ISO 10241-1 / ISO 25964 exactMatch / SKOS)
     - equivalent
+    # Approximate equivalence (ISO 25964 closeMatch / SKOS)
+    - close_match
+    # Cross-vocabulary mapping (SKOS)
+    - broad_match
+    - narrow_match
+    - related_match
+    # Comparative (ISO 10241-1)
     - compare
     - contrast
+    # Associative (ISO 10241-1 / ISO 25964 RT / TBX crossReference)
     - see
+    # Associative sub-types (ISO 25964 / TBX)
+    - related_concept
+    - related_concept_broader
+    - related_concept_narrower
+    # Associative — spatial/temporal (ISO 25964 / TBX)
+    - sequentially_related_concept
+    - spatially_related_concept
+    - temporally_related_concept
+    # Lexical (ISO 12620 / TBX)
+    - homograph
+    - false_friend
+
+iso12620:
+  term_type:
+    # ISO 12620 term type classification for designations.
+    # Orthographic / structural:
+    - abbreviation
+    - acronym
+    - clipped_term
+    - full_form
+    - initialism
+    - short_form
+    - transliterated_form
+    - transcribed_form
+    - variant
+    # Symbolic / formulaic:
+    - equation
+    - formula
+    - logical_expression
+    - symbol
+    # Usage / provenance:
+    - common_name
+    - entry_term
+    - internationalism
+    - international_scientific_term
+    - part_number
+    - phraseological_unit
+    - shortcut
+    - sku
+    - standard_text
+    - synonym
+    - synonymous_phrase
 
 abbreviation:
   type:

data/glossarist.gemspec

@@ -32,9 +32,9 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
 
   spec.add_dependency "lutaml-model", "~> 0.8.5"
-  spec.add_dependency "sts", "~> 0.5.6"
   spec.add_dependency "relaton", ">= 2.0.0", "< 3"
   spec.add_dependency "rubyzip", ">= 2.3", "< 3"
+  spec.add_dependency "sts", "~> 0.5.6"
   spec.add_dependency "tbx", "~> 0.1"
   spec.add_dependency "thor"
 end

data/lib/glossarist/asset_reference.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Glossarist
+  class AssetReference
+    attr_reader :path, :location
+
+    def initialize(path:, location: nil)
+      @path = path
+      @location = location
+    end
+
+    def dedup_key
+      path
+    end
+  end
+end

data/lib/glossarist/bibliographic_reference.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Glossarist
+  class BibliographicReference
+    attr_reader :anchor, :location
+
+    def initialize(anchor:, location: nil)
+      @anchor = anchor
+      @location = location
+    end
+
+    def dedup_key
+      anchor
+    end
+  end
+end

data/lib/glossarist/concept.rb

@@ -4,7 +4,7 @@ module Glossarist
     attribute :id, :string
     attribute :uuid, :string
     attribute :subject, :string
-    attribute :non_verb_rep, :string
+    attribute :non_verb_rep, NonVerbRep, collection: true
     attribute :extension_attributes, :string
     attribute :lineage_source, :string
     attribute :localizations, :hash

data/lib/glossarist/concept_data.rb

@@ -30,6 +30,8 @@ module Glossarist
     # Language code should be exactly 3 char long.
     # TODO: use min_length, max_length once added in lutaml-model
     attribute :language_code, :string, pattern: /^.{3}$/
+    attribute :script, :string
+    attribute :system, :string
     attribute :entry_status, :string
 
     key_value do
@@ -48,6 +50,8 @@ module Glossarist
       map :references, to: :references
       map :domain, to: :domain
       map %i[language_code languageCode], to: :language_code
+      map :script, to: :script
+      map :system, to: :system
       map %i[entry_status entryStatus], to: :entry_status
       map %i[review_date reviewDate], to: :review_date
       map %i[review_decision_date reviewDecisionDate], to: :review_decision_date

data/lib/glossarist/concept_enricher.rb

@@ -8,6 +8,7 @@ module Glossarist
       concepts.each do |mc|
         mc.localizations.each do |l10n|
           refs = extractor.extract_from_localized_concept(l10n)
+            .grep(ConceptReference)
           next if refs.empty?
 
