Merge pull request #5758 from avalonmediasystem/caption_transcript_facet

Add 'Has Captions' and 'Has Transcripts' facets

Author: masaball

app/controllers/catalog_controller.rb

@@ -88,6 +88,8 @@ class CatalogController < ApplicationController
     config.add_facet_field 'collection_ssim', label: 'Collection', limit: 5
     config.add_facet_field 'unit_ssim', label: 'Unit', limit: 5
     config.add_facet_field 'language_ssim', label: 'Language', limit: 5
+    config.add_facet_field 'has_captions_bsi', label: 'Has Captions', helper_method: :display_has_caption_or_transcript
+    config.add_facet_field 'has_transcripts_bsi', label: 'Has Transcripts', helper_method: :display_has_caption_or_transcript
     # Hide these facets if not a Collection Manager
     config.add_facet_field 'workflow_published_sim', label: 'Published', limit: 5, if: Proc.new {|context, config, opts| Ability.new(context.current_user, context.user_session).can? :create, MediaObject}, group: "workflow"
     config.add_facet_field 'avalon_uploader_ssi', label: 'Created by', limit: 5, if: Proc.new {|context, config, opts| Ability.new(context.current_user, context.user_session).can? :create, MediaObject}, group: "workflow"

app/helpers/application_helper.rb

@@ -104,6 +104,10 @@ def display_metadata(label, value, default=nil)
     content_tag(:dt, label) + contents
   end
 
+  def display_has_caption_or_transcript value
+    value = value == "true" ? 'Yes' : 'No'
+  end
+
   def search_result_label item
     if item['title_tesi'].present?
       label = truncate(item['title_tesi'], length: 100)

app/models/concerns/master_file_behavior.rb

@@ -171,6 +171,10 @@ def supplemental_file_captions
     supplemental_files(tag: 'caption')
   end
 
+  def supplemental_file_transcripts
+    supplemental_files(tag: 'transcript')
+  end
+
   # Supplies the route to the master_file as an rdf formatted URI
   # @return [String] the route as a uri
   # @example uri for a mf on avalon.iu.edu with a id of: avalon:1820

app/models/concerns/media_object_behavior.rb

@@ -57,6 +57,14 @@ def access_text
     "This item is accessible by: #{actors.join(', ')}."
   end
 
+  def has_captions
+    master_files.any? { |mf| mf.has_captions? }
+  end
+
+  def has_transcripts
+    master_files.any? { |mf| mf.has_transcripts? }
+  end
+
   # CDL methods
   def lending_status
     Checkout.active_for_media_object(id).any? ? "checked_out" : "available"

app/models/master_file.rb

@@ -492,6 +492,10 @@ def has_captions?
     !captions.empty? || !supplemental_file_captions.empty?
   end
 
+  def has_transcripts?
+    supplemental_file_transcripts.present?
+  end
+
   def has_waveform?
     !waveform.empty?
   end
@@ -508,6 +512,7 @@ def to_solr *args
     super.tap do |solr_doc|
       solr_doc['file_size_ltsi'] = file_size if file_size.present?
       solr_doc['has_captions?_bs'] = has_captions?
+      solr_doc['has_transcripts?_bs'] = has_transcripts?
       solr_doc['has_waveform?_bs'] = has_waveform?
       solr_doc['has_poster?_bs'] = has_poster?
       solr_doc['has_thumbnail?_bs'] = has_thumbnail?

app/models/media_object.rb

@@ -236,6 +236,8 @@ def fill_in_solr_fields_that_need_master_files(solr_doc)
     solr_doc['section_id_ssim'] = ordered_master_file_ids
     solr_doc["other_identifier_sim"] +=  master_files.collect {|mf| mf.identifier.to_a }.flatten
     solr_doc["date_digitized_ssim"] = master_files.collect {|mf| mf.date_digitized }.compact.map {|t| Time.parse(t).strftime "%F" }
+    solr_doc["has_captions_bsi"] = has_captions
+    solr_doc["has_transcripts_bsi"] = has_transcripts
     solr_doc["section_label_tesim"] = section_labels
     solr_doc['section_physical_description_ssim'] = section_physical_descriptions
     solr_doc['all_comments_ssim'] = all_comments

