Updated Ruby RDF vocabulary handling.

Reworked vocabulary registration to work with recent upstream changes.
Updated local vocabularies.
Added a couple social vocabularies that we will be using in the future.
Added some tests (per issue #88).

commands/gen-vocabs.rb

@@ -30,7 +30,7 @@ def run
     end
 
     vocabs.each do |id, v|
-      next if v[:alias]
+      next if v[:alias] || v[:skip]
 
       puts "Generate lib/vocabs/#{id}.rb"
 
@@ -39,23 +39,24 @@ def run
         File.open("lib/vocabs/#{id}.rb_p", 'w') { |f| f.write v[:patch] }
         cmd += " patch --patch-file lib/vocabs/#{id}.rb_p"
       end
-      cmd += " serialize --uri '#{v[:uri]}' --output-format vocabulary"
+      cmd += " serialize --uri '#{v[:uri]}' --output-format vocabulary --ordered"
       cmd += " --module-name #{v.fetch(:module_name, 'RDF::Vocab')}"
       cmd += " --class-name #{v[:class_name] || id.to_s.upcase}"
       cmd += ' --strict' if v.fetch(:strict, true)
-      cmd += " --extra #{URI.encode(v[:extra].to_json)}" if v[:extra]
+      cmd += ' --noDoc'
+      cmd += " --extra #{URI.encode_www_form_component(v[:extra].to_json)}" if v[:extra]
       cmd += " -o lib/vocabs/#{id}.rb_t"
       cmd += ' "' + v.fetch(:source, v[:uri]) + '"'
 
       puts "  #{cmd}"
 
       begin
-        `#{cmd} && mv lib/vocabs/#{id}.rb_t lib/vocabs/#{id}.rb`
+        %x{#{cmd} && sed 's/\r//g' lib/vocabs/#{id}.rb_t > lib/vocabs/#{id}.rb}
       rescue
         require 'English'
         puts "Failed to load #{id}: #{$ERROR_INFO.message}"
       ensure
-        `rm -f lib/vocabs/#{id}.rb_t lib/vocabs/#{id}.rb_p`
+        %x{rm -f lib/vocabs/#{id}.rb_t lib/vocabs/#{id}.rb_p}
       end
     end
   end
@@ -88,9 +89,9 @@ def self.env_for_site(site)
     view_context = view_context_for(site)
 
     {
-      items: Nanoc::ItemCollectionWithRepsView.new(site.items, view_context),
-      layouts: Nanoc::LayoutCollectionView.new(site.layouts, view_context),
-      config: Nanoc::ConfigView.new(site.config, view_context),
+      items: Nanoc::Core::ItemCollectionWithRepsView.new(site.items, view_context),
+      layouts: Nanoc::Core::LayoutCollectionView.new(site.layouts, view_context),
+      config: Nanoc::Core::ConfigView.new(site.config, view_context),
     }
   end
 end

etc/vocabs_additional.yaml

@@ -15,10 +15,6 @@ c4o:
   uri: "http://purl.org/spar/c4o/"
   source: "http://www.sparontologies.net/ontologies/c4o/source.ttl"
 
-cert:
-  uri: "https://www.w3.org/ns/auth/cert#"
-  source: "http://www.w3.org/ns/auth/cert.n3"
-
 co:
   uri: "http://purl.org/co/"
   source: "var/vocabs/collections.owl"
@@ -31,10 +27,6 @@ ctag:
   uri: "http://commontag.org/ns#"
   source: "var/vocabs/ctag_2009-06-08.n3"
 
-dbo:
-  uri: "http://dbpedia.org/ontology/"
-  source: "http://mappings.dbpedia.org/server/ontology/dbpedia.owl"
-
 dbr:
   uri: "http://dbpedia.org/resource/"
   strict: false
@@ -65,25 +57,25 @@ essglobal:
   uri: "http://purl.org/essglobal/standard/activities/"
   source: "http://www.maltas.org/ess/standard/activities.skos"
   strict: false
-  class_name: "ESSGLOBALActivities"
+  class_name: "ESSGlobalActivities"
 
 "essglobal_legalform":
   uri: "http://purl.org/essglobal/standard/legal-form/"
   source: "http://www.maltas.org/ess/standard/legal-form.skos"
   strict: false
-  class_name: "ESSGLOBALLegalform"
+  class_name: "ESSGlobalLegalform"
 
 "essglobal_qualifiers":
   uri: "http://purl.org/essglobal/standard/qualifiers/"
   source: "http://www.maltas.org/ess/standard/qualifiers.skos"
   strict: false
-  class_name: "ESSGLOBALQualifiers"
+  class_name: "ESSGlobalQualifiers"
 
