Merge pull request #5763 from avalonmediasystem/caption_as_transcript

Add option to treat a caption file as transcript

app/controllers/supplemental_files_controller.rb

@@ -151,12 +151,20 @@ def edit_structure_path
     end
 
     def edit_file_information
-      ident = "machine_generated_#{params[:id]}".to_sym
-
-      if params[ident] && !@supplemental_file.machine_generated?
-        @supplemental_file.tags += ['machine_generated']
-      elsif !params[ident] && @supplemental_file.machine_generated?
-        @supplemental_file.tags -= ['machine_generated']
+      file_params = [ 
+        { param: "machine_generated_#{params[:id]}".to_sym, tag: "machine_generated", method: :machine_generated? },
+        { param: "treat_as_transcript_#{params[:id]}".to_sym, tag: "transcript", method: :caption_transcript? }
+      ]
+
+      file_params.each do |v|
+        param_name = v[:param]
+        tag = v[:tag]
+        method = v[:method]
+        if params[param_name] && !@supplemental_file.send(method)
+          @supplemental_file.tags += [tag]
+        elsif !params[param_name] && @supplemental_file.send(method)
+          @supplemental_file.tags -= [tag]
+        end
       end
       @supplemental_file.label = supplemental_file_params[:label]
       return unless supplemental_file_params[:language].present?

app/models/iiif_canvas_presenter.rb

@@ -41,7 +41,7 @@ def display_content
   end
 
   def annotation_content
-    supplemental_captions_transcripts.collect { |file| supplementing_content_data(file) }
+    supplemental_captions_transcripts.uniq.collect { |file| supplementing_content_data(file) }.flatten
   end
 
   def sequence_rendering
@@ -103,16 +103,28 @@ def audio_display_content(quality)
     end
 
     def supplementing_content_data(file)
-      url = if !file.is_a?(SupplementalFile)
-              Rails.application.routes.url_helpers.captions_master_file_url(master_file.id)
-            elsif file.tags.include?('caption')
-              Rails.application.routes.url_helpers.captions_master_file_supplemental_file_url(master_file.id, file.id)
-            elsif file.tags.include?('transcript')
-              Rails.application.routes.url_helpers.transcripts_master_file_supplemental_file_url(master_file.id, file.id)
-            else
-              Rails.application.routes.url_helpers.master_file_supplemental_file_url(master_file.id, file.id)
-            end
-      IIIFManifest::V3::AnnotationContent.new(body_id: url, **supplemental_attributes(file))
+      unless file.is_a?(SupplementalFile)
+        url = Rails.application.routes.url_helpers.captions_master_file_url(master_file.id)
+        return IIIFManifest::V3::AnnotationContent.new(body_id: url, **supplemental_attributes(file))
+      end
+
+      tags = file.tags.reject { |t| t == 'machine_generated' }.compact
+      case tags
+      when ['caption']
+        url = Rails.application.routes.url_helpers.captions_master_file_supplemental_file_url(master_file.id, file.id)
+        IIIFManifest::V3::AnnotationContent.new(body_id: url, **supplemental_attributes(file, type: 'caption'))
+      when ['transcript']
+        url = Rails.application.routes.url_helpers.transcripts_master_file_supplemental_file_url(master_file.id, file.id)
+        IIIFManifest::V3::AnnotationContent.new(body_id: url, **supplemental_attributes(file, type: 'transcript'))
+      when ['caption', 'transcript']
+        caption_url = Rails.application.routes.url_helpers.captions_master_file_supplemental_file_url(master_file.id, file.id)
+        transcript_url = Rails.application.routes.url_helpers.transcripts_master_file_supplemental_file_url(master_file.id, file.id)
+        [IIIFManifest::V3::AnnotationContent.new(body_id: caption_url, **supplemental_attributes(file, type: 'caption')),
+         IIIFManifest::V3::AnnotationContent.new(body_id: transcript_url, **supplemental_attributes(file, type: 'transcript'))]
+      else
+        url = Rails.application.routes.url_helpers.master_file_supplemental_file_url(master_file.id, file.id)
+        IIIFManifest::V3::AnnotationContent.new(body_id: url, **supplemental_attributes(file))
+      end
     end
 
     def stream_urls
@@ -206,10 +218,10 @@ def manifest_attributes(quality, media_type)
       end
     end
 
