Delete derivatives when FileSet is destroyed.

curation_concerns-models/app/models/concerns/curation_concerns/file_set/derivatives.rb

@@ -7,6 +7,7 @@ module Derivatives
         Hydra::Derivatives.source_file_service = CurationConcerns::LocalFileService
         Hydra::Derivatives.output_file_service = CurationConcerns::PersistDerivatives
         Hydra::Derivatives::FullTextExtract.output_file_service = CurationConcerns::PersistDirectlyContainedOutputFileService
+        after_destroy :cleanup_derivatives
       end
 
       # This completely overrides the version in Hydra::Works so that we
@@ -46,9 +47,19 @@ def create_derivatives(filename)
         # The destination_name parameter has to match up with the file parameter
         # passed to the DownloadsController
         def derivative_url(destination_name)
-          path = DerivativePath.derivative_path_for_reference(self, destination_name)
+          path = derivative_path_factory.derivative_path_for_reference(self, destination_name)
           URI("file://#{path}").to_s
         end
+
+        def cleanup_derivatives
+          derivative_path_factory.derivatives_for_reference(self).each do |path|
+            FileUtils.rm_f(path)
+          end
+        end
+
+        def derivative_path_factory
+          DerivativePath
+        end
     end
   end
 end

curation_concerns-models/app/services/curation_concerns/derivative_path.rb

@@ -7,16 +7,33 @@ def derivative_path_for_reference(object, destination_name)
         derivative_path(object, extension_for(destination_name), destination_name)
       end
 
+      # @return [Array<String>] Array of paths to derivatives for this object.
+      def derivatives_for_reference(object)
+        Dir.glob(root_path(object).join("*")).select do |path|
+          path.start_with?(path_prefix(object).to_s)
+        end
+      end
+
       private
 
+        # @param [#id] object Object whose ID is used to generate root path
+        # @return [String] Returns the root path where derivatives will be generated into.
+        def root_path(object)
+          Pathname.new(derivative_path(object, "", "")).dirname
+        end
+
+        # @return <Pathname> Full prefix of the path for object.
+        def path_prefix(object)
+          Pathname.new(CurationConcerns.config.derivatives_path).join(pair_path(object.id))
+        end
+
         def derivative_path(object, extension, destination_name)
           file_name = destination_name + extension
-          File.join(CurationConcerns.config.derivatives_path, pair_path(object.id, file_name))
+          "#{path_prefix(object)}-#{file_name}"
         end
 
-        def pair_path(id, file_name)
-          pair = id.split('').each_slice(2).map(&:join).join('/')
-          "#{pair}-#{file_name}"
+        def pair_path(id)
+          id.split('').each_slice(2).map(&:join).join('/')
         end
 
         def extension_for(destination_name)

spec/models/curation_concerns/file_set/derivatives_spec.rb

@@ -81,4 +81,18 @@
       end
     end
   end
+
+  describe "cleanup" do
+    let(:mime_type) { 'image/jpg' }
+    it "cleans up all created derivatives" do
+      allow(CurationConcerns::DerivativePath).to receive(:derivatives_for_reference).with(file_set).and_return([
+        "tmp/1/2.jpg"
+      ])
+      allow(FileUtils).to receive(:rm_f).with("tmp/1/2.jpg")
+
+      file_set.destroy
+
+      expect(FileUtils).to have_received(:rm_f).with("tmp/1/2.jpg")
+    end
+  end
 end

spec/services/derivative_path_spec.rb

@@ -13,4 +13,28 @@
 
     it { is_expected.to eq 'tmp/12/3-thumbnail.jpeg' }
   end
+
+  describe "#derivatives_for_reference" do
+    subject { described_class.derivatives_for_reference(object) }
+    before do
+      FileUtils.mkdir_p("tmp/12")
+      File.open("tmp/12/3-thumbnail.jpeg", 'w') do |f|
+        f.write "test"
+      end
+      File.open("tmp/12/4-thumbnail.jpeg", 'w') do |f|
+        f.write "test"
+      end
+    end
+    after do
+      FileUtils.rm_rf("tmp/12")
+    end
+
+    let(:object) { double(id: '123') }
+
+    it "lists all the paths to derivatives" do
+      expect(subject).to eq [
+        "tmp/12/3-thumbnail.jpeg"
+      ]
+    end
+  end
 end