Merge pull request #5719 from avalonmediasystem/transform_srt

Convert SRT to VTT captions for cosumption into Video.js

app/controllers/master_files_controller.rb

@@ -286,7 +286,7 @@ def caption_manifest
     @caption_url = if caption_id == 'master_file_caption'
                      captions_master_file_path
                    else
-                     master_file_supplemental_file_path(master_file_id: @master_file.id, id: caption_id)
+                     captions_master_file_supplemental_file_path(master_file_id: @master_file.id, id: caption_id)
                    end
   end
 

app/controllers/supplemental_files_controller.rb

@@ -58,12 +58,7 @@ def create
   end
 
   def show
-    # TODO: Use a master file presenter which reads from solr instead of loading the masterfile from fedora
-    # FIXME: authorize supplemental file directly (needs supplemental file to have reference to masterfile)
-    raise Avalon::NotFound, "Supplemental file: #{params[:id]} not found" unless SupplementalFile.exists? params[:id].to_s
-
-    @supplemental_file = SupplementalFile.find(params[:id])
-    raise Avalon::NotFound, "Supplemental file: #{@supplemental_file.id} not found" unless @object.supplemental_files.any? { |f| f.id == @supplemental_file.id }
+    find_supplemental_file
 
     # Redirect or proxy the content
     if Settings.supplemental_files.proxy
@@ -107,6 +102,15 @@ def destroy
     end
   end
 
+  def captions
+    find_supplemental_file
+
+    file_content = @supplemental_file.file.download
+    content = @supplemental_file.file.content_type == 'text/srt' ? SupplementalFile.convert_from_srt(file_content) : file_content
+
+    send_data content, filename: @supplemental_file.file.filename.to_s, type: 'text/vtt', disposition: 'attachment'
+  end
+
   private
 
     def set_object
@@ -118,6 +122,16 @@ def supplemental_file_params
       params.fetch(:supplemental_file, {}).permit(:label, :language, :file, tags: [])
     end
 
+    def find_supplemental_file
+      # TODO: Use a master file presenter which reads from solr instead of loading the masterfile from fedora
+      # FIXME: authorize supplemental file directly (needs supplemental file to have reference to masterfile)
+      raise Avalon::NotFound, "Supplemental file: #{params[:id]} not found" unless SupplementalFile.exists? params[:id].to_s
+
+      @supplemental_file = SupplementalFile.find(params[:id])
+      raise Avalon::NotFound, "Supplemental file: #{@supplemental_file.id} not found" unless @object.supplemental_files.any? { |f| f.id == @supplemental_file.id }
+    end
+
+
     def handle_error(message:, status:)
       if request.format == :json
         render json: { errors: message }, status: status
@@ -158,7 +172,7 @@ def object_supplemental_file_path
     end
 
     def authorize_object
-      action = action_name.to_sym == :show ? :show : :edit
+      action = [:show, :captions].include?(action_name.to_sym) ? :show : :edit
       authorize! action, @object, message: "You do not have sufficient privileges to #{action_name} this supplemental file"
     end
 end

app/models/concerns/iiif_supplemental_file_behavior.rb

@@ -36,7 +36,7 @@ def object_supplemental_file_url(object, supplemental_file)
 
   def determine_rendering_type(mime)
     case mime
-    when 'application/pdf', 'application/msword', 'application/vnd.oasis.opendocument.text', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document', 'text/html', 'text/plain', 'text/vtt'
+    when 'application/pdf', 'application/msword', 'application/vnd.oasis.opendocument.text', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document', 'text/html', 'text/plain', 'text/srt', 'text/vtt'
       'Text'
     when /image\/.+/
       'Image'

app/models/supplemental_file.rb

@@ -45,4 +45,20 @@ def caption?
   def machine_generated?
     tags.include?('machine_generated')
   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
+    # This Regex looks for malformed time stamp pieces such as '00:1:00,000', '0:01:00,000', etc.
+    # When it finds a match it prepends a 0 to the capture group so both of the above examples 
+    # would return '00:01:00,000'
+    conversion = srt.gsub(/(:|^)(\d)(,|:)/, '\10\2\3')
+    # convert timestamps and save the file
+    # VTT uses '.' as its decimal separator, SRT uses ',' so we convert the punctuation
+    conversion.gsub!(/([0-9]{2}:[0-9]{2}:[0-9]{2})([,])([0-9]{3})/, '\1.\3')
+    # normalize new line character
+    conversion.gsub!("\r\n", "\n")
+
+    "WEBVTT\n\n#{conversion}".strip
+  end
 end

