glossarist 2.8.2 → 2.8.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f602beb2e985225705695343d1309f242117b6b59398f27b9b7d4966e2300c98
-  data.tar.gz: e5891cad0fd4efecac7140d007f36efe0a2f83609013a771ac02621199ba0306
+  metadata.gz: bb636339e439da216185715fcfe387842aabd1b38147dd49b5e4fdb5909b6d2d
+  data.tar.gz: 2704b81c1f4d179d8909bc1d245b9caee83d2e54fa56a6148699892f021e9d6f
 SHA512:
-  metadata.gz: 97a81b187d34665f43c2800b3ac7636de9ae71b83c832bb9fdba0f5d9895089f6c3229ba80d38b3e9b0479ab71f16bf282b924c4ee1f9292059c250ade950cfe
-  data.tar.gz: 5d5c43b11134a5f072d80d99c5110471dae8145a4c3212aa9cce49ac1c491fd478f748cd4f57adf97a5c3212a41c7ca679c31a07b6018cda6ecf77cdef3fff64
+  metadata.gz: 5e39df7a02295426fb6463f93189cf6e3d457e60457b9bb5cde1881eeb549fe1afd4730e3ac280030be8de60734dbc2e0532b14030890e107b9df35dabbd1ede
+  data.tar.gz: 5e4ef4cf71118c020f29b105bb81bb19d753dcb9484f367d0a2caab0b51927e7bcf69e325549bb4e7f44f9f1ec571ffb4223710ca8cc7987dd10eb4750148514

data/.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2026-05-13 07:45:58 UTC using RuboCop version 1.86.1.
+# on 2026-06-04 14:04:56 UTC using RuboCop version 1.86.1.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
@@ -11,30 +11,49 @@ Gemspec/RequiredRubyVersion:
   Exclude:
     - 'glossarist.gemspec'
 
-# Offense count: 3
+# Offense count: 6
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: EnforcedStyle, IndentationWidth.
 # SupportedStyles: with_first_argument, with_fixed_indentation
 Layout/ArgumentAlignment:
   Exclude:
-    - 'lib/glossarist/designation/expression.rb'
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
-# Offense count: 1
+# Offense count: 3
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: EnforcedStyleAlignWith.
 # SupportedStylesAlignWith: either, start_of_block, start_of_line
 Layout/BlockAlignment:
   Exclude:
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
-# Offense count: 1
+# Offense count: 3
 # This cop supports safe autocorrection (--autocorrect).
 Layout/BlockEndNewline:
   Exclude:
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
-# Offense count: 7
+# Offense count: 1
+# This cop supports safe autocorrection (--autocorrect).
+Layout/EmptyLineAfterGuardClause:
+  Exclude:
+    - 'lib/glossarist/section.rb'
+
+# Offense count: 1
+# This cop supports safe autocorrection (--autocorrect).
+Layout/EmptyLines:
+  Exclude:
+    - 'Gemfile'
+
+# Offense count: 2
+# This cop supports safe autocorrection (--autocorrect).
+# Configuration parameters: EnforcedStyle, IndentationWidth.
+# SupportedStyles: special_inside_parentheses, consistent, align_braces
+Layout/FirstHashElementIndentation:
+  Exclude:
+    - 'spec/unit/section_spec.rb'
+
+# Offense count: 8
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: AllowMultipleStyles, EnforcedHashRocketStyle, EnforcedColonStyle, EnforcedLastArgumentHashStyle.
 # SupportedHashRocketStyles: key, separator, table
@@ -42,32 +61,39 @@ Layout/BlockEndNewline:
 # SupportedLastArgumentHashStyles: always_inspect, always_ignore, ignore_implicit, ignore_explicit
 Layout/HashAlignment:
   Exclude:
-    - 'spec/unit/concept_data_spec.rb'
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/dataset_register_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
-# Offense count: 2
+# Offense count: 6
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: Width, EnforcedStyleAlignWith, AllowedPatterns.
 # SupportedStylesAlignWith: start_of_line, relative_to_receiver
 Layout/IndentationWidth:
   Exclude:
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
-# Offense count: 267
+# Offense count: 424
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: Max, AllowHeredoc, AllowURI, AllowQualifiedName, URISchemes, AllowRBSInlineAnnotation, AllowCopDirectives, AllowedPatterns, SplitStrings.
 # URISchemes: http, https
 Layout/LineLength:
   Enabled: false
 