           existing = l10n.data.references || []

data/lib/glossarist/concept_reference.rb

@@ -1,17 +1,29 @@
 # frozen_string_literal: true
 
 module Glossarist
+  # A typed reference to another concept, either local (within the same
+  # glossary) or external (in another concept registry).
+  #
+  # Local references use +concept_id+ without +source+. External references
+  # use +source+ (a registry URN prefix) and +concept_id+ to identify the
+  # target concept, or a direct +urn+ field for formal URN references.
   class ConceptReference < Lutaml::Model::Serializable
     attribute :term, :string
     attribute :concept_id, :string
     attribute :source, :string
     attribute :ref_type, :string
+    attribute :urn, :string
 
     key_value do
       map :term, to: :term
       map :concept_id, to: :concept_id
       map :source, to: :source
       map :ref_type, to: :ref_type
+      map :urn, to: :urn
+    end
+
+    def self.domain(concept_id)
+      new(concept_id: concept_id, ref_type: "domain")
     end
 
     def local?
@@ -23,23 +35,8 @@ module Glossarist
       !local?
     end
 
-    def to_urn
-      return nil unless external?
-      return nil unless source && concept_id
-
-      case source
-      when /\Aurn:iec/ then "#{source}-#{concept_id}"
-      when /\Aurn:iso/ then "#{source}:term:#{concept_id}"
-      else "#{source}/#{concept_id}"
-      end
-    end
-
-    def to_gcr_hash
-      h = { "term" => term }
-      h["concept_id"] = concept_id if concept_id
-      h["source"] = source if source
-      h["ref_type"] = ref_type if ref_type
-      h.compact
+    def dedup_key
+      concept_id ? [source, concept_id] : [source, concept_id, term]
     end
   end
 end

data/lib/glossarist/concept_validator.rb

@@ -2,9 +2,6 @@
 
 module Glossarist
   class ConceptValidator
-    LANG_CODES = Glossarist::LANG_CODES
-    VALID_ENTRY_STATUSES = %w[valid superseded withdrawn draft].freeze
-
     attr_reader :path, :errors, :warnings
 
     def initialize(path)
@@ -14,23 +11,46 @@ module Glossarist
     end
 
     def validate_all
-      seen_ids = {}
+      result = ValidationResult.new
+      context = Validation::Rules::DatasetContext.new(@path)
+      concept_rules = Validation::Rules::Registry.for_scope(:concept)
       file_idx = 0
 
       ConceptCollector.each_concept(@path) do |concept|
         fname = concept_file_name(concept, file_idx)
-        validate_concept(concept, fname, seen_ids)
+        concept_context = Validation::Rules::ConceptContext.new(
+          concept, file_name: fname, collection_context: context
+        )
+
+        concept_rules.each do |rule|
+          next unless rule.applicable?(concept_context)
+
+          rule.check(concept_context).each { |i| result.add_issue(i) }
+        end
+
         file_idx += 1
       end
 
       if file_idx.zero?
         yaml_files = find_yaml_files
         if yaml_files.any?
-          @errors << "YAML files found but no parseable concepts"
+          result.add_error("YAML files found but no parseable concepts")
         end
       end
 
-      ValidationResult.new(errors: @errors, warnings: @warnings)
+      # Run collection-level rules
+      collection_rules = Validation::Rules::Registry.for_scope(:collection)
+      collection_rules.each do |rule|
+        next unless rule.applicable?(context)
+
+        rule.check(context).each { |i| result.add_issue(i) }
+      end
+
+      # Sync legacy arrays for backward compatibility
+      @errors = result.errors
+      @warnings = result.warnings
+
+      result
     end
 