-    def supplemental_attributes(file)
+    def supplemental_attributes(file, type: nil)
       if file.is_a?(SupplementalFile)
         label = file.tags.include?('machine_generated') ? file.label + ' (machine generated)' : file.label
-        format = if file.file.content_type == 'text/srt' && file.tags.include?('caption')
+        format = if file.file.content_type == 'text/srt' && type == 'caption'
                    'text/vtt'
                  else
                    file.file.content_type

app/models/supplemental_file.rb

@@ -46,6 +46,10 @@ def machine_generated?
     tags.include?('machine_generated')
   end
 
+  def caption_transcript?
+    tags.include?('caption') && tags.include?('transcript')
+  end
+
   # Adapted from https://github.com/opencoconut/webvtt-ruby/blob/e07d59220260fce33ba5a0c3b355e3ae88b99457/lib/webvtt/parser.rb#L11-L30
   def self.convert_from_srt(srt)
     # normalize timestamps in srt

app/views/media_objects/_supplemental_files_list.html.erb

@@ -18,22 +18,26 @@ Unless required by applicable law or agreed to in writing, software distributed
             <div class="file_view <% if tag == "caption" %>captions<% end %>">
               <% if tag == "caption" %>
                 <div class="row supplemental-file-data edit-item" id="edit-label-row">
-                  <div class="col-md-4 col-4">
-                    <label class="caption-form-label">Label</label>
-                  </div>
-                  <div class="col-md-4 col-4">
-                    <label class="caption-form-label">Language</label>
+                  <div class="row col-sm-9 col-md-8 col-9">
+                    <div class="col-md-4 col-4">
+                      <label class="caption-form-label">Label</label>
+                    </div>
+                    <div class="col-md-4 col-4">
+                      <label class="caption-form-label">Language</label>
+                    </div>
                   </div>
                 </div>
               <% end %>
               <% files.each do |file| %>
+                <% next if tag == 'transcript' and file.tags.include?('caption') %>
                 <div class="supplemental-file-data" data-file-id="<%= file.id %>" data-masterfile-id="<%= section.id %>" data-tag="<%= tag %>" >
                     <span name="label_<%= section.id + "_" + file.id.to_s %>" class="display-item"><%= file.label %></span>
                     <%= form_for :supplemental_file, url: object_supplemental_file_path(section, file), remote: true,
                           html: { method: "put", class: "supplemental-file-form edit-item", id: "form-#{file.id}" },
                           data: { file_id: file.id, masterfile_id: section.id } do |form| %>
                         <div class="col-sm-9 col-md-8 col-9 form-row p-0">
-                          <div class="form-group col-md-6 col-sm-6 col-6">
+                          <% col = tag == 'caption' ? 4 : 6 %>
+                          <div class="form-group col-md-<%= col %> col-sm-<%= col %> col-<%= col %>">
                             <%= form.text_field :label, id: "supplemental_file_input_#{section.id}_#{file.id}", value: file.label %>
                           </div>
                           <% if tag == 'transcript' %>
@@ -45,11 +49,17 @@ Unless required by applicable law or agreed to in writing, software distributed
                             </span>
                           <% end %>
                           <% if tag == 'caption' %>
-                            <div class="form-group col-md-6 col p-0">
+                            <div class="form-group col-md-4 col p-0">
                               <%= form.text_field :language, id: "supplemental_file_language_#{section.id}_#{file.id}", value: LanguageTerm.find(file.language).text,
                                   class: "typeahead from-model form-control",
                                   data: { model: 'languageTerm', validate: false } %>
                             </div>
+                            <span class="form-group col-md-4 col-sm-4 col-4 p-0">
+                              <%= label_tag "treat_as_transcript_#{file.id}", class: "ml-3" do %>
+                                <%= check_box_tag "treat_as_transcript_#{file.id}", '1', file.caption_transcript? %>
+                                Treat as transcript
+                              <% end %>
+                            </span>
                           <% end %>
                         </div>
                         <div class="col-sm-3 col-md-4 col-3 p-0">

spec/models/iiif_canvas_presenter_spec.rb

@@ -269,14 +269,20 @@
         expect(subject.any? { |content| content.body_id =~ /master_files\/#{master_file.id}\/captions/ }).to eq false
       end
 
