Make derivatives configurable

This commit introduces a new configuration option so it will make it
easier for consuming applications to override default derivative
settings.

Author: kirkkwang

app/services/hyrax/file_set_derivatives_service.rb

@@ -67,49 +67,39 @@ def supported_mime_types
     end
 
     def create_pdf_derivatives(filename)
-      Hydra::Derivatives::PdfDerivatives.create(filename,
-                                                outputs: [{
-                                                  label: :thumbnail,
-                                                  format: 'jpg',
-                                                  size: '338x493',
-                                                  url: derivative_url('thumbnail'),
-                                                  layer: 0
-                                                }])
-      extract_full_text(filename, derivative_url('extracted_text'))
+      create_derivatives_for(:pdf, filename, Hydra::Derivatives::PdfDerivatives)
     end
 
     def create_office_document_derivatives(filename)
-      Hydra::Derivatives::DocumentDerivatives.create(filename,
-                                                     outputs: [{
-                                                       label: :thumbnail, format: 'jpg',
-                                                       size: '200x150>',
-                                                       url: derivative_url('thumbnail'),
-                                                       layer: 0
-                                                     }])
-      extract_full_text(filename, derivative_url('extracted_text'))
+      create_derivatives_for(:office, filename, Hydra::Derivatives::DocumentDerivatives)
     end
 
     def create_audio_derivatives(filename)
-      Hydra::Derivatives::AudioDerivatives.create(filename,
-                                                  outputs: [{ label: 'mp3', format: 'mp3', url: derivative_url('mp3'), mime_type: 'audio/mpeg', container: 'service_file' },
-                                                            { label: 'ogg', format: 'ogg', url: derivative_url('ogg'), mime_type: 'audio/ogg', container: 'service_file' }])
+      create_derivatives_for(:audio, filename, Hydra::Derivatives::AudioDerivatives)
     end
 
     def create_video_derivatives(filename)
-      Hydra::Derivatives::VideoDerivatives.create(filename,
-                                                  outputs: [{ label: :thumbnail, format: 'jpg', url: derivative_url('thumbnail'), mime_type: 'image/jpeg' },
-                                                            { label: 'webm', format: 'webm', url: derivative_url('webm'), mime_type: 'video/webm', container: 'service_file' },
-                                                            { label: 'mp4', format: 'mp4', url: derivative_url('mp4'), mime_type: 'video/mp4', container: 'service_file' }])
+      create_derivatives_for(:video, filename, Hydra::Derivatives::VideoDerivatives)
     end
 
     def create_image_derivatives(filename)
-      # We're asking for layer 0, becauase otherwise pyramidal tiffs flatten all the layers together into the thumbnail
-      Hydra::Derivatives::ImageDerivatives.create(filename,
-                                                  outputs: [{ label: :thumbnail,
-                                                              format: 'jpg',
-                                                              size: '200x150>',
-                                                              url: derivative_url('thumbnail'),
-                                                              layer: 0 }])
+      create_derivatives_for(:image, filename, Hydra::Derivatives::ImageDerivatives)
+    end
+
+    # Routes +extracted_text+ outputs to full text extraction; the rest to +processor+.
+    def create_derivatives_for(type, filename, processor)
+      text_outputs, processor_outputs = derivative_outputs(type).partition { |output| output[:container] == 'extracted_text' }
+      processor.create(filename, outputs: processor_outputs) if processor_outputs.any?
+      extract_full_text(filename, text_outputs) if text_outputs.any?
+    end
+
+    # Resolves configured specs for +type+: callable values are invoked with the
+    # file set, and +:url+ destination names become real derivative URLs.
+    def derivative_outputs(type)
+      Hyrax.config.derivative_options.fetch(type).map do |output|
+        resolved = output.transform_values { |value| value.respond_to?(:call) ? value.call(file_set) : value }
+        resolved.merge(url: derivative_url(resolved[:url]))
+      end
     end
 
     def derivative_path_factory
@@ -119,12 +109,11 @@ def derivative_path_factory
     # Calls the Hydra::Derivates::FulltextExtraction unless the extract_full_text
     # configuration option is set to false
     # @param [String] filename of the object to be used for full text extraction
-    # @param [String] uri to the file set (deligated to file_set)
-    def extract_full_text(filename, uri)
+    # @param [Array<Hash>] outputs the resolved extracted_text output specs
+    def extract_full_text(filename, outputs)
       return unless Hyrax.config.extract_full_text?
 