+# Offense count: 1
+# This cop supports safe autocorrection (--autocorrect).
+# Configuration parameters: EnforcedStyle.
+# SupportedStyles: symmetrical, new_line, same_line
+Layout/MultilineHashBraceLayout:
+  Exclude:
+    - 'spec/unit/section_spec.rb'
+
 # Offense count: 7
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: AllowInHeredoc.
 Layout/TrailingWhitespace:
   Exclude:
-    - 'lib/glossarist/designation/expression.rb'
-    - 'spec/unit/concept_data_spec.rb'
-    - 'spec/unit/designations_spec.rb'
+    - 'spec/unit/dataset_register_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
 # Offense count: 1
 # Configuration parameters: AllowedMethods.
@@ -76,6 +102,12 @@ Lint/ConstantDefinitionInBlock:
   Exclude:
     - 'spec/unit/utilities/common_functions_spec.rb'
 
+# Offense count: 1
+# Configuration parameters: IgnoreLiteralBranches, IgnoreConstantBranches, IgnoreDuplicateElseBranch.
+Lint/DuplicateBranch:
+  Exclude:
+    - 'scripts/migrate_dataset.rb'
+
 # Offense count: 1
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: AllowUnusedKeywordArguments, IgnoreEmptyMethods, IgnoreNotImplementedMethods, NotImplementedExceptions.
@@ -84,55 +116,36 @@ Lint/UnusedMethodArgument:
   Exclude:
     - 'lib/glossarist/dataset_validator.rb'
 
-# Offense count: 37
+# Offense count: 85
 # Configuration parameters: AllowedMethods, AllowedPatterns, CountRepeatedAttributes, Max.
 Metrics/AbcSize:
   Enabled: false
 
-# Offense count: 24
+# Offense count: 7
+# Configuration parameters: CountComments, CountAsOne, AllowedMethods, AllowedPatterns.
+# AllowedMethods: refine
+Metrics/BlockLength:
+  Max: 41
+
+# Offense count: 52
 # Configuration parameters: AllowedMethods, AllowedPatterns, Max.
 Metrics/CyclomaticComplexity:
-  Exclude:
-    - 'lib/glossarist/cli/export_command.rb'
-    - 'lib/glossarist/concept_validator.rb'
-    - 'lib/glossarist/designation/expression.rb'
-    - 'lib/glossarist/gcr_metadata.rb'
-    - 'lib/glossarist/gcr_statistics.rb'
-    - 'lib/glossarist/gcr_validator.rb'
-    - 'lib/glossarist/managed_concept.rb'
-    - 'lib/glossarist/reference_extractor.rb'
-    - 'lib/glossarist/reference_resolver.rb'
-    - 'lib/glossarist/resolution_adapter/local.rb'
-    - 'lib/glossarist/schema_migration.rb'
-    - 'lib/glossarist/transforms/concept_to_skos_transform.rb'
-    - 'lib/glossarist/transforms/concept_to_tbx_transform.rb'
-    - 'lib/glossarist/validation/bibliography_index.rb'
+  Enabled: false
 
-# Offense count: 48
+# Offense count: 107
 # Configuration parameters: CountComments, CountAsOne, AllowedMethods, AllowedPatterns.
 Metrics/MethodLength:
-  Max: 42
+  Max: 47
 
-# Offense count: 4
+# Offense count: 5
 # Configuration parameters: CountKeywordArgs, MaxOptionalParameters.
 Metrics/ParameterLists:
   Max: 6
 
-# Offense count: 18
+# Offense count: 38
 # Configuration parameters: AllowedMethods, AllowedPatterns, Max.
 Metrics/PerceivedComplexity:
-  Exclude:
-    - 'lib/glossarist/concept_validator.rb'
-    - 'lib/glossarist/designation/expression.rb'
-    - 'lib/glossarist/gcr_metadata.rb'
-    - 'lib/glossarist/gcr_validator.rb'
-    - 'lib/glossarist/reference_extractor.rb'
-    - 'lib/glossarist/reference_resolver.rb'
-    - 'lib/glossarist/resolution_adapter/local.rb'
-    - 'lib/glossarist/schema_migration.rb'
-    - 'lib/glossarist/transforms/concept_to_skos_transform.rb'
-    - 'lib/glossarist/transforms/concept_to_tbx_transform.rb'
-    - 'lib/glossarist/validation/bibliography_index.rb'
+  Enabled: false
 
 # Offense count: 6
 # Configuration parameters: MinNameLength, AllowNamesEndingInNumbers, AllowedNames, ForbiddenNames.
@@ -141,13 +154,16 @@ Naming/MethodParameterName:
   Exclude:
     - 'lib/glossarist/schema_migration.rb'
 
