Import files from CSV

Working solution

Refactor

Tests

lib/assets/csv_test.csv

@@ -0,0 +1,3 @@
+id,type,title,description,resource_type,contributor,contributor,date_created,file
+11111111-2465-1111-5468-000123456789,ETD,Hammock Tacos,,image,Tracy S Gertler,Cynthia V Stack,2015-03-29T22:12:12.100363+00:00,example3.tiff
+22222222-0971-2222-1353-000123456789,ETD,Freegan Intelligentsia,,text,Tracy S Gertler,Cynthia V Stack,2015-03-29T22:12:12.100363+00:00,sample.docx

lib/importer/attach_files_to_work.rb

@@ -0,0 +1,14 @@
+# Called in 'object_factory'
+module Importer
+  class AttachFilesToWork < AttachFilesToWorkJob
+    # Overrides hyrax/app/jobs/attach_files_to_work_job because
+    # the original `perform` method iterates through an array of files and gives a NoMethod Error `each`
+    def perform(work, uploaded_file, actor)
+      actor.file_set.permissions_attributes = work.permissions.map(&:to_hash)
+      actor.create_metadata
+      actor.create_content(uploaded_file)
+      actor.attach_to_work(work)
+      uploaded_file.update(file_set_uri: actor.file_set.uri)
+    end
+  end
+end

lib/importer/factory/collection_factory.rb

@@ -16,7 +16,7 @@ def update
         run_callbacks(:save) do
           object.save!
         end
-        log_updated(object)
+        # log_updated(object)
       end
     end
   end

lib/importer/factory/image_factory.rb

@@ -3,7 +3,7 @@ module Factory
     class ImageFactory < ObjectFactory
       include WithAssociatedCollection
 
-      self.klass = GenericWork
+      self.klass = Image
       # A way to identify objects that are not Hydra minted identifiers
       self.system_identifier_field = :identifier
 

lib/importer/factory/object_factory.rb

@@ -1,4 +1,5 @@
 require 'importer/log_subscriber'
+require 'importer/attach_files_to_work'
 module Importer
   module Factory
     class ObjectFactory
@@ -25,12 +26,13 @@ def run
         object
       end
 
+      ## FOR CONSIDERATION: handle a row (i.e. Work) with more than one file:
+      ## currently the file_set is replaced on update
       def update
         raise "Object doesn't exist" unless object
-        run_callbacks(:save) do
-          work_actor.update(environment(update_attributes))
-        end
-        log_updated(object)
+        work_actor.update(environment(update_attributes))
+        destroy_existing_file_set if object.file_sets.present?
+        attach_file_to_work
       end
 
       def create_attributes
@@ -64,23 +66,21 @@ def search_by_identifier
       def create
         attrs = create_attributes
         @object = klass.new
-        run_callbacks :save do
-          run_callbacks :create do
-            klass == Collection ? create_collection(attrs) : work_actor.create(environment(attrs))
-          end
-        end
-        log_created(object)
+        klass == Collection ? create_collection(attrs) : work_actor.create(environment(attrs))
+        attach_file_to_work
       end
 
-      def log_created(obj)
-        msg = "Created #{klass.model_name.human} #{obj.id}"
-        Rails.logger.info("#{msg} (#{Array(attributes[system_identifier_field]).first})")
-      end
-
-      def log_updated(obj)
-        msg = "Updated #{klass.model_name.human} #{obj.id}"
-        Rails.logger.info("#{msg} (#{Array(attributes[system_identifier_field]).first})")
-      end
+      ## below methods are commented out to pass rubocop inspection ("Class has too many line");
+      ## the logs already mention the CREATE and UPDATE actions once done
+      # def log_created(obj)
+      #   msg = "Created #{klass.model_name.human} #{obj.id}"
+      #   Rails.logger.info("#{msg} (#{Array(attributes[system_identifier_field]).first})")
+      # end
+      #
+      # def log_updated(obj)
+      #   msg = "Updated #{klass.model_name.human} #{obj.id}"
+      #   Rails.logger.info("#{msg} (#{Array(attributes[system_identifier_field]).first})")
+      # end
 
       private
 
@@ -94,6 +94,10 @@ def work_actor
           Hyrax::CurationConcern.actor
         end
 
