handle multiple caption files (#1765)

app/components/embed/media_tag_component.rb

@@ -5,6 +5,16 @@ class MediaTagComponent < ViewComponent::Base # rubocop:disable Metrics:ClassLen
     include StacksImage
     with_collection_parameter :resource
     SUPPORTED_MEDIA_TYPES = %i[audio video].freeze
+    # Hardcoding some language code to label mappings, based on the mappings we currently need.
+    # This approach should be revisited once we have more robust BCP 47 code to label mapping integrated.
+    FILE_LANGUAGE_CAPTION_LABELS = {
+      'en' => 'English',
+      'ru' => 'Russian',
+      'de' => 'German',
+      'et' => 'Estonian',
+      'lv' => 'Latvian',
+      'es' => 'Spanish'
+    }.freeze
 
     # @param [Purl::Resource] resource This resource is expected to have a primary file.
     # @param [#index] resource_iteration Information about what part of the collection we are in
@@ -102,14 +112,34 @@ def enabled_streaming_sources
       )
     end
 
+    # Generate the video caption elements
     def transcript
       return unless render_captions?
 
-      tag.track(src: @resource.vtt.file_url, kind: 'captions', srclang: 'en', label: 'English')
+      # A video clip may have multiple caption files in different languages.
+      # We want to enable the user to select from any of these options.
+      # We also want the different language options to be listed alphabetically.
+      safe_join(
+        sort_caption_tracks(@resource.caption_files).map do |caption_file|
+          lang_code = caption_file.language || 'en'
+          lang_label = caption_language(lang_code)
+          tag.track(src: caption_file.file_url, kind: 'captions', srclang: lang_code, label: lang_label)
+        end
+      )
+    end
+
+    # Sort the caption files by language label so we can display them in alphabetical order in the
+    # captions options list.
+    def sort_caption_tracks(caption_files)
+      caption_files.sort_by { |cfile| caption_language(cfile.language) }
+    end
+
+    def caption_language(language_code)
+      FILE_LANGUAGE_CAPTION_LABELS.fetch(language_code, 'Caption')
     end
 
     def render_captions?
-      @include_transcripts && @resource.vtt
+      @include_transcripts && @resource.caption_files
     end
 
     # NOTE: This is only for the legacy media player. We can remove it when we switch to the new player.

app/models/embed/purl/file_xml_deserializer.rb

@@ -36,12 +36,14 @@ def duration
         Embed::MediaDuration.new(@file.xpath('./*[@duration]').first).to_s
       end
 
+      # rubocop:disable Metrics/AbcSize
       def deserialize
         ResourceFile.new(
           druid: @druid,
           label: @description,
           mimetype: @file.attributes['mimetype']&.value,
           size: @file.attributes['size']&.value.to_i,
+          language: @file.attributes['language']&.value,
           filename:,
           stanford_only:,
           location_restricted:,
@@ -50,6 +52,7 @@ def deserialize
           file.duration = duration
         end
       end
+      # rubocop:enable Metrics/AbcSize
     end
   end
 end

app/models/embed/purl/resource.rb

@@ -44,9 +44,9 @@ def thumbnail
         files.find(&:image?)
       end
 
-      # @return [ResourceFile]
-      def vtt
-        files.find(&:vtt?)
+      # @return [Array<ResourceFile>]
+      def caption_files
+        files.select(&:vtt?)
       end
     end
   end

app/models/embed/purl/resource_file.rb

@@ -14,7 +14,7 @@ def attributes=(hash)
         end
       end
 
-      attr_accessor :druid, :label, :filename, :mimetype, :size, :duration,
+      attr_accessor :druid, :label, :filename, :mimetype, :size, :duration, :language,
                     :world_downloadable, :stanford_only, :location_restricted
 
       alias title filename

spec/components/embed/media_tag_component_spec.rb

@@ -223,4 +223,31 @@
       expect(page).to have_css('track[src="https://stacks.stanford.edu/file/druid:bc123df4567/abc_123_cap.webvtt"]')
     end
   end
+
+  context 'with captions for multiple languages' do
+    let(:resource) do
+      build(:resource, :video, files: [build(:resource_file, :video),
+                                       build(:resource_file, :vtt, language: 'en'),
+                                       build(:resource_file, :vtt, language: 'ru')])
+    end
+
+    let(:include_transcripts) { true }
+
+    it 'has track elements with multiple languages' do
+      expect(page).to have_css('track[srclang="en"][label="English"]')
+      expect(page).to have_css('track[srclang="ru"][label="Russian"]')
+    end
+  end
+
+  context 'with caption with no specified language' do
+    let(:resource) do
+      build(:resource, :video, files: [build(:resource_file, :video),
+                                       build(:resource_file, :vtt)])
+    end
+    let(:include_transcripts) { true }
+
+    it 'has a track element with code and label for English' do
+      expect(page).to have_css('track[srclang="en"][label="English"]')
+    end
+  end
 end

spec/fixtures/purl_fixtures.rb

@@ -848,32 +848,6 @@ def single_video_purl
     XML
   end
 
-  def video_purl_with_vtt
-    <<-XML
-      <publicObject>
-        <identityMetadata>
-          <objectLabel>Title of the single video</objectLabel>
-        </identityMetadata>
-        <contentMetadata type="media">
-          <resource sequence="1" id="abc123_1" type="video">
-            <file id="abc_123.mp4" mimetype="video/mp4" size="152000000"></file>
-            <file id="abc_123_cap.webvtt" mimetype="text/vtt" size="176218"></file>
-          </resource>
-        </contentMetadata>
-        <rightsMetadata>
-          <access type="read">
-            <machine>
-              <location>spec</location>
-            </machine>
-          </access>
-        </rightsMetadata>
-        <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/">
-          <dc:title>title of video with vtt</dc:title>
-        </oai_dc>
-      </publicObject>
-    XML
-  end
-
   def video_purl
     <<-XML
       <publicObject>

spec/models/embed/purl/resource_spec.rb

@@ -60,17 +60,17 @@
     end
   end
 
-  describe '#vtt' do
+  describe '#caption_files' do
     context 'when it has a vtt transcript' do
-      subject { resource.vtt.title }
+      subject { resource.caption_files[0].title }
 
       let(:resource) { build(:resource, :video) }
 
       it { is_expected.to eq 'abc_123_cap.webvtt' }
     end
 
     context 'when it does not have a vtt transcript' do
-      subject { resource.vtt }
+      subject { resource.caption_files[0] }
 
       let(:resource) { build(:resource, :video, files: [build(:resource_file, :video)]) }