     private
@@ -48,54 +68,5 @@ module Glossarist
       id = concept.data&.id
       id ? "concept-#{id}.yaml" : "concept-#{idx}.yaml"
     end
-
-    def validate_concept(concept, fname, seen_ids)
-      validate_id(concept, fname, seen_ids)
-      validate_localizations(concept, fname)
-      validate_entry_statuses(concept, fname)
-    end
-
-    def validate_id(concept, fname, seen_ids)
-      id = concept.data&.id
-      unless id
-        @errors << "#{fname}: missing concept id"
-        return
-      end
-
-      id_str = id.to_s
-      if seen_ids[id_str]
-        @errors << "#{fname}: duplicate id '#{id_str}' (first seen in #{seen_ids[id_str]})"
-      else
-        seen_ids[id_str] = fname
-      end
-    end
-
-    def validate_localizations(concept, fname)
-      l10ns = concept.localizations&.values || []
-      if l10ns.empty?
-        @errors << "#{fname}: no localizations found"
-        return
-      end
-
-      l10ns.each do |l10n|
-        lang = l10n.language_code || "unknown"
-        terms = l10n.data&.terms
-        unless terms.is_a?(Array) && terms.any?
-          @errors << "#{fname}/#{lang}: must have at least 1 term"
-        end
-      end
-    end
-
-    def validate_entry_statuses(concept, fname)
-      (concept.localizations&.values || []).each do |l10n|
-        lang = l10n.language_code || "unknown"
-        status = l10n.data&.entry_status
-        next unless status
-
-        unless VALID_ENTRY_STATUSES.include?(status)
-          @errors << "#{fname}/#{lang}: invalid entry_status '#{status}' (expected one of: #{VALID_ENTRY_STATUSES.join(', ')})"
-        end
-      end
-    end
   end
 end

data/lib/glossarist/dataset_validator.rb

@@ -3,43 +3,53 @@
 module Glossarist
   class DatasetValidator
     def validate(path, strict: false, reference_path: nil)
-      result = validate_path(path)
+      if File.extname(path).downcase == ".gcr"
+        validate_gcr(path, reference_path: reference_path)
+      else
+        validate_directory(path, reference_path: reference_path)
+      end
+    end
+
+    private
+
+    def validate_gcr(path, reference_path: nil)
+      result = GcrValidator.new.validate(path)
 
       if reference_path
-        ref_result = validate_cross_references(path, reference_path)
+        ref_result = validate_gcr_cross_references(path, reference_path)
         result.merge(ref_result)
       end
 
       result
     end
 
-    private
+    def validate_directory(path, reference_path: nil)
+      result = ConceptValidator.new(path).validate_all
 
-    def validate_path(path)
-      if File.extname(path).downcase == ".gcr"
-        validate_gcr(path)
-      else
-        validate_directory(path)
+      if reference_path
+        ref_result = validate_directory_cross_references(path, reference_path)
+        result.merge(ref_result)
       end
-    end
 
-    def validate_gcr(path)
-      GcrValidator.new.validate(path)
+      result
     end
 
-    def validate_directory(path)
-      ConceptValidator.new(path).validate_all
+    def validate_gcr_cross_references(path, reference_path)
+      extractor = ReferenceExtractor.new
+      resolver = build_resolver(reference_path)
+      pkg = GcrPackage.load(path)
+      uri_prefix = pkg.metadata&.dig("uri_prefix") || pkg.metadata&.dig("shortname")
+      resolver.register_self(pkg.concepts)
+      resolver.register_package(pkg, uri_prefix: uri_prefix)
+      resolver.validate_all(pkg, extractor: extractor)
     end
 
-    def validate_cross_references(path, reference_path)
+    def validate_directory_cross_references(path, reference_path)
       extractor = ReferenceExtractor.new
       resolver = build_resolver(reference_path)
-
-      if File.extname(path).downcase == ".gcr"
-        validate_gcr_refs(resolver, path, extractor)
-      else
-        validate_directory_refs(resolver, path, extractor)
-      end
+      concepts = ConceptCollector.collect(path)
+      resolver.register_self(concepts)
+      resolver.validate_all(concepts, extractor: extractor)
     end
 
