Merge pull request #5704 from avalonmediasystem/batch_captions

Support creation of multiple captions in batch ingest

app/jobs/ingest_batch_entry_job.rb

@@ -69,6 +69,7 @@ def process_success(batch_entry, entry)
       old_media_object_id = batch_entry.media_object_pid
       batch_entry.media_object_pid = entry.media_object.id
       batch_entry.complete = true
+      batch_entry.current_status = 'Completed'
       batch_entry.save!
       # Delete pre-existing media object
       MediaObject.find(old_media_object_id).destroy if old_media_object_id.present? && MediaObject.exists?(old_media_object_id)

lib/avalon/batch/entry.rb

@@ -71,7 +71,7 @@ def self.from_json(json)
         json_hash = JSON.parse(json)
         opts = json_hash.except("fields", "files", "position")
         opts[:collection] = Admin::Collection.find(json_hash["collection"])
-        self.new(json_hash["fields"].symbolize_keys, json_hash["files"].map(&:symbolize_keys!), opts.symbolize_keys, json_hash["position"], nil)
+        self.new(json_hash["fields"].symbolize_keys, json_hash["files"].map(&:deep_symbolize_keys!), opts.symbolize_keys, json_hash["position"], nil)
       end
 
       def user_key
@@ -176,17 +176,21 @@ def self.offset_valid?( offset )
         true
       end
 
-      def self.attach_datastreams_to_master_file( master_file, filename )
+      def self.attach_datastreams_to_master_file( master_file, filename, captions )
         structural_file = "#{filename}.structure.xml"
         if FileLocator.new(structural_file).exist?
           master_file.structuralMetadata.content=FileLocator.new(structural_file).reader
           master_file.structuralMetadata.original_name = structural_file
         end
-        captions_file = "#{filename}.vtt"
-        if FileLocator.new(captions_file).exist?
-          master_file.captions.content=FileLocator.new(captions_file).reader
-          master_file.captions.mime_type='text/vtt'
-          master_file.captions.original_name = captions_file
+        captions.each do |c|
+          next unless c.present? && c[:caption_file].present? && FileLocator.new(c[:caption_file]).exist?
+          filename = c[:caption_file].split('/').last
+          label = c[:caption_label].presence || filename
+          language = c[:caption_language].present? ? caption_language(c[:caption_language]) : Settings.caption_default.language
+          supplemental_file = SupplementalFile.new(label: label, tags: ['caption'], language: language)
+          supplemental_file.file.attach(io: FileLocator.new(c[:caption_file]).reader, filename: filename)
+          supplemental_file.save
+          master_file.supplemental_files += [supplemental_file]
         end
       end
 
@@ -198,7 +202,8 @@ def process!
           # master_file.save(validate: false) #required: need id before setting media_object
           # master_file.media_object = media_object
           files = self.class.gatherFiles(file_spec[:file])
-          self.class.attach_datastreams_to_master_file(master_file, file_spec[:file])
+          captions = gather_captions(file_spec).values
+          self.class.attach_datastreams_to_master_file(master_file, file_spec[:file], captions)
           master_file.setContent(files, dropbox_dir: media_object.collection.dropbox_absolute_path)
 
           # Overwrite files hash with working file paths to pass to matterhorn
@@ -263,11 +268,25 @@ def self.derivativePath(filename, quality)
         filename.dup.insert(filename.rindex('.'), ".#{quality}")
       end
 
+      def self.caption_language(language)
+        begin
+          LanguageTerm.find(language.capitalize).code
+        rescue LanguageTerm::LookupError
+          Settings.caption_default.language
+        end
+      end
+      private_class_method :caption_language
+
       private
 
         def hidden
           !!opts[:hidden]
         end
+
+        def gather_captions(file)
+          [] unless file.keys.any? { |k| k.to_s.include?('caption') }
+          file.select { |f| f.to_s.include?('caption') }
+        end
     end
   end
 end

lib/avalon/batch/manifest.rb

@@ -21,7 +21,7 @@ class Manifest
       extend Forwardable
 
       EXTENSIONS = ['csv','xls','xlsx','ods']
-      FILE_FIELDS = [:file,:label,:offset,:skip_transcoding,:absolute_location,:date_digitized]
+      FILE_FIELDS = [:file,:label,:offset,:skip_transcoding,:absolute_location,:date_digitized, :caption_file, :caption_label, :caption_language]
       SKIP_FIELDS = [:collection]
 
       def_delegators :@entries, :each
@@ -102,6 +102,18 @@ def true?(value)
         not (value.to_s =~ /^(y(es)?|t(rue)?)$/i).nil?
       end
 
