glossarist 2.6.1 → 2.6.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b4e48f1d237829b78cd052e39afdd7c952c468be894e25856f79f93471ea2dc6
-  data.tar.gz: f6502f7012e30138193c4c28eee2ab551020edf27ff0c6e08cc39939e5d0c0c3
+  metadata.gz: a4a05468d25c9ac0d13c748454dc98d7dc031718fe98d298c3db89d6874963a7
+  data.tar.gz: 203b26205a85b9593942d7d6095dab85faa93ed1e99b9f474f20c67a77e61c5d
 SHA512:
-  metadata.gz: bcfb39fe52138d23d489cfbff45f62b0f00161d3183ef66dbb5718fcbc30694ef5e426bc9f7797329480dddb01a7c96b8f6e1dfc43e392032441bb62487c723a
-  data.tar.gz: 289911c5a5851389cb5e37fbc07bd45825b00cdd51fa8cac5679168af3ec9e2b7313f6afbd7093fb93b35384bfc7318dcb97a22d72869f78a321883f4ab299a4
+  metadata.gz: dc177b1d927f7b309cb47fac5158c151001202ba0332b35f0eaf87a8e1e2d9eda400dd077145b10d4acc5ee968676f39dd95172a32ad16e6078ff3357c299317
+  data.tar.gz: eb60cfa90dd26008e2287cbafb35aaeb4e48bb60c120cc254fba024d945f4a6cdb41e9b51a2b8c68a30ac6e31e7e40f38243ec44437a5995126eba47f47110dd

data/.rubocop_todo.yml

@@ -1,18 +1,11 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2026-05-06 18:35:41 UTC using RuboCop version 1.86.1.
+# on 2026-05-07 13:57:54 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
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 5
-# This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: TreatCommentsAsGroupSeparators, ConsiderPunctuation.
-Bundler/OrderedGems:
-  Exclude:
-    - 'Gemfile'
-
 # Offense count: 1
 Gemspec/RequiredRubyVersion:
   Exclude:
@@ -20,25 +13,48 @@ Gemspec/RequiredRubyVersion:
 
 # Offense count: 1
 # This cop supports safe autocorrection (--autocorrect).
-Layout/ClosingParenthesisIndentation:
+# Configuration parameters: EnforcedStyle, IndentationWidth.
+# SupportedStyles: with_first_argument, with_fixed_indentation
+Layout/ArgumentAlignment:
   Exclude:
-    - 'spec/unit/concept_set_spec.rb'
+    - 'spec/unit/gcr_package_spec.rb'
 
-# Offense count: 1
+# Offense count: 2
 # This cop supports safe autocorrection (--autocorrect).
-# Configuration parameters: EnforcedStyle, IndentationWidth.
-# SupportedStyles: consistent, consistent_relative_to_receiver, special_for_inner_method_call, special_for_inner_method_call_in_parentheses
-Layout/FirstArgumentIndentation:
+# Configuration parameters: EnforcedStyleAlignWith.
+# SupportedStylesAlignWith: either, start_of_block, start_of_line
+Layout/BlockAlignment:
+  Exclude:
+    - 'lib/glossarist/gcr_validator.rb'
+
+# Offense count: 2
+# This cop supports safe autocorrection (--autocorrect).
+Layout/BlockEndNewline:
+  Exclude:
+    - 'lib/glossarist/gcr_validator.rb'
+
+# Offense count: 4
+# This cop supports safe autocorrection (--autocorrect).
+# Configuration parameters: Width, EnforcedStyleAlignWith, AllowedPatterns.
+# SupportedStylesAlignWith: start_of_line, relative_to_receiver
+Layout/IndentationWidth:
   Exclude:
-    - 'spec/unit/concept_set_spec.rb'
+    - 'lib/glossarist/gcr_validator.rb'
 
-# Offense count: 202
+# Offense count: 214
 # 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: AllowInHeredoc.
+Layout/TrailingWhitespace:
+  Exclude:
+    - 'spec/unit/gcr_package_spec.rb'
+
 # Offense count: 1
 # Configuration parameters: AllowedMethods.
 # AllowedMethods: enums
@@ -54,7 +70,7 @@ Lint/UnusedMethodArgument:
   Exclude:
     - 'lib/glossarist/dataset_validator.rb'
 
-# Offense count: 23
+# Offense count: 24
 # Configuration parameters: AllowedMethods, AllowedPatterns, CountRepeatedAttributes, Max.
 Metrics/AbcSize:
   Exclude:
@@ -73,12 +89,6 @@ Metrics/AbcSize:
     - 'lib/glossarist/utilities/uuid.rb'
     - 'spec/unit/concept_collector_spec.rb'
 