-      context 'srt captions' do
+      context 'srt files' do
         let(:srt_caption_file) { FactoryBot.create(:supplemental_file, :with_caption_srt_file, :with_caption_tag) }
-        let(:supplemental_files) { [supplemental_file, transcript_file, caption_file, srt_caption_file] }
-        it 'sets format to "text/vtt"' do
+        let(:srt_transcript_file) { FactoryBot.create(:supplemental_file, :with_caption_srt_file, :with_transcript_tag) }
+        let(:supplemental_files) { [transcript_file, caption_file, srt_caption_file, srt_transcript_file] }
+        it 'sets caption format to "text/vtt"' do
           captions = subject.select { |s| s.body_id.include?('captions') }
           expect(captions.none? { |content| content.format == 'text/srt' }).to eq true
           expect(captions.all? { |content| content.format == 'text/vtt' }).to eq true
         end
+        it 'sets other formats to the original file content_type' do
+          supplemental_files = subject.reject { |s| s.body_id.include?('captions') }
+          expect(supplemental_files[0].format).to eq transcript_file.file.content_type
+          expect(supplemental_files[1].format).to eq srt_transcript_file.file.content_type
+        end
       end
 
       context 'legacy master file captions' do
@@ -298,6 +304,22 @@
           expect(subject.any? { |content| content.label['eng'][0] =~ /#{transcript_file.label} \(machine generated\)/ }).to eq true
         end
       end
+
+      context 'caption being treated as a transcript' do
+        let(:caption_file) { FactoryBot.create(:supplemental_file, :with_caption_file, tags: ['caption', 'transcript']) }
+        let(:srt_caption_file) { FactoryBot.create(:supplemental_file, :with_caption_srt_file, tags: ['caption', 'transcript']) }
+        let(:supplemental_files) { [caption_file, srt_caption_file] }
+
+        it 'returns a caption entry and a transcript entry with proper formats' do
+          captions = subject.select { |s| s.body_id.include?('captions') }
+          transcripts = subject.select { |s| s.body_id.include?('transcripts') }
+          expect(captions.count).to eq 2
+          expect(transcripts.count).to eq 2
+          expect(captions.all? { |content| content.format == 'text/vtt' }).to eq true
+          expect(transcripts.any? { |content| content.format == 'text/vtt' }).to eq true
+          expect(transcripts.any? { |content| content.format == 'text/srt' }).to eq true
+        end
+      end
     end
   end
 

spec/support/supplemental_files_controller_examples.rb

@@ -292,6 +292,34 @@
         end
       end
 
+      context "caption treated as transcript" do
+        let(:transcript_param) { "treat_as_transcript_#{supplemental_file.id}".to_sym }
+        context "missing transcript tag" do
+          let(:supplemental_file) { FactoryBot.create(:supplemental_file, :with_caption_file, :with_caption_tag, label: 'label') }
+          it "adds transcript note to tags" do
+            expect {
+              put :update, params: { class_id => object.id, id: supplemental_file.id, transcript_param => 1, supplemental_file: valid_update_attributes, format: :html }, session: valid_session
+            }.to change { master_file.reload.supplemental_files.first.tags }.from(['caption']).to(['caption', 'transcript'])
+          end
+        end
+        context "with transcript tag" do
+          let(:supplemental_file) { FactoryBot.create(:supplemental_file, :with_transcript_file, tags: ['caption', 'transcript'], label: 'label') }
+          it "does not add more instances of transcript note" do
+            expect {
+              put :update, params: { class_id => object.id, id: supplemental_file.id, transcript_param => 1, supplemental_file: valid_update_attributes, format: :html }, session: valid_session
+            }.to not_change { master_file.reload.supplemental_files.first.tags }.from(['caption', 'transcript'])
+          end
+        end
+        context "removing transcript designation" do
+          let(:supplemental_file) { FactoryBot.create(:supplemental_file, :with_transcript_file, tags: ['caption', 'transcript'], label: 'label (machine generated)') }
+          it "removes transcript note from tags" do
+            expect {
+              put :update, params: { class_id => object.id, id: supplemental_file.id, supplemental_file: valid_update_attributes, format: :html }, session: valid_session
+            }.to change { master_file.reload.supplemental_files.first.tags }.from(['caption', 'transcript']).to(['caption'])
+          end
+        end
+      end
+
       context "with invalid params" do
         it "returns a 400" do
           put :update, params: { class_id => object.id, id: supplemental_file.id, supplemental_file: invalid_update_attributes, format: :html }, session: valid_session