spec/controllers/catalog_controller_spec.rb

@@ -232,6 +232,28 @@
       end
     end
 
+    describe "facet fields" do
+      let(:media_object) { FactoryBot.create(:fully_searchable_media_object, :with_master_file, :with_completed_workflow, avalon_uploader: 'archivist1', governing_policies: [lease]) }
+      let(:lease) { FactoryBot.create(:lease, inherited_read_groups: ['ExternalGroup']) }
+      before(:each) do
+        MediaObjectIndexingJob.perform_now(media_object.id)
+      end
+      ["avalon_resource_type_ssim", "creator_ssim", "date_sim", "genre_ssim", "series_ssim", "collection_ssim", "unit_ssim", "language_ssim", "has_captions_bsi", "has_transcripts_bsi",
+       "workflow_published_sim", "avalon_uploader_ssi", "read_access_group_ssim", "read_access_virtual_group_ssim", "date_digitized_ssim", "date_ingested_ssim"].each do |field|  
+        it "should facet results on #{field}" do
+          query = Array(media_object.to_solr(include_child_fields:true)[field]).first
+          # The following line is to check that the test is using a valid solr field name
+          # since an incorrect one will lead to an empty query resulting in a false positive below
+          expect(query.to_s).not_to be_empty
+          get :index, params: { 'f' => { field => [query] } }
+          expect(response).to be_successful
+          expect(response).to render_template('catalog/index')
+          expect(assigns(:response).documents.count).to eq 1
+          expect(assigns(:response).documents.map(&:id)).to contain_exactly(media_object.id)
+        end
+      end
+    end
+
     describe "gated discovery" do
       context "with bad ldap groups" do
         let(:ldap_groups) { ['good-group', 'bad group'] }

spec/helpers/application_helper_spec.rb

@@ -163,6 +163,16 @@
     end
   end
 
+  describe "#display_has_caption_or_transcript" do
+    it "should return 'Yes' for 'true'" do
+      expect(helper.display_has_caption_or_transcript("true")).to eq("Yes")
+    end
+
+    it "should return 'No' for 'false'" do
+      expect(helper.display_has_caption_or_transcript("false")).to eq("No")
+    end
+  end
+
   describe "#pretty_time" do
     it 'returns a formatted time' do
       expect(helper.pretty_time(0)).to eq '00:00:00.000'

spec/models/master_file_spec.rb

@@ -721,14 +721,26 @@
     let(:caption_file) { FactoryBot.create(:supplemental_file, :with_caption_file, :with_caption_tag) }
     let(:transcript_file) { FactoryBot.create(:supplemental_file, :with_transcript_tag) }
     let(:master_file) { FactoryBot.create(:master_file, supplemental_files: [caption_file, transcript_file]) }
-    it 'has a caption' do
+    it 'returns only caption files' do
       expect(master_file.supplemental_file_captions).to_not be_empty
       expect(master_file.supplemental_file_captions).to all(be_kind_of(SupplementalFile))
       expect(master_file.supplemental_file_captions).to_not include(transcript_file)
       expect(master_file.supplemental_file_captions).to include(caption_file)
     end
   end
 
+  describe 'supplemental_file_transcripts' do
+    let(:caption_file) { FactoryBot.create(:supplemental_file, :with_caption_file, :with_caption_tag) }
+    let(:transcript_file) { FactoryBot.create(:supplemental_file, :with_transcript_tag) }
+    let(:master_file) { FactoryBot.create(:master_file, supplemental_files: [caption_file, transcript_file]) }
+    it 'returns only transcript files' do
+      expect(master_file.supplemental_file_transcripts).to_not be_empty
+      expect(master_file.supplemental_file_transcripts).to all(be_kind_of(SupplementalFile))
+      expect(master_file.supplemental_file_transcripts).to include(transcript_file)
+      expect(master_file.supplemental_file_transcripts).to_not include(caption_file)
+    end
+  end
+
   describe 'waveforms' do
     let(:master_file) { FactoryBot.create(:master_file) }
     it 'sets original_name to default value' do

spec/models/media_object_spec.rb

@@ -1162,4 +1162,33 @@
       expect(MediaObject.autocomplete('te', mo1.id)).to include({ id: 'Test 2', display: 'Test 2' })
     end
   end
+
+  describe "#has_captions" do
+    let(:captionless_media_object) { FactoryBot.create(:media_object, :with_master_file) }
+    let(:captioned_media_object) { FactoryBot.create(:media_object, master_files: [master_file1, master_file2]) }
+    let(:master_file1) { FactoryBot.create(:master_file) }
+    let(:master_file2) { FactoryBot.create(:master_file, :with_captions) }
+    it "returns false when child master files contain no captions" do
+      expect(captionless_media_object.has_captions).to be false
+    end
+
+    it "returns true when any child master file contains a caption" do
+      expect(captioned_media_object.has_captions).to be true
+    end
+  end
+
+  describe "#has_transcripts" do
+    let(:transcriptless_media_object) { FactoryBot.create(:media_object, :with_master_file) }
+    let(:transcript_media_object) { FactoryBot.create(:media_object, master_files: [master_file1, master_file2]) }
+    let(:master_file1) { FactoryBot.create(:master_file) }
+    let(:master_file2) { FactoryBot.create(:master_file, supplemental_files: [transcript]) }
+    let(:transcript) { FactoryBot.create(:supplemental_file, :with_transcript_tag) }
+    it "returns false when child master files contain no transcript" do
+      expect(transcriptless_media_object.has_transcripts).to be false
+    end
+
+    it "returns true when any child master file contains a transcript" do
+      expect(transcript_media_object.has_transcripts).to be true
+    end
+  end
 end