-      Hydra::Derivatives::FullTextExtract.create(filename,
-                                                 outputs: [{ url: uri, container: "extracted_text" }])
+      Hydra::Derivatives::FullTextExtract.create(filename, outputs: outputs)
     end
   end
 end

lib/hyrax/configuration.rb

@@ -284,6 +284,43 @@ def derivative_mime_type_mappings
           video: lookup_mimes(:video_mime_types) }
     end
 
+    attr_writer :derivative_options
+
+    ##
+    # The +outputs:+ specs passed to +Hydra::Derivatives+ per derivative type.
+    # +:url+ is a destination name resolved to a real URL at create time; a
+    # callable value is invoked with the file set. An +extracted_text+ container
+    # routes to full text extraction (gated also by {#extract_full_text?}) — its
+    # presence is the per-type toggle. Tune these instead of decorating
+    # +Hyrax::FileSetDerivativesService+:
+    #
+    #   Hyrax.config do |config|
+    #     config.derivative_options[:pdf].first[:size] = '676x986'                            # bigger thumbnail
+    #     config.derivative_options[:pdf].reject! { |o| o[:container] == 'extracted_text' }   # no PDF text
+    #
+    #     # transcode video to its source dimensions (falls back to 320x240)
+    #     sizer = ->(fs) { (w = Array(fs.width).first) && (h = Array(fs.height).first) ? "#{w}x#{h}" : '320x240' }
+    #     config.derivative_options[:video].each { |o| o[:size] = sizer }
+    #   end
+    #
+    # @return [Hash{Symbol => Array<Hash>}]
+    # @see Hyrax::FileSetDerivativesService
+    def derivative_options
+      @derivative_options ||= {
+        pdf: [{ label: :thumbnail, format: 'jpg', size: '338x493', url: 'thumbnail', layer: 0 },
+              { url: 'extracted_text', container: 'extracted_text' }],
+        office: [{ label: :thumbnail, format: 'jpg', size: '200x150>', url: 'thumbnail', layer: 0 },
+                 { url: 'extracted_text', container: 'extracted_text' }],
+        audio: [{ label: 'mp3', format: 'mp3', url: 'mp3', mime_type: 'audio/mpeg', container: 'service_file' },
+                { label: 'ogg', format: 'ogg', url: 'ogg', mime_type: 'audio/ogg', container: 'service_file' }],
+        video: [{ label: :thumbnail, format: 'jpg', url: 'thumbnail', mime_type: 'image/jpeg' },
+                { label: 'webm', format: 'webm', url: 'webm', mime_type: 'video/webm', container: 'service_file' },
+                { label: 'mp4', format: 'mp4', url: 'mp4', mime_type: 'video/mp4', container: 'service_file' }],
+        # image uses layer: 0 because otherwise pyramidal tiffs flatten all the layers together into the thumbnail
+        image: [{ label: :thumbnail, format: 'jpg', size: '200x150>', url: 'thumbnail', layer: 0 }]
+      }
+    end
+
     attr_writer :derivative_services
     # The registered candidate derivative services.  In the array, the first `valid?` candidate will
     # handle the derivative generation.

spec/lib/hyrax/configuration_spec.rb

@@ -47,6 +47,8 @@
   it { is_expected.to respond_to(:default_admin_set_id) }
   it { is_expected.to respond_to(:derivative_services) }
   it { is_expected.to respond_to(:derivative_services=) }
+  it { is_expected.to respond_to(:derivative_options) }
+  it { is_expected.to respond_to(:derivative_options=) }
   it { is_expected.to respond_to(:display_media_download_link=) }
   it { is_expected.to respond_to(:display_media_download_link?) }
   it { is_expected.to respond_to(:display_microdata?) }
@@ -132,6 +134,31 @@
     end
   end
 
+  describe "#derivative_options" do
+    it "defaults to output specs for each derivative type" do
+      expect(described_class.new.derivative_options.keys)
+        .to contain_exactly(:pdf, :office, :audio, :video, :image)
+    end
+
+    it "resolves url to a destination name rather than a full url" do
+      expect(described_class.new.derivative_options[:image])
+        .to contain_exactly(hash_including(label: :thumbnail, url: 'thumbnail', layer: 0))
+    end
+
+    it "is overridable" do
+      config = described_class.new
+      config.derivative_options = { image: [{ label: :big, format: 'jpg', size: '999x999', url: 'big' }] }
+      expect(config.derivative_options[:image].first).to include(size: '999x999')
+    end
+
+    it "extracts text for pdf and office by default" do
+      %i[pdf office].each do |type|
+        expect(described_class.new.derivative_options[type])
+          .to include(hash_including(container: 'extracted_text'))
+      end
+    end
+  end
+
   describe "#use_valkyrie?" do
     before { stub_const("ENV", "HYRAX_SKIP_WINGS" => "true") }
     it "returns true if wings is disabled" do