-# Offense count: 3
-# Configuration parameters: EnforcedStyle, CheckMethodNames, CheckSymbols, AllowedIdentifiers, AllowedPatterns.
-# SupportedStyles: snake_case, normalcase, non_integer
-# AllowedIdentifiers: TLS1_1, TLS1_2, capture3, iso8601, rfc1123_date, rfc822, rfc2822, rfc3339, x86_64
-Naming/VariableNumber:
+# Offense count: 5
+# This cop supports safe autocorrection (--autocorrect).
+# Configuration parameters: EnforcedStyle, ProceduralMethods, FunctionalMethods, AllowedMethods, AllowedPatterns, AllowBracesOnProceduralOneLiners, BracesRequiredMethods.
+# SupportedStyles: line_count_based, semantic, braces_for_chaining, always_braces
+# ProceduralMethods: benchmark, bm, bmbm, create, each_with_object, measure, new, realtime, tap, with_object
+# FunctionalMethods: let, let!, subject, watch
+# AllowedMethods: lambda, proc, it
+Style/BlockDelimiters:
   Exclude:
-    - 'spec/unit/rdf/skos_vocabulary_spec.rb'
+    - 'spec/unit/section_spec.rb'
 
 # Offense count: 6
 # This cop supports safe autocorrection (--autocorrect).
@@ -156,18 +172,16 @@ Naming/VariableNumber:
 Style/FormatStringToken:
   EnforcedStyle: unannotated
 
-# Offense count: 1
-# Configuration parameters: AllowedClasses.
-Style/OneClassPerFile:
+# Offense count: 2
+# This cop supports unsafe autocorrection (--autocorrect-all).
+Style/IdenticalConditionalBranches:
   Exclude:
-    - 'spec/**/*'
-    - 'test/**/*'
-    - 'lib/glossarist.rb'
+    - 'scripts/migrate_dataset.rb'
 
-# Offense count: 4
+# Offense count: 3
 # Configuration parameters: Max.
 Style/SafeNavigationChainLength:
   Exclude:
     - 'lib/glossarist/managed_concept.rb'
-    - 'lib/glossarist/transforms/concept_to_skos_transform.rb'
     - 'lib/glossarist/transforms/concept_to_tbx_transform.rb'
+    - 'lib/glossarist/validation/rules/source_urn_format_rule.rb'

data/glossarist.gemspec

@@ -32,7 +32,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
 
   spec.add_dependency "lutaml-model", "~> 0.8.15"
-  spec.add_dependency "lutaml-store", "~> 0.1.1"
+  spec.add_dependency "lutaml-store", "~> 0.2.0"
   spec.add_dependency "paint", "~> 2.3"
   spec.add_dependency "relaton", ">= 2.0.0", "< 3"
   spec.add_dependency "rubyzip", ">= 2.3", "< 3"

data/lib/glossarist/bibliography_data.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+module Glossarist
+  class BibliographyData < Lutaml::Model::Serializable
+    attribute :shortname, :string, default: -> { "bibliography" }
+    attribute :entries, BibliographyEntry, collection: true,
+                                           initialize_empty: true
+
+    key_value do
+      map nil, to: :entries,
+               with: { from: :entries_from_hash, to: :entries_to_hash }
+    end
+
+    def find(citation_key)
+      entries.find { |e| e.citation_key == citation_key }
+    end
+
+    def keys
+      entries.map(&:citation_key)
+    end
+
+    def [](citation_key)
+      entry = find(citation_key)
+      entry&.data
+    end
+
+    def entries_from_hash(model, value)
+      return unless value.is_a?(Hash)
+
+      model.entries = value.map do |key, data|
+        BibliographyEntry.new(citation_key: key, data: data || {})
+      end
+    end
+
+    def entries_to_hash(model, doc)
+      model.entries.each do |entry|
+        doc[entry.citation_key] = entry.data
+      end
+    end
+  end
+end

data/lib/glossarist/bibliography_entry.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Glossarist
+  class BibliographyEntry < Lutaml::Model::Serializable
+    attribute :citation_key, :string
+    attribute :data, :hash, default: -> { {} }
+
+    key_value do
+      map "citation_key", to: :citation_key
+      map "data", to: :data
+    end
+  end
+end

data/lib/glossarist/citation.rb

@@ -56,8 +56,14 @@ module Glossarist
 
       doc["locality"] = {}
       doc["locality"]["type"] = model.locality.type
-      doc["locality"]["reference_from"] = model.locality.reference_from if model.locality.reference_from
-      doc["locality"]["reference_to"] = model.locality.reference_to if model.locality.reference_to
+      if model.locality.reference_from
+        doc["locality"]["reference_from"] =
+          model.locality.reference_from
+      end
+      if model.locality.reference_to
+        doc["locality"]["reference_to"] =
+          model.locality.reference_to
+      end
     end
   end
 end