+      def process_captions(field, content, values, i)
+        if field.to_s.include?('file')
+          @caption_count += 1
+          @caption_key = "caption_#{@caption_count}".to_sym
+          content.last[@caption_key] = {}
+          # Set file path to caption file
+          content.last[@caption_key][field] = path_to(values[i])
+        end
+        # Set caption metadata fields
+        content.last[@caption_key][field] ||= values[i]
+      end
+
       def create_entries!
         first = @spreadsheet.first_row + 2
         last = @spreadsheet.last_row
@@ -117,10 +129,15 @@ def create_entries!
           content=[]
 
           fields = Hash.new { |h,k| h[k] = [] }
+          @caption_count = 0
           @field_names.each_with_index do |f,i|
             unless f.blank? || SKIP_FIELDS.include?(f) || values[i].blank?
               if FILE_FIELDS.include?(f)
                 content << {} if f == :file
+                if f.to_s.include?('caption')
+                  process_captions(f, content, values, i)
+                  next
+                end
                 content.last[f] = f == :skip_transcoding ? true?(values[i]) : values[i]
               else
                 fields[f] << values[i]
@@ -140,7 +157,6 @@ def create_entries!
           entries << Entry.new(fields.select { |f| !FILE_FIELDS.include?(f) }, files, opts, index, self)
         end
       end
-
     end
   end
 end

spec/lib/avalon/batch/entry_spec.rb

@@ -176,20 +176,46 @@
       expect(master_file.absolute_location).to eq(Avalon::FileResolver.new.path_to(master_file.file_location))
       expect(master_file.date_digitized).to eq('2015-10-30T00:00:00Z')
     end
+
+    context 'with caption files' do
+      let(:caption_file) { File.join(Rails.root, 'spec/fixtures/dropbox/example_batch_ingest/assets/sheephead_mountain.mov.vtt')}
+      let(:caption) {{ :caption_file => caption_file, :caption_label => 'Sheephead Captions', :caption_language => 'English' }}
+      let(:entry_files) { [{ file: File.join(testdir, filename), offset: '00:00:00.500', label: 'Quis quo', date_digitized: '2015-10-30', skip_transcoding: false, caption_1: caption }] }
+
+      it 'adds captions to masterfile' do
+        expect(master_file.supplemental_file_captions).to be_present
+      end
+    end
   end
 
   describe '#attach_datastreams_to_master_file' do
-    let(:master_file) { FactoryBot.build(:master_file) }
+    let(:master_file) { FactoryBot.create(:master_file) }
     let(:filename) { File.join(Rails.root, 'spec/fixtures/dropbox/example_batch_ingest/assets/sheephead_mountain.mov') }
+    let(:caption_file) { File.join(Rails.root, 'spec/fixtures/dropbox/example_batch_ingest/assets/sheephead_mountain.mov.vtt')}
+    let(:caption) { [{ :caption_file => caption_file, :caption_label => 'Sheephead Captions', :caption_language => 'English' }] }
+
     before do
-      Avalon::Batch::Entry.attach_datastreams_to_master_file(master_file, filename)
+      Avalon::Batch::Entry.attach_datastreams_to_master_file(master_file, filename, caption)
     end
 
     it 'should attach structural metadata' do
       expect(master_file.structuralMetadata.has_content?).to be_truthy
     end
     it 'should attach captions' do
-      expect(master_file.captions.has_content?).to be_truthy
+      expect(master_file.supplemental_file_captions).to be_present
+    end
+
+    context 'with multiple captions' do
+      let(:caption) { [{ :caption_file => caption_file, :caption_label => 'Sheephead Captions', :caption_language => 'english' },
+                      { :caption_file => caption_file, :caption_label => 'Second Caption', :caption_language => 'fre' }] }
+      it 'should attach all captions to master file' do
+        expect(master_file.supplemental_file_captions).to be_present
+        expect(master_file.supplemental_file_captions.count).to eq 2
+        expect(master_file.supplemental_file_captions[0].label).to eq 'Sheephead Captions'
+        expect(master_file.supplemental_file_captions[1].label).to eq 'Second Caption'
+        expect(master_file.supplemental_file_captions[0].language).to eq 'eng'
+        expect(master_file.supplemental_file_captions[1].language).to eq 'fre'
+      end
     end
   end