 "essglobal_typeoflabour":
   uri: "http://purl.org/essglobal/standard/type-of-labour/"
   source: "http://www.maltas.org/ess/standard/type-of-labour.skos"
   strict: false
-  class_name: "ESSGLOBALTypeoflabour"
+  class_name: "ESSGlobalTypeoflabour"
 
 fabio:
   uri: "http://purl.org/spar/fabio/"
@@ -111,9 +103,6 @@ profile:
 rov:
   uri: "http://www.w3.org/ns/regorg#"
 
-sioct:
-  uri: "http://rdfs.org/sioc/types#"
-
 vann:
   uri: "http://purl.org/vocab/vann/"
   source: "var/vocabs/vann.ttl"
@@ -133,16 +122,17 @@ pentandra:
   uri: <%= "#{@config[:base_url]}#{@config[:company_root]}#" %>
   source: <%= "#{@config[:output_dir]}#{@config[:static_root]}#{@config[:company_root]}/index.html" %>
   strict: false
+  class_name: Pentandra
 
 "pentandra_blog":
   uri: <%= "#{@config[:base_url]}#{@config[:blog_root]}#" %>
   source: <%= "#{@config[:output_dir]}#{@config[:static_root]}#{@config[:blog_root]}/index.html" %>
   strict: false
-  class_name: PENTANDRABlog
+  class_name: PentandraBlog
 
 "pentandra_website":
   uri: <%= "#{@config[:base_url]}#" %>
   source: <%= "#{@config[:output_dir]}#{@config[:static_root]}/index.html" %>
   strict: false
-  class_name: PENTANDRAWebsite
+  class_name: PentandraWebsite
 ...

layouts/blog/blogpost.erb

@@ -1,5 +1,5 @@
 <% render('/default.*') do %>
-  <article class="group h-entry" prefix="<%= prefix_mappings_for(:sioct) %>" resource="#blogpost" typeof="cc:Work schema:BlogPosting sioct:BlogPost ctag:TaggedContent" property="schema:mainEntity">
+  <article class="group h-entry" prefix="<%= prefix_mappings_for(:sioctypes) %>" resource="#blogpost" typeof="cc:Work schema:BlogPosting sioctypes:BlogPost ctag:TaggedContent" property="schema:mainEntity">
     <header class="section">
       <h1 class="p-name" property="dc:title schema:headline" id="title"><%= @item[:title] %></h1>
       <div class="group meta" id="document-context">

layouts/blog/note.erb

@@ -1,5 +1,5 @@
 <% render('/default.*') do %>
-  <article class="group h-entry" prefix="<%= prefix_mappings_for(:sioct) %>" resource="#blogpost" typeof="cc:Work schema:BlogPosting sioct:BlogPost ctag:TaggedContent" property="schema:mainEntity">
+  <article class="group h-entry" prefix="<%= prefix_mappings_for(:sioctypes) %>" resource="#blogpost" typeof="cc:Work schema:BlogPosting sioctypes:BlogPost ctag:TaggedContent" property="schema:mainEntity">
     <header class="section">
       <h1 class="p-name" property="dc:title schema:headline" id="title"><%= @item[:title] %></h1>
       <div class="group meta" id="document-context">

layouts/partials/vocab_metadata.erb

@@ -1,5 +1,5 @@
 <ul class="none">
-  <% vocabularies_for(group).each do |vocab| %>
+  <% vocabularies_for(category).each do |vocab| %>
     <li>
       <dl resource="<%= h vocab.fetch(:vocab_uri) %>" property="void:vocabulary" typeof="voaf:Vocabulary">
         <dt><a property="vann:preferredNamespaceUri" content="<%= h vocab.fetch(:namespace_uri) %>" href="<%= h vocab.fetch(:vocab_uri) %>"><code class="prefix" property="vann:preferredNamespacePrefix"><%= h vocab.fetch(:prefix) %></code></a>: <span class="foreign" property="dc:title"><%= h vocab.fetch(:title) %></span></dt>

layouts/specifications/vocabulary.haml

@@ -13,7 +13,7 @@
     %title
       = ont["rdfs:label"].first['@value']
     %link{rel: "stylesheet", href: "/css/specifications.css"}
-  - pfx = RDF::Vocabulary.find(ont['@id']).__name__.split(':').last.downcase
+  - pfx = ont['vann:preferredNamespacePrefix']&.first['@value'] || RDF::Vocabulary.find(ont['@id']).__name__.split(':').last.downcase
   - pfxs = prefixes.inject([]) {|m, (k,v)| m << "#{k}: #{v}"}.join(' ')
   - typeof = ont['@type'].map {|t| RDF::URI(t).pname}.join(' ')
   %body{resource: ont['@id'], typeof: typeof, prefix: pfxs}

lib/data_sources/vocabularies.rb

@@ -10,22 +10,43 @@ module DataSources
     class Vocabularies < Nanoc::DataSource
       identifier :vocabularies
 