data/lib/glossarist/cli/export_command.rb

@@ -74,14 +74,18 @@ module Glossarist
 
       def export_jsonld(concepts, name, output_dir)
         require "glossarist/transforms/concept_to_gloss_transform"
-        transform = Transforms::ConceptToGlossTransform.new(nil, transform_options)
-        File.write(File.join(output_dir, "#{name}.jsonld"), transform.to_jsonld(concepts))
+        transform = Transforms::ConceptToGlossTransform.new(nil,
+                                                            transform_options)
+        File.write(File.join(output_dir, "#{name}.jsonld"),
+                   transform.to_jsonld(concepts))
       end
 
       def export_turtle(concepts, name, output_dir)
         require "glossarist/transforms/concept_to_gloss_transform"
-        transform = Transforms::ConceptToGlossTransform.new(nil, transform_options)
-        File.write(File.join(output_dir, "#{name}.ttl"), transform.to_turtle(concepts))
+        transform = Transforms::ConceptToGlossTransform.new(nil,
+                                                            transform_options)
+        File.write(File.join(output_dir, "#{name}.ttl"),
+                   transform.to_turtle(concepts))
       end
 
       def export_tbx(concepts, name, output_dir)
@@ -95,7 +99,8 @@ module Glossarist
         require "glossarist/transforms/concept_to_gloss_transform"
         File.open(File.join(output_dir, "#{name}.jsonl"), "w") do |f|
           concepts.each do |concept|
-            transform = Transforms::ConceptToGlossTransform.new(concept, transform_options)
+            transform = Transforms::ConceptToGlossTransform.new(concept,
+                                                                transform_options)
             f.write(transform.to_jsonl_line)
             f.write("\n")
           end

data/lib/glossarist/cli/validate_command.rb

@@ -37,7 +37,8 @@ module Glossarist
         filled = (current.to_f / total * bar_width).round
         bar = "#{'█' * filled}#{'░' * (bar_width - filled)}"
 
-        $stderr.print "\r  #{Paint['Validating', :bold]} #{bar} #{current}/#{total} (#{pct}%)"
+        $stderr.print "\r  #{Paint['Validating',
+                                   :bold]} #{bar} #{current}/#{total} (#{pct}%)"
         $stderr.flush
       end
 
@@ -96,18 +97,33 @@ module Glossarist
         msg_col = 21
 
         puts "    #{label}  #{code}  #{issue.message}"
-        puts "#{' ' * msg_col}#{Paint[issue.suggestion, :green]}" if issue.suggestion
+        if issue.suggestion
+          puts "#{' ' * msg_col}#{Paint[issue.suggestion,
+                                        :green]}"
+        end
       end
 
       def print_summary_line(result)
         error_count = result.issues.count(&:error?)
         warning_count = result.issues.count(&:warning?)
 
-        status = error_count.positive? ? Paint["INVALID", :red, :bold] : Paint["VALID", :green, :bold]
+        status = if error_count.positive?
+                   Paint["INVALID", :red,
+                         :bold]
+                 else
+                   Paint["VALID", :green,
+                         :bold]
+                 end
 
         details = []
-        details << Paint["#{error_count} error(s)", :red] if error_count.positive?
-        details << Paint["#{warning_count} warning(s)", :yellow] if warning_count.positive?
+        if error_count.positive?
+          details << Paint["#{error_count} error(s)",
+                           :red]
+        end
+        if warning_count.positive?
+          details << Paint["#{warning_count} warning(s)",
+                           :yellow]
+        end
 
         puts "  #{status}  #{details.join(', ')}"
       end

data/lib/glossarist/collection.rb

@@ -15,8 +15,8 @@ module Glossarist
       @index = {}
     end
 
-    def each(&block)
-      @index.each_value(&block)
+    def each(&)
+      @index.each_value(&)
     end
 
     # Returns concept with given ID, if it is present in collection, or +nil+

data/lib/glossarist/collections/bibliography_collection.rb

@@ -7,7 +7,7 @@ module Glossarist
     class BibliographyCollection < Relaton::Db
       def initialize(_concepts, global_cache, local_cache)
         @version_mismatch = check_cache_version(local_cache) ||
-                            check_cache_version(global_cache)
+          check_cache_version(global_cache)
         super(global_cache, local_cache)
       end
 
@@ -40,6 +40,7 @@ module Glossarist
         concepts.each do |concept|
           concept.default_lang.sources.each do |source|
             next if source.origin.ref.nil?
+
             ref_text = source.origin.ref.source
             next if ref_text.nil?
 