-# Offense count: 1
-# Configuration parameters: CountComments, CountAsOne, AllowedMethods, AllowedPatterns.
-# AllowedMethods: refine
-Metrics/BlockLength:
-  Max: 28
-
 # Offense count: 19
 # Configuration parameters: AllowedMethods, AllowedPatterns, Max.
 Metrics/CyclomaticComplexity:
@@ -101,7 +111,7 @@ Metrics/CyclomaticComplexity:
 Metrics/MethodLength:
   Max: 42
 
-# Offense count: 4
+# Offense count: 3
 # Configuration parameters: CountKeywordArgs, MaxOptionalParameters.
 Metrics/ParameterLists:
   Max: 6
@@ -135,6 +145,17 @@ Naming/VariableNumber:
   Exclude:
     - 'spec/unit/rdf/skos_vocabulary_spec.rb'
 
+# Offense count: 3
+# 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:
+    - 'lib/glossarist/gcr_validator.rb'
+
 # Offense count: 6
 # This cop supports safe autocorrection (--autocorrect).
 # Configuration parameters: MaxUnannotatedPlaceholdersAllowed, Mode, AllowedMethods, AllowedPatterns.

data/lib/glossarist/concept_collector.rb

@@ -41,13 +41,35 @@ module Glossarist
       def v1_concepts?(dir)
         concepts_dir = File.join(dir, "concepts")
         File.directory?(concepts_dir) &&
+          !v2_flat_concepts?(dir) &&
+          !managed_concepts?(dir) &&
           Dir.glob(File.join(concepts_dir, "*.yaml")).any? do |f|
             V1::Concept.from_file(f)&.termid?
           end
       end
 
       def v2_concepts?(dir)
-        File.directory?(File.join(dir, "geolexica-v2"))
+        File.directory?(File.join(dir, "geolexica-v2")) ||
+          v2_flat_concepts?(dir)
+      end
+
+      def v2_flat_concepts?(dir)
+        return false if managed_concepts?(dir)
+
+        concepts_dir = File.join(dir, "concepts")
+        return false unless File.directory?(concepts_dir)
+
+        Dir.glob(File.join(concepts_dir, "*.yaml")).first(5).any? do |f|
+          v2_flat_concept_file?(f)
+        end
+      end
+
+      def v2_flat_concept_file?(path)
+        raw = File.read(path, encoding: "utf-8")
+        doc = ConceptDocument.from_yamls(raw)
+        !!doc.concept&.data&.id
+      rescue StandardError
+        false
       end
 
       def managed_concepts?(dir)
@@ -74,24 +96,32 @@ module Glossarist
       end
 
       def collect_v2_concepts(dir)
-        v2_dir = File.join(dir, "geolexica-v2")
-        if File.directory?(File.join(v2_dir, "concepts"))
-          collect_managed_concepts(v2_dir)
+        if v2_flat_concepts?(dir)
+          collect_grouped_v2_concepts(File.join(dir, "concepts"))
         else
-          collect_grouped_v2_concepts(v2_dir)
+          v2_dir = File.join(dir, "geolexica-v2")
+          if File.directory?(File.join(v2_dir, "concepts"))
+            collect_managed_concepts(v2_dir)
+          else
+            collect_grouped_v2_concepts(v2_dir)
+          end
         end
       end
 
       def each_v2_concept(dir, &block)
-        v2_dir = File.join(dir, "geolexica-v2")
-        if File.directory?(File.join(v2_dir, "concepts"))
-          each_managed_concept(v2_dir, &block)
+        if v2_flat_concepts?(dir)
+          each_grouped_v2_concepts(File.join(dir, "concepts"), &block)
         else
-          each_grouped_v2_concept(v2_dir, &block)
+          v2_dir = File.join(dir, "geolexica-v2")
+          if File.directory?(File.join(v2_dir, "concepts"))
+            each_managed_concept(v2_dir, &block)
+          else
+            each_grouped_v2_concepts(v2_dir, &block)
+          end
         end
       end
 
-      def each_grouped_v2_concept(v2_dir, &block)
+      def each_grouped_v2_concepts(v2_dir, &block)
         collection = ManagedConceptCollection.new
         manager = ConceptManager.new(path: v2_dir)
         manager.load_from_files(collection: collection)

data/lib/glossarist/gcr_package.rb

@@ -2,6 +2,7 @@
 
 require "zip"
 require "fileutils"
+require "pathname"
 
 module Glossarist
   class GcrPackage
@@ -14,12 +15,18 @@ module Glossarist
 
     KNOWN_COMPILED_FORMATS = COMPILED_EXTENSIONS.keys.freeze
 
-    attr_reader :zip_path, :metadata, :concepts
+    DATASET_ASSETS = [
+      { path: "bibliography.yaml", type: :file, attr: :bibliography },
+      { path: "images", type: :directory },
+    ].freeze
+
+    attr_reader :zip_path, :metadata, :concepts, :bibliography
 
     def initialize(zip_path)
       @zip_path = zip_path
       @metadata = nil
       @concepts = []
+      @bibliography = nil
     end
 
     def self.create(concepts:, metadata:, output_path:, register_data: nil,
@@ -66,12 +73,43 @@ module Glossarist
       end
     end
 