spec/services/hyrax/file_set_derivatives_service_spec.rb

@@ -42,6 +42,66 @@
         end
       end
 
+      context "when an app overrides only the image thumbnail size" do
+        let(:valid_file_set) do
+          FactoryBot.valkyrie_create(:hyrax_file_metadata, :image, file_set_id: SecureRandom.uuid)
+        end
+        let(:custom_size) { '9999x9999' }
+
+        before do
+          allow(Hyrax.config).to receive(:derivative_options).and_return(
+            image: [Hyrax::Configuration.new.derivative_options[:image].first.merge(size: custom_size)]
+          )
+        end
+
+        it "creates the thumbnail at the configured size but keeps the other defaults" do
+          allow(Hydra::Derivatives::ImageDerivatives).to receive(:create)
+          described_class.new(valid_file_set).create_derivatives('foo.jpg')
+          expect(Hydra::Derivatives::ImageDerivatives).to have_received(:create).with(
+            'foo.jpg',
+            outputs: contain_exactly(
+              hash_including(size: custom_size, label: :thumbnail, format: 'jpg', layer: 0)
+            )
+          )
+        end
+      end
+
+      context "when given a pdf file" do
+        let(:valid_file_set) do
+          FactoryBot.valkyrie_create(:hyrax_file_metadata, mime_type: 'application/pdf', file_set_id: SecureRandom.uuid)
+        end
+
+        before do
+          allow(Hydra::Derivatives::PdfDerivatives).to receive(:create)
+          allow(Hydra::Derivatives::FullTextExtract).to receive(:create)
+        end
+
+        it "creates the thumbnail and extracts full text" do
+          described_class.new(valid_file_set).create_derivatives('foo.pdf')
+          expect(Hydra::Derivatives::PdfDerivatives).to have_received(:create).with(
+            'foo.pdf', outputs: contain_exactly(hash_including(label: :thumbnail))
+          )
+          expect(Hydra::Derivatives::FullTextExtract).to have_received(:create).with(
+            'foo.pdf', outputs: contain_exactly(hash_including(container: 'extracted_text'))
+          )
+        end
+
+        context "when the extracted_text output is removed from the config" do
+          before do
+            pdf_without_text = Hyrax::Configuration.new.derivative_options[:pdf].reject { |o| o[:container] == 'extracted_text' }
+            allow(Hyrax.config).to receive(:derivative_options).and_return(pdf: pdf_without_text)
+          end
+
+          it "still creates the thumbnail but skips full text extraction" do
+            described_class.new(valid_file_set).create_derivatives('foo.pdf')
+            expect(Hydra::Derivatives::PdfDerivatives).to have_received(:create).with(
+              'foo.pdf', outputs: contain_exactly(hash_including(label: :thumbnail))
+            )
+            expect(Hydra::Derivatives::FullTextExtract).not_to have_received(:create)
+          end
+        end
+      end
+
       context "when given a video file" do
         let(:valid_file_set) do
           FactoryBot.valkyrie_create(:hyrax_file_metadata, :video_file, file_set_id: SecureRandom.uuid)
@@ -59,6 +119,26 @@
             )
           )
         end
+
+        context "when an output size is a callable" do
+          let(:received_file_set) { [] }
+
+          before do
+            size_from = ->(file_set) { received_file_set << file_set && '1280x720' }
+            sized = Hyrax::Configuration.new.derivative_options[:video].map { |output| output.merge(size: size_from) }
+            allow(Hyrax.config).to receive(:derivative_options).and_return(video: sized)
+          end
+
+          it "invokes it with the file set and passes the result through" do
+            allow(Hydra::Derivatives::VideoDerivatives).to receive(:create)
+            described_class.new(valid_file_set).create_derivatives('foo.mov')
+
+            expect(received_file_set).to all(eq(valid_file_set))
+            expect(Hydra::Derivatives::VideoDerivatives).to have_received(:create) do |_filename, outputs:|
+              expect(outputs).to all(include(size: '1280x720'))
+            end
+          end
+        end
       end
     end
   end