data/lib/glossarist/collections/collection.rb

@@ -18,8 +18,8 @@ module Glossarist
         @collection << @klass.new(object)
       end
 
-      def each(&block)
-        @collection.each(&block)
+      def each(&)
+        @collection.each(&)
       end
 
       def empty?

data/lib/glossarist/collections/localization_collection.rb

@@ -27,12 +27,12 @@ module Glossarist
         to_a
       end
 
-      def each_key(&block)
-        keys.each(&block)
+      def each_key(&)
+        keys.each(&)
       end
 
-      def each_value(&block)
-        values.each(&block)
+      def each_value(&)
+        values.each(&)
       end
     end
   end

data/lib/glossarist/concept_collector.rb

@@ -136,25 +136,25 @@ module Glossarist
         end
       end
 
-      def each_v2_concept(dir, &block)
+      def each_v2_concept(dir, &)
         if v2_flat_concepts?(dir)
-          each_grouped_v2_concepts(File.join(dir, "concepts"), &block)
+          each_grouped_v2_concepts(File.join(dir, "concepts"), &)
         else
           v2_dir = File.join(dir, "geolexica-v2")
           if File.directory?(File.join(v2_dir, "concepts"))
-            each_managed_concept(v2_dir, &block)
+            each_managed_concept(v2_dir, &)
           else
-            each_grouped_v2_concepts(v2_dir, &block)
+            each_grouped_v2_concepts(v2_dir, &)
           end
         end
       end
 
-      def each_grouped_v2_concepts(v2_dir, &block)
+      def each_grouped_v2_concepts(v2_dir, &)
         collection = ManagedConceptCollection.new
         manager = ConceptManager.new(path: v2_dir)
         manager.version = detect_schema_version(v2_dir)
         manager.load_from_files(collection: collection)
-        collection.each(&block)
+        collection.each(&)
       end
 
       def collect_grouped_v2_concepts(v2_dir)

data/lib/glossarist/concept_document.rb

@@ -2,6 +2,7 @@
 
 module Glossarist
   class ConceptDocument < Lutaml::Model::Serializable
+    attribute :id, :string
     attribute :concept, ManagedConcept
     attribute :localizations, LocalizedConcept, collection: true
 
@@ -9,7 +10,7 @@ module Glossarist
       sequence do
         map_document 0, to: :concept, type: ManagedConcept
         map_document 1.., to: :localizations, type: LocalizedConcept,
-                                           collection: true
+                          collection: true
       end
     end
 

data/lib/glossarist/concept_manager.rb

@@ -132,13 +132,12 @@ module Glossarist
       if v1_collection?
         File.join(path, "concept-*.{yaml,yml}")
       else
-        # normal v2 collection
-        concepts_glob = File.join(path, "concept", "*.{yaml,yml}")
-        if Dir.glob(concepts_glob).empty?
-          # multiple content YAML files
-          concepts_glob = File.join(path, "*.{yaml,yml}")
-        end
-        concepts_glob
+        candidates = [
+          File.join(path, "concept", "*.{yaml,yml}"),
+          File.join(path, "concepts", "*.{yaml,yml}"),
+          File.join(path, "*.{yaml,yml}"),
+        ]
+        candidates.find { |g| !Dir.glob(g).empty? }
       end
     end
 

data/lib/glossarist/concept_reference.rb

@@ -26,6 +26,10 @@ module Glossarist
       new(concept_id: concept_id, ref_type: "domain")
     end
 
+    def self.section(concept_id)
+      new(concept_id: concept_id, ref_type: "section")
+    end
+
     def local?
       %w[local designation].include?(ref_type) ||
         (ref_type.nil? && (source.nil? || source.empty?))

data/lib/glossarist/concept_set.rb

@@ -36,7 +36,7 @@ module Glossarist
     end
 
     def to_latex_from_file(entries_file)
-      File.readlines(entries_file).map do |concept_name|
+      File.readlines(entries_file).filter_map do |concept_name|
         concept = concept_map[concept_name.strip.downcase]
 
         if concept.nil?
@@ -44,7 +44,7 @@ module Glossarist
         else
           latex_template(concept)
         end
-      end.compact.join("\n")
+      end.join("\n")
     end
 
     def read_concepts(concepts)
@@ -72,9 +72,9 @@ module Glossarist
     end
 
     def concept_map
-      @concept_map ||= concepts.managed_concepts.map do |concept|
+      @concept_map ||= concepts.managed_concepts.to_h do |concept|
         [concept.default_designation.downcase, concept]
-      end.to_h
+      end
     end
   end
 end