+      class UnknownVocabularyPrefix < ::Nanoc::Core::Error
+        def initialize(prefix)
+          super("A vocabulary with the prefix “#{prefix}” (specified in the site's configuration file) was not found.")
+        end
+      end
+
       def up
-        @voaf_metadata =
-          File.open('var/voaf_metadata.yaml') do |file|
-            YAML.safe_load(file.read,
-                           filename: file.path,
-                           permitted_classes: [Symbol])
-          end
+        load_extra_metadata(@config[:extra_metadata])
+
+        local_vocabs =
+          LifePreserver::Vocab.constants
+                              .map(&LifePreserver::Vocab.method(:const_get))
+                              .select { |constant| constant.is_a? Class }
+        register_vocabularies(local_vocabs, @config[:prefix_overrides])
+
+        if (categories = @config[:categories])
+          prefixes = categories.values.flatten
+          RDF::Vocabulary.limit_vocabs(*prefixes.map(&:to_sym))
+        end
       end
 
+      # Returns the collection of Vocabulary items, one per vocabulary prefix.
+      #
+      # @raise [UnknownVocabularyPrefix] if a vocabulary prefix specified in site config
+      #   does not have a mapping to an {RDF::Vocabulary} class.
+      # @return [Enumerable] The collection of items.
       def items
         items = []
 
-        @config[:prefixes_used].each do |group_name, group|
-          group.each do |prefix|
-            vocab = RDF::Vocabulary.find_by_prefix(prefix)
-            items << vocabulary_to_item(vocab, group_name)
+        @config.fetch(:categories, {}).each do |category, prefixes|
+          prefixes.each do |prefix|
+            # TODO: submit patch for RDF::Vocabulary to handle this edge case?
+            vocab = RDF::Vocabulary.vocab_map[prefix == 'rdf' ? :rdfv : prefix.to_sym]
+            raise UnknownVocabularyPrefix.new(prefix) unless vocab
+
+            vocab_class = vocab.fetch(:class, RDF::Vocabulary.from_sym(vocab.fetch(:class_name).to_sym))
+            items << vocabulary_to_item(prefix, vocab_class, category)
           end
         end
 
@@ -34,43 +55,36 @@ def items
 
       protected
 
-      def vocabulary_to_item(vocab, group_name)
-        prefix = vocab.__prefix__
-        metadata = extract_metadata_from(vocab)
+      def vocabulary_to_item(prefix, vocab, category)
+        metadata = find_vocab_metadata(vocab, @extra_metadata)
 
         attributes = {
           kind: 'vocabulary',
           prefix: prefix,
-          group: group_name,
+          category: category,
           namespace_uri: vocab.to_uri.value,
           is_hidden: true,
         }
 
         new_item(
           '',
           attributes.merge(metadata),
-          File.join(File::SEPARATOR, group_name.to_s.parameterize, prefix.to_s.parameterize),
+          File.join(File::SEPARATOR, category.to_s.parameterize, prefix.to_s.parameterize),
         )
       end
 
-      # @return [Hash]
-      def extract_metadata_from(vocab)
-        vocab_uri = vocab_uri(vocab)
-        metadata = Hash(@voaf_metadata[vocab_uri.to_sym] || @voaf_metadata[vocab_uri.split(%r{[/#]$})[0].to_sym])
-        metadata[:vocab_uri] = vocab_uri
-
-        if metadata.key?(:description)
-          metadata[:description] = cleanup(metadata[:description])
-        end
-
-        metadata
-      end
-
+      # Clean up punctuation of given text. Start with capital letter
+      # and end with a full stop if it doesn't already end with some sort of
+      # terminal punctuation.
       def cleanup(text)
         text = text.upcase_first
         text[/[.!?]$/] ? text : text << '.'
       end
 