config/routes.rb

@@ -166,7 +166,7 @@
     # Supplemental Files
     resources :supplemental_files, except: [:new, :index, :edit] do
       member do
-        get 'captions', :to => redirect('/master_files/%{master_file_id}/supplemental_files/%{id}')
+        get 'captions'
         get 'transcripts', :to => redirect('/master_files/%{master_file_id}/supplemental_files/%{id}')
       end
     end

spec/controllers/supplemental_files_controller_spec.rb

@@ -17,4 +17,55 @@
 RSpec.describe SupplementalFilesController, type: :controller do
   it_behaves_like "a nested controller for", MasterFile
   it_behaves_like "a nested controller for", MediaObject
+
+  describe 'captions endpoint for MasterFile' do
+    let(:supplemental_file) { FactoryBot.create(:supplemental_file, :with_caption_file, :with_caption_tag) }
+    # This should return the minimal set of values that should be in the session
+    # in order to pass any filters (e.g. authentication) defined in
+    # SupplementalFilesController. Be sure to keep this updated too.
+    let(:valid_session) { {} }
+
+
+    describe 'security' do
+      let(:master_file) { FactoryBot.create(:master_file, :with_media_object, supplemental_files: [supplemental_file]) }
+
+      context 'with unauthenticated user' do
+        it 'should return 401' do
+          expect(get :captions, params: { master_file_id: master_file.id, id: supplemental_file.id }).to have_http_status(401)
+        end
+      end
+      context 'with end-user without permissions' do
+        before do
+          login_as :user
+        end
+        it 'should return 401' do
+          expect(get :captions, params: { master_file_id: master_file.id, id: supplemental_file.id }).to have_http_status(401)
+        end
+      end
+    end
+
+    describe "GET #captions" do
+      let(:public_media_object) { FactoryBot.create(:fully_searchable_media_object) }
+      let(:master_file) { FactoryBot.create(:master_file, media_object: public_media_object, supplemental_files: [supplemental_file]) }
+      before { allow(Settings.supplemental_files).to receive(:proxy).and_return(true) }
+
+      it "returns the caption file content" do
+        get :captions, params: {  master_file_id: master_file.id, id: supplemental_file.id }, session: valid_session
+        expect(response).to have_http_status(200)
+        expect(response.header["Content-Type"]).to eq 'text/vtt'
+        expect(response.body).to eq supplemental_file.file.download
+      end
+
+      context 'with SRT caption' do
+        let(:supplemental_file) { FactoryBot.create(:supplemental_file, :with_caption_tag, :with_caption_srt_file) }
+        let(:file) { Rails.root.join('spec', 'fixtures', 'captions.srt')}
+        it 'returns the caption file content in VTT format' do
+          get :captions, params: {  master_file_id: master_file.id, id: supplemental_file.id }, session: valid_session
+          expect(response).to have_http_status(200)
+          expect(response.header["Content-Type"]).to eq 'text/vtt'
+          expect(response.body).to eq SupplementalFile.convert_from_srt(File.read(file))
+        end
+      end
+    end
+  end
 end

spec/models/supplemental_file_spec.rb

@@ -90,4 +90,12 @@
       expect(subject.reload.language).to eq "ger"
     end
   end
+
+  describe '.convert_from_srt' do
+    let(:input) { "1\n00:00:03,498 --> 00:00:05,000\n- Example Captions\n" }
+    let(:output) { "WEBVTT\n\n1\n00:00:03.498 --> 00:00:05.000\n- Example Captions" }
+    it 'converts SRT format captions into VTT captions' do
+      expect(SupplementalFile.convert_from_srt(input)).to eq output
+    end
+  end
 end

spec/routing/supplemental_files_routing_spec.rb

@@ -28,11 +28,12 @@
     it "routes to #update" do
       expect(:put => "/master_files/abc1234/supplemental_files/edf567").to route_to("supplemental_files#update", master_file_id: 'abc1234', id: 'edf567')
     end
+    it "routes to #captions" do
+      expect(:get => "/master_files/abc1234/supplemental_files/edf567/captions").to route_to("supplemental_files#captions", master_file_id: 'abc1234', id: 'edf567')
+    end
     # Redirects are not testable from the routing spec out of the box.
     # Forcing the tests to `type: :request` to keep routing tests in one place.
     it "redirects to supplemental_files#show", type: :request do
-      get "/master_files/abc1234/supplemental_files/edf567/captions"
-      expect(response).to redirect_to("/master_files/abc1234/supplemental_files/edf567")
       get "/master_files/abc1234/supplemental_files/edf567/transcripts"
       expect(response).to redirect_to("/master_files/abc1234/supplemental_files/edf567")
     end