Create FileSetDescription to find characterization for FileSets

Valkyrie-based FileSet objects don't have direct access to characterization
information. In ActiveFedora, we used a `#characterization_proxy` to provide
that information. Here, we provide a small service that can find the "primary"
file (`FileMetadata`) for a given FileSet, and ask it about its description.

app/models/hyrax/file_metadata.rb

@@ -2,14 +2,16 @@
 
 module Hyrax
   class FileMetadata < Valkyrie::Resource
+    GENERIC_MIME_TYPE = 'application/octet-stream'
+
     attribute :file_identifiers, ::Valkyrie::Types::Set # id of the file stored by the storage adapter
     attribute :alternate_ids, Valkyrie::Types::Set.of(Valkyrie::Types::ID) # id of the Hydra::PCDM::File which holds metadata and the file in ActiveFedora
     attribute :file_set_id, ::Valkyrie::Types::ID # id of parent file set resource
 
     # all remaining attributes are on AF::File metadata_node unless otherwise noted
     attribute :label, ::Valkyrie::Types::Set
     attribute :original_filename, ::Valkyrie::Types::Set
-    attribute :mime_type, ::Valkyrie::Types::Set
+    attribute :mime_type, ::Valkyrie::Types::String.default(GENERIC_MIME_TYPE)
     attribute :use, ::Valkyrie::Types::Set # AF::File type
     attribute :content, ::Valkyrie::Types::Set
 
@@ -79,14 +81,27 @@ def self.for(file:)
           use: file.try(:use) || [::Valkyrie::Vocab::PCDMUse.OriginalFile])
     end
 
+    ##
+    # @param [Symbol] use
+    # @return [Boolean]
+    def used_for?(use)
+      public_send("#{use}?")
+    end
+
+    ##
+    # @return [Boolean]
     def original_file?
       use.include?(::Valkyrie::Vocab::PCDMUse.OriginalFile)
     end
 
+    ##
+    # @return [Boolean]
     def thumbnail_file?
       use.include?(::Valkyrie::Vocab::PCDMUse.ThumbnailImage)
     end
 
+    ##
+    # @return [Boolean]
     def extracted_file?
       use.include?(::Valkyrie::Vocab::PCDMUse.ExtractedImage)
     end

app/services/hyrax/characterization/file_set_description.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+module Hyrax
+  module Characterization
+    ##
+    # @api public
+    class FileSetDescription
+      include Hydra::Works::MimeTypes
+
+      ##
+      # @!attribute [rw] file_set
+      #   @return [Hyrax::FileSet]
+      attr_accessor :file_set
+
+      delegate :mime_type, to: :primary_file
+
+      ##
+      # @param [Hyrax::FileSet] file_set
+      # @param [Symbol] primary_file  a symbol mapping to the file_set member
+      #   used for characterization
+      def initialize(file_set:, primary_file: :original_file)
+        self.file_set = file_set
+        @primary_file = primary_file
+      end
+
+      ##
+      # @api public
+      # @return [Hyrax::FileMetadata] the member file to use for characterization
+      def primary_file
+        files.find { |f| f.used_for?(@primary_file) } || Hyrax::FileMetadata.new
+      end
+
+      private
+
+        ##
+        # @api private
+        def files
+          @__files__ ||= Hyrax.query_service.custom_queries.find_files(file_set: file_set)
+        end
+    end
+  end
+end

spec/models/hyrax/file_metadata_spec.rb

@@ -7,21 +7,31 @@
     let(:resource_klass) { described_class }
   end
 
-  let(:file) { Rack::Test::UploadedFile.new('spec/fixtures/world.png', 'image/png') }
-  let(:subject) do
+  subject(:file_metadata) do
     described_class.for(file: file).new(id: 'test_id', format_label: 'test_format_label')
   end
+
+  let(:file) { Rack::Test::UploadedFile.new('spec/fixtures/world.png', 'image/png') }
   let(:pcdm_file_uri) { RDF::URI('http://pcdm.org/models#File') }
 
   it 'sets the proper attributes' do
     expect(subject.id.to_s).to eq 'test_id'
     expect(subject.label).to contain_exactly('world.png')
     expect(subject.original_filename).to contain_exactly('world.png')
-    expect(subject.mime_type).to contain_exactly('image/png')
+    expect(subject.mime_type).to eq('image/png')
     expect(subject.format_label).to contain_exactly('test_format_label')
     expect(subject.use).to contain_exactly(Valkyrie::Vocab::PCDMUse.OriginalFile)
   end
 
+  describe '#used_for?' do
+    it 'is true when use matches' do
+      expect { file_metadata.use = [Valkyrie::Vocab::PCDMUse.ThumbnailImage] }
+        .to change { file_metadata.used_for?(:original_file) }
+        .from(true)
+        .to false
+    end
+  end
+
   describe '#original_file?' do
     context 'when use says file is the original file' do
       before { subject.use = [Valkyrie::Vocab::PCDMUse.OriginalFile, pcdm_file_uri] }

spec/services/hyrax/characterization/file_set_description_spec.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+RSpec.describe Hyrax::Characterization::FileSetDescription, valkyrie_adapter: :test_adapter do
+  subject(:description) { described_class.new(file_set: file_set) }
+
+  let(:ctype) { 'image/png' }
+  let(:file) { Rack::Test::UploadedFile.new('spec/fixtures/world.png', ctype) }
+  let(:file_set) { FactoryBot.valkyrie_create(:hyrax_file_set, file_ids: file_ids) }
+  let(:file_ids) { [] }
+
+  describe '#mime_type' do
+    context 'before the file set is saved' do
+      let(:file_set) { FactoryBot.build(:hyrax_file_set) }
+
+      it 'has a generic MIME type' do
+        expect(description.mime_type).to eq 'application/octet-stream'
+      end
+    end
+
+    context 'when it has no files' do
+      it 'has a generic MIME type' do
+        expect(description.mime_type).to eq 'application/octet-stream'
+      end
+    end
+
+    context 'when it has an original file' do
+      let(:file_ids) { [original_file.id] }
+      let(:original_file) { Hyrax.persister.save(resource: Hyrax::FileMetadata.for(file: file)) }
+
+      it { is_expected.to be_image }
+
+      it 'has a mime type from a file' do
+        expect(description.mime_type).to eq ctype
+      end
+    end
+  end
+end

spec/wings/services/file_metadata_builder_spec.rb

@@ -32,7 +32,7 @@
       expect(built_file_metadata.file_set_id.id).to eq file_set.id.id
       expect(built_file_metadata.label).to contain_exactly(original_name)
       expect(built_file_metadata.original_filename).to contain_exactly(original_name)
-      expect(built_file_metadata.mime_type).to contain_exactly(mime_type)
+      expect(built_file_metadata.mime_type).to eq mime_type
       expect(built_file_metadata.use).to contain_exactly(use)
     end
   end