+    def self.each_dataset_asset(source_dir)
+      base = Pathname.new(source_dir)
+      DATASET_ASSETS.each do |asset|
+        path = File.join(source_dir, asset[:path])
+        case asset[:type]
+        when :file
+          yield_file_asset(path, asset[:path]) { |*a| yield(*a) }
+        when :directory
+          yield_directory_assets(path, base) { |*a| yield(*a) }
+        end
+      end
+    end
+
+    def self.yield_file_asset(path, entry_name)
+      return unless File.exist?(path)
+
+      yield entry_name, File.binread(path)
+    end
+
+    def self.yield_directory_assets(dir_path, base_path)
+      return unless File.directory?(dir_path)
+
+      Dir.glob(File.join(dir_path, "**", "*")).each do |file|
+        next unless File.file?(file)
+
+        relative = Pathname.new(file).relative_path_from(base_path).to_s
+        yield relative, File.binread(file)
+      end
+    end
+
     def validate
       GcrValidator.new.validate(@zip_path)
     end
 
-    def write(concepts, metadata, register_data, compiled_formats: [],
-              shortname: nil, **opts)
+    def write(concepts, metadata, register_data, # rubocop:disable Metrics/ParameterLists
+              compiled_formats: [],
+              shortname: nil, source_dir: nil, **opts)
       Zip::File.open(@zip_path, create: true) do |zf|
         zf.get_output_stream("metadata.yaml") do |f|
           f.write(metadata.to_yaml)
@@ -87,6 +125,12 @@ module Glossarist
           write_concept(zf, mc)
         end
 
+        if source_dir
+          self.class.each_dataset_asset(source_dir) do |name, content|
+            zf.get_output_stream(name) { |f| f.write(content) }
+          end
+        end
+
         if compiled_formats.any?
           write_compiled(zf, concepts, compiled_formats, shortname: shortname,
                                                          **opts)
@@ -94,29 +138,51 @@ module Glossarist
       end
     end
 
-    def write_concept(zip_file, concept)
-      termid = concept.data.id.to_s
-      doc = ConceptDocument.from_managed_concept(concept)
-      zip_file.get_output_stream("concepts/#{termid}.yaml") do |f|
-        f.write(doc.to_yamls)
-      end
-    end
-
     def read
       @concepts = []
 
       Zip::File.open(@zip_path) do |zf|
-        if (entry = zf.find_entry("metadata.yaml"))
-          @metadata = GcrMetadata.from_yaml(entry.get_input_stream.read)
-        end
+        read_metadata(zf)
+        read_file_assets(zf)
+        read_concepts(zf)
+      end
+    end
 
-        zf.entries.each do |entry|
-          next unless entry.name.start_with?("concepts/") && entry.name.end_with?(".yaml")
+    def read_metadata(zip_file)
+      entry = zip_file.find_entry("metadata.yaml")
+      return unless entry
 
-          raw = entry.get_input_stream.read
-          doc = ConceptDocument.from_yamls(raw)
-          @concepts << doc.to_managed_concept
-        end
+      @metadata = GcrMetadata.from_yaml(entry.get_input_stream.read)
+    end
+
+    def read_file_assets(zip_file)
+      DATASET_ASSETS.each do |asset|
+        next unless asset[:type] == :file && asset[:attr]
+
+        entry = zip_file.find_entry(asset[:path])
+        next unless entry
+
+        instance_variable_set("@#{asset[:attr]}", entry.get_input_stream.read)
+      end
+    end
+
+    def read_concepts(zip_file)
+      zip_file.entries.each do |entry|
+        next unless entry.name.start_with?("concepts/") && entry.name.end_with?(".yaml")
+
+        raw = entry.get_input_stream.read
+        doc = ConceptDocument.from_yamls(raw)
+        @concepts << doc.to_managed_concept
+      end
+    end
+
+    private
+
+    def write_concept(zip_file, concept)
+      termid = concept.data.id.to_s
+      doc = ConceptDocument.from_managed_concept(concept)
+      zip_file.get_output_stream("concepts/#{termid}.yaml") do |f|
+        f.write(doc.to_yamls)
       end
     end
 