     def build_resolver(reference_path)
@@ -51,19 +61,5 @@ module Glossarist
       end
       resolver
     end
-
-    def validate_gcr_refs(resolver, path, extractor)
-      pkg = GcrPackage.load(path)
-      uri_prefix = pkg.metadata&.dig("uri_prefix") || pkg.metadata&.dig("shortname")
-      resolver.register_self(pkg.concepts)
-      resolver.register_package(pkg, uri_prefix: uri_prefix)
-      resolver.validate_all(pkg, extractor: extractor)
-    end
-
-    def validate_directory_refs(resolver, path, extractor)
-      concepts = ConceptCollector.collect(path)
-      resolver.register_self(concepts)
-      resolver.validate_all(concepts, extractor: extractor)
-    end
   end
 end

data/lib/glossarist/designation/abbreviation.rb

@@ -1,7 +1,6 @@
 module Glossarist
   module Designation
     class Abbreviation < Expression
-      attribute :international, :boolean
       attribute :type, :string, default: -> { "abbreviation" }
 
       Glossarist::GlossaryDefinition::ABBREVIATION_TYPES.each do |name|
@@ -9,7 +8,6 @@ module Glossarist
       end
 
       key_value do
-        map :international, to: :international
         map :type, to: :type, render_default: true
         Glossarist::GlossaryDefinition::ABBREVIATION_TYPES.each do |name|
           map name.to_sym, to: name.to_sym

data/lib/glossarist/designation/base.rb

@@ -6,12 +6,30 @@ module Glossarist
       attribute :normative_status, :string,
                 values: Glossarist::GlossaryDefinition::DESIGNATION_BASE_NORMATIVE_STATUSES
       attribute :type, :string
+      attribute :language, :string
+      attribute :script, :string
+      attribute :system, :string
+      attribute :international, :boolean
+      attribute :absent, :boolean
+      attribute :pronunciation, Pronunciation, collection: true
+      attribute :sources, ConceptSource, collection: true
+      attribute :term_type, :string
+      attribute :related, RelatedConcept, collection: true
 
       key_value do
         map :type, to: :type
         map %i[normative_status normativeStatus], to: :normative_status
         map %i[geographical_area geographicalArea], to: :geographical_area
         map :designation, to: :designation
+        map :language, to: :language
+        map :script, to: :script
+        map :system, to: :system
+        map :international, to: :international
+        map :absent, to: :absent
+        map :pronunciation, to: :pronunciation
+        map :sources, to: :sources
+        map %i[term_type termType], to: :term_type
+        map :related, to: :related
       end
 
       def self.of_yaml(hash, options = {})
@@ -34,7 +52,9 @@ module Glossarist
       end
 
       def self.infer_designation_type(hash)
-        if hash["international"] || hash["abbreviation_type"]
+        if hash["abbreviation_type"]
+          "abbreviation"
+        elsif hash["international"]
           "symbol"
         else
           "expression"

data/lib/glossarist/designation/expression.rb

@@ -5,6 +5,7 @@ module Glossarist
     class Expression < Base
       attribute :prefix, :string
       attribute :usage_info, :string
+      attribute :field_of_application, :string
 
       attribute :gender, :string
       attribute :plurality, :string
@@ -16,6 +17,8 @@ module Glossarist
         map :type, to: :type, render_default: true
         map :prefix, to: :prefix
         map %i[usage_info usageInfo], to: :usage_info
+        map %i[field_of_application fieldOfApplication],
+            to: :field_of_application
         map %i[grammar_info grammarInfo], to: :grammar_info
       end
 

data/lib/glossarist/designation/letter_symbol.rb

@@ -2,13 +2,9 @@ module Glossarist
   module Designation
     class LetterSymbol < Symbol
       attribute :text, :string
-      attribute :language, :string
-      attribute :script, :string
 
       key_value do
         map :text, to: :text
-        map :language, to: :language
-        map :script, to: :script
       end
 
       def self.of_yaml(hash, options = {})

data/lib/glossarist/designation/symbol.rb

@@ -1,11 +1,9 @@
 module Glossarist
   module Designation
     class Symbol < Base
-      attribute :international, :boolean
       attribute :type, :string, default: -> { "symbol" }
 
       key_value do
-        map :international, to: :international
         map :type, to: :type, render_default: true
       end