+      # Return the URI of the given vocabulary.
+      #
+      # @param vocab [RDF::Vocabulary] The vocabulary.
+      # @return [String]
       def vocab_uri(vocab)
         # HACK: until we have better handling of ontology definitions at a
         # different uri than the vocabulary namespace.
@@ -80,6 +94,75 @@ def vocab_uri(vocab)
           vocab.ontology ? vocab.ontology.value : vocab.to_uri.value
         end
       end
+
+      # Find the descriptive metadata of the given vocabulary, searching
+      # within the given (normalized) metadata.
+      #
+      # Provided metadata is expected to have the following form:
+      #
+      # @example YAML vocabulary metadata format
+      #   vocab_uri:
+      #     title: Vocab title
+      #     description: Vocab description
+      #
+      # @param vocab [RDF::Vocabulary] The vocabulary.
+      # @return [Hash] Descriptive metadata for the given vocabulary, if found.
+      #   Otherwise, an empty Hash.
+      def find_vocab_metadata(vocab, extra_metadata = {})
+        vocab_uri = vocab_uri(vocab)
+
+        metadata = Hash(extra_metadata[vocab_uri.to_sym] || extra_metadata[vocab_uri.split(%r{[/#]$})[0].to_sym])
+        metadata[:vocab_uri] = vocab_uri
+
+        if metadata.key?(:description)
+          metadata[:description] = cleanup(metadata[:description])
+        end
+
+        metadata
+      end
+
+      # Load descriptive metadata for RDF vocabularies into the class.
+      #
+      # @note called by {#up} when the site is loading.
+      # @param metadata [Hash,String] the extra metadata. If a String, a
+      #   relative path to a YAML file, which will be loaded. If Hash, is used
+      #   directly as metadata.
+      # @return [Hash{Symbol=>Symbol}] the extra metadata.
+      def load_extra_metadata(metadata)
+        return if metadata.nil? || metadata.empty?
+
+        @extra_metadata =
+          case metadata
+          when String
+            File.open(metadata) do |file|
+              YAML.safe_load(file.read,
+                             filename: file.path,
+                             permitted_classes: [Symbol])
+            end
+          when Hash
+            metadata
+          end
+      end
+
+      # Register local vocabularies with the Ruby RDF framework.
+      #
+      # Sets the vocabulary class name if specified in the {:prefix_overrides}
+      # key of the data source configuration.
+      #
+      # @note called by {#up} when the site is loading.
+      #
+      # @param vocabs [Array<RDF::Vocabulary>] Vocabulary classes to register.
+      # @param prefixes [Hash{Symbol=>String}] Optional prefixes with to register
+      #   the vocabularies.
+      #
+      # @return [void]
+      def register_vocabularies(vocabs, prefixes = {})
+        Array(vocabs).each do |vocab|
+          class_name = vocab.__name__.to_s.demodulize
+          prefix = prefixes[class_name.to_sym] || class_name.downcase
+          RDF::Vocabulary.register(prefix.to_sym, vocab, class_name: class_name)
+        end
+      end
     end
   end
 end

lib/filters/rdf_distiller.rb

@@ -59,7 +59,7 @@ def run(_content, params = {})
 
         repository = RDF::Repository.new
 
-        RDF::Reader.for(input.to_sym).new(+assigns[:content], options) { |reader| repository << reader }
+        RDF::Reader.for(input.to_sym).new(+assigns[:content], **options) { |reader| repository << reader }
 
         if repository.has_statement?(RDF::Statement(RDF::URI.new(base_uri), RDF.type, RDF::OWL.Ontology))
           vocab = RDF::Vocabulary.find(base_uri) || RDF::Vocabulary.from_graph(repository, url: base_uri, class_name: prefix.to_s.upcase)
@@ -76,10 +76,10 @@ def run(_content, params = {})
 
             vocab.to_html(graph: repository, prefixes: prefixes, template: assigns[:layout][:filename])
           else
-            repository.dump(output_format, options)
+            repository.dump(output_format, **options)
           end
         else
-          repository.dump(output_format, options)
+          repository.dump(output_format, **options)
         end
       end
 

lib/helpers/vocabulary.rb

@@ -27,8 +27,8 @@ def prefix_mappings_for(*args)
       # Finds a vocabulary or vocabularies for the given name.
       #
       # @param [Array<String, Symbol>] args The vocabulary prefixes or prefix
-      #   groups. If a group name, it should return the group. If a prefix,
-      #   it should return a single vocabulary.
+      #   categories. If a category name, it should return the entire category of
+      #   vocabularies. If a prefix, it should return a single vocabulary.
       #
       # @return [Array<Nanoc::Core::BasicItemView>] Any applicable items of kind +vocabulary+.
       def vocabularies_for(*args)
@@ -37,7 +37,7 @@ def vocabularies_for(*args)
         Array(args).flatten.each do |arg|
           vocabularies_root = @config.fetch(:vocabularies_root)
           vocabs = @items.find_all(File.join(vocabularies_root, arg.to_s, '*'))
-          # Look for a single prefix if no group is found
+          # Look for a single prefix if no category is found
           vocabs = vocabs.present? ? vocabs : @items[File.join(vocabularies_root, '*', arg.to_s)]
 
           if vocabs.blank?