@@ -206,6 +272,7 @@ compiled_formats: [], **opts)
           output_path: File.expand_path(output),
           compiled_formats: compiled_formats,
           shortname: shortname,
+          source_dir: dir,
           **opts,
         )
       end
@@ -219,7 +286,7 @@ compiled_formats: [], **opts)
         concept_count = 0
         languages = Set.new
 
-        Zip::OutputStream.open(output_path) do |zos|
+        Zip::OutputStream.open(output_path) do |zos| # rubocop:disable Metrics/BlockLength
           if register_data
             zos.put_next_entry("register.yaml")
             zos.write(register_data.to_yaml)
@@ -253,6 +320,11 @@ compiled_formats: [], **opts)
                                               register_data: register_data, **opts)
           zos.put_next_entry("metadata.yaml")
           zos.write(metadata.to_yaml)
+
+          each_dataset_asset(dir) do |name, content|
+            zos.put_next_entry(name)
+            zos.write(content)
+          end
         end
 
         new(output_path)

data/lib/glossarist/gcr_validator.rb

@@ -13,27 +13,8 @@ module Glossarist
       end
 
       begin
-        Zip::File.open(zip_path) do |zf|
-          unless zf.find_entry("metadata.yaml")
-            result.add_error("Missing metadata.yaml")
-            return result
-          end
-
-          metadata = GcrMetadata.from_yaml(
-            zf.find_entry("metadata.yaml").get_input_stream.read,
-          )
-          validate_metadata(metadata, result)
-
-          concept_entries = zf.entries.select do |e|
-            e.name.start_with?("concepts/") && e.name.end_with?(".yaml")
-          end
-          if concept_entries.empty?
-            result.add_error("No concept files found in concepts/")
-          end
-
-          concept_entries.each do |entry|
-            validate_concept_entry(entry, metadata, result)
-          end
+        Zip::File.open(zip_path) do |zip_file|
+          validate_zip_contents(zip_file, result)
         end
       rescue StandardError => e
         result.add_error("Failed to read ZIP: #{e.message}")
@@ -44,6 +25,31 @@ module Glossarist
 
     private
 
+    def validate_zip_contents(zip_file, result) # rubocop:disable Metrics/AbcSize
+      unless zip_file.find_entry("metadata.yaml")
+        result.add_error("Missing metadata.yaml")
+        return
+      end
+
+      metadata = GcrMetadata.from_yaml(
+        zip_file.find_entry("metadata.yaml").get_input_stream.read,
+      )
+      validate_metadata(metadata, result)
+
+      concept_entries = zip_file.entries.select do |e|
+        e.name.start_with?("concepts/") && e.name.end_with?(".yaml")
+      end
+      if concept_entries.empty?
+        result.add_error("No concept files found in concepts/")
+      end
+
+      concept_entries.each do |entry|
+        validate_concept_entry(entry, metadata, result)
+      end
+
+      validate_assets(zip_file, result)
+    end
+
     def validate_metadata(metadata, result)
       unless metadata&.concept_count
         result.add_error("metadata.yaml missing required fields (concept_count)")
@@ -94,5 +100,36 @@ module Glossarist
         result.add_warning("#{entry.name}: no concept URI (data.uri) and no concept_uri_template or uri_prefix in metadata")
       end
     end
+
+    def validate_assets(zip_file, result)
+      GcrPackage::DATASET_ASSETS.each do |asset|
+        case asset[:type]
+        when :file
+          validate_file_asset_entry(zip_file, asset[:path], result)
+        when :directory
+          validate_directory_asset(zip_file, asset[:path], result)
+        end
+      end
+    end
+
+    def validate_file_asset_entry(zip_file, path, result)
+      entry = zip_file.find_entry(path)
+      return unless entry
+
+      YAML.safe_load(entry.get_input_stream.read)
+    rescue Psych::SyntaxError => e
+      result.add_error("#{path}: invalid YAML at line #{e.line}: #{e.message}")
+    end
+
+    def validate_directory_asset(zip_file, dir_path, result)
+      dir_entries = zip_file.entries.select do |e|
+        e.name.start_with?("#{dir_path}/")
+      end
+      return unless dir_entries.any? && dir_entries.all? do |e|
+        e.name.end_with?("/")
+      end
+
+      result.add_warning("#{dir_path}/ directory is empty")
+    end
   end
 end

data/lib/glossarist/version.rb

@@ -4,5 +4,5 @@
 #
 
 module Glossarist
-  VERSION = "2.6.1"
+  VERSION = "2.6.2"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: glossarist
 version: !ruby/object:Gem::Version
-  version: 2.6.1
+  version: 2.6.2
 platform: ruby
 authors:
 - Ribose
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-05-06 00:00:00.000000000 Z
+date: 2026-05-07 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: lutaml-model