+        def file_set_actor(w)
+          Hyrax::Actors::FileSetActor.new(FileSet.create, w.user)
+        end
+
         def create_collection(attrs)
           @object.attributes = attrs
           @object.apply_depositor_metadata(User.batch_user)
@@ -107,14 +111,33 @@ def transform_attributes
                                 .merge(file_attributes)
         end
 
-        # NOTE: This approach is probably broken since the actor that handled `:files` attribute was removed:
-        # https://github.com/samvera/hyrax/commit/3f1b58195d4381c51fde8b9149016c5b09f0c9b4
         def file_attributes
           files_directory.present? && files.present? ? { files: file_paths } : {}
         end
 
         def file_paths
-          files.map { |file_name| File.join(files_directory, file_name) }
+          attributes[:file].map { |file_name| File.join(files_directory, file_name) } if attributes[:file]
+        end
+
+        def import_file(path)
+          u = Hyrax::UploadedFile.new
+          u.user_id = User.find_by_user_key(User.batch_user_key).id if User.find_by_user_key(User.batch_user_key)
+          u.file = CarrierWave::SanitizedFile.new(path)
+          u.save
+          u
+        end
+
+        ## If no file name is provided in the CSV file, `attach_file_to_work` is not performed
+        ## TO DO: handle invalid file in CSV
+        ## currently the importer stops if no file corresponding to a given file_name is found
+        def attach_file_to_work
+          imported_file = import_file(file_paths.first) if file_paths
+          AttachFilesToWork.new.perform(object, imported_file, file_set_actor(imported_file)) if imported_file
+        end
+
+        def destroy_existing_file_set
+          f = object.file_sets.first
+          f.destroy if attributes[:file] != f.title
         end
 
         # Regardless of what the MODS Parser gives us, these are the properties we are prepared to accept.

lib/importer/factory/with_associated_collection.rb

@@ -13,6 +13,7 @@ def create_attributes
       # Strip out the :collection key, and add the member_of_collection_ids,
       # which is used by Hyrax::Actors::AddAsMemberOfCollectionsActor
       def update_attributes
+        return super if attributes[:collection].nil?
         super.except(:collection).merge(member_of_collection_ids: [collection.id])
       end
 

spec/lib/importer/factory/etd_factory_spec.rb

@@ -2,52 +2,50 @@
 
 RSpec.describe Importer::Factory::ETDFactory, :clean do
   let(:factory) { described_class.new(attributes) }
-  let(:actor) { double }
-  before do
-    allow(Hyrax::CurationConcern).to receive(:actor).and_return(actor)
-  end
   let(:files) { [] }
-  let(:attributes) do
-    {
-      collection: { id: coll.id },
-      files: files,
-      identifier: ['123'],
-      title: ['Test image'],
-      read_groups: ['public'],
-      depositor: 'bob',
-      edit_users: ['bob']
-    }
-  end
-
-  context 'with files' do
-    let(:factory) { described_class.new(attributes, 'tmp/files', files) }
-    let(:files) { ['img.png'] }
-    let(:coll) { create(:collection) }
-
-    context "for a new image" do
-      it 'calls the actor with the files' do
-        expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
-          expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
-        end
-        factory.run
-      end
-    end
+  let(:work) { GenericWork }
 
-    context "for an existing image without files" do
-      let(:work) { create(:generic_work) }
-      let(:factory) { described_class.new(attributes.merge(id: work.id), 'tmp/files', files) }
-
-      it 'creates file sets' do
-        expect(actor).to receive(:update).with(Hyrax::Actors::Environment) do |k|
-          expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
-        end
-        factory.run
-      end
-    end
-  end
+  # context 'with files' do
+  #   let(:factory) { described_class.new(attributes, 'tmp/files', files) }
+  #   let(:files) { ['img.png'] }
+  #   let(:coll) { create(:collection) }
+  #
+  #   context "for a new image" do
+  #     it 'calls the actor with the files' do
+  #       expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
+  #         expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
+  #       end
+  #       factory.run
+  #     end
+  #   end
+  #
+  #   context "for an existing image without files" do
+  #     let(:work) { create(:generic_work) }
+  #     let(:factory) { described_class.new(attributes.merge(id: work.id), 'tmp/files', files) }
+  #
+  #     it 'creates file sets' do
+  #       expect(actor).to receive(:update).with(Hyrax::Actors::Environment) do |k|
+  #         expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
+  #       end
+  #       factory.run
+  #     end
+  #   end
+  # end
 
   context 'when a collection already exists' do
     let!(:coll) { create(:collection) }
+    let(:attributes) do
+      {
+        collection: { id: coll.id },
+        files: files,
+        identifier: ['123'],
+        title: ['Test image'],
+        read_groups: ['public'],
+        depositor: 'bob',
+        edit_users: ['bob']
+      }
+    end
+    let(:actor) { Hyrax::CurationConcern.actor }
 
     it 'does not create a new collection' do
       expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
@@ -58,4 +56,6 @@
       end.to change(Collection, :count).by(0)
     end
   end
+
+  include_examples("csv_importer")
 end

spec/lib/importer/factory/image_factory_spec.rb

@@ -2,52 +2,50 @@
 
 RSpec.describe Importer::Factory::ImageFactory, :clean do
   let(:factory) { described_class.new(attributes) }
-  let(:actor) { double }
-  before do
-    allow(Hyrax::CurationConcern).to receive(:actor).and_return(actor)
-  end
   let(:files) { [] }
-  let(:attributes) do
-    {
-      collection: { id: coll.id },
-      files: files,
-      identifier: ['123'],
-      title: ['Test image'],
-      read_groups: ['public'],
-      depositor: 'bob',
-      edit_users: ['bob']
-    }
-  end
-
-  context 'with files' do
-    let(:factory) { described_class.new(attributes, 'tmp/files', files) }
-    let(:files) { ['img.png'] }
-    let!(:coll) { create(:collection) }
-
-    context "for a new image" do
-      it 'creates file sets with access controls' do
-        expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
-          expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
-        end
-        factory.run
-      end
-    end
+  let(:work) { Image }
 
-    context "for an existing image without files" do
-      let(:work) { create(:generic_work) }
-      let(:factory) { described_class.new(attributes.merge(id: work.id), 'tmp/files', files) }
-
-      it 'creates file sets' do
-        expect(actor).to receive(:update).with(Hyrax::Actors::Environment) do |k|
-          expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
-        end
-        factory.run
-      end
-    end
-  end
+  # context 'with files' do
+  #   let(:factory) { described_class.new(attributes, 'tmp/files', files) }
+  #   let(:files) { ['img.png'] }
+  #   let!(:coll) { create(:collection) }
+  #
+  #   context "for a new image" do
+  #     it 'creates file sets with access controls' do
+  #       expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
+  #         expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
+  #       end
+  #       factory.run
+  #     end
+  #   end
+  #
+  #   context "for an existing image without files" do
+  #     let(:work) { create(:generic_work) }
+  #     let(:factory) { described_class.new(attributes.merge(id: work.id), 'tmp/files', files) }
+  #
+  #     it 'creates file sets' do
+  #       expect(actor).to receive(:update).with(Hyrax::Actors::Environment) do |k|
+  #         expect(k.attributes).to include(member_of_collection_ids: [coll.id], files: ['tmp/files/img.png'])
+  #       end
+  #       factory.run
+  #     end
+  #   end
+  # end
 
   context 'when a collection already exists' do
     let!(:coll) { create(:collection) }
+    let(:attributes) do
+      {
+        collection: { id: coll.id },
+        files: files,
+        identifier: ['123'],
+        title: ['Test image'],
+        read_groups: ['public'],
+        depositor: 'bob',
+        edit_users: ['bob']
+      }
+    end
+    let(:actor) { Hyrax::CurationConcern.actor }
 
     it 'does not create a new collection' do
       expect(actor).to receive(:create).with(Hyrax::Actors::Environment) do |k|
@@ -58,4 +56,6 @@
       end.to change(Collection, :count).by(0)
     end
   end
+
+  include_examples("csv_importer")
 end