valkyrize file_actor adding file_node to hold metadata_node values from active fedora

app/actors/hyrax/actors/file_actor.rb

@@ -1,3 +1,5 @@
+require 'wings/services/file_node_builder'
+
 module Hyrax
   module Actors
     # Actions for a file identified by file_set and relation (maps to use predicate)
@@ -8,10 +10,11 @@ class FileActor
       # @param [FileSet] file_set the parent FileSet
       # @param [Symbol, #to_sym] relation the type/use for the file
       # @param [User] user the user to record as the Agent acting upon the file
-      def initialize(file_set, relation, user)
+      def initialize(file_set, relation, user, use_valkyrie: false)
         @file_set = file_set
-        @relation = relation.to_sym
+        @relation = normalize_relation(relation, use_valkyrie: use_valkyrie)
         @user = user
+        @use_valkyrie = use_valkyrie
       end
 
       # Persists file as part of file_set and spawns async job to characterize and create derivatives.
@@ -21,16 +24,7 @@ def initialize(file_set, relation, user)
       # @see IngestJob
       # @todo create a job to monitor the temp directory (or in a multi-worker system, directories!) to prune old files that have made it into the repo
       def ingest_file(io)
-        # Skip versioning because versions will be minted by VersionCommitter as necessary during save_characterize_and_record_committer.
-        Hydra::Works::AddFileToFileSet.call(file_set,
-                                            io,
-                                            relation,
-                                            versioning: false)
-        return false unless file_set.save
-        repository_file = related_file
-        Hyrax::VersioningService.create(repository_file, user)
-        pathhint = io.uploaded_file.uploader.path if io.uploaded_file # in case next worker is on same filesystem
-        CharacterizeJob.perform_later(file_set, repository_file.id, pathhint || io.path)
+        perform_ingest_file(io, use_valkyrie: @use_valkyrie)
       end
 
       # Reverts file and spawns async job to characterize and create derivatives.
@@ -55,7 +49,72 @@ def ==(other)
 
         # @return [Hydra::PCDM::File] the file referenced by relation
         def related_file
-          file_set.public_send(relation) || raise("No #{relation} returned for FileSet #{file_set.id}")
+          file_set.public_send(normalize_relation(relation)) || raise("No #{relation} returned for FileSet #{file_set.id}")
+        end
+
+        # Persists file as part of file_set and records a new version.
+        # Also spawns an async job to characterize and create derivatives.
+        # @param [JobIoWrapper] io the file to save in the repository, with mime_type and original_name
+        # @return [FileNode, FalseClass] the created file node on success, false on failure
+        # @todo create a job to monitor the temp directory (or in a multi-worker system, directories!) to prune old files that have made it into the repo
+        def perform_ingest_file(io, use_valkyrie: false)
+          use_valkyrie ? perform_ingest_file_through_valkyrie(io) : perform_ingest_file_through_active_fedora(io)
+        end
+
+        def perform_ingest_file_through_active_fedora(io)
+          # Skip versioning because versions will be minted by VersionCommitter as necessary during save_characterize_and_record_committer.
+          Hydra::Works::AddFileToFileSet.call(file_set,
+                                              io,
+                                              relation,
+                                              versioning: false)
+          return false unless file_set.save
+          repository_file = related_file
+          Hyrax::VersioningService.create(repository_file, user)
+          pathhint = io.uploaded_file.uploader.path if io.uploaded_file # in case next worker is on same filesystem
+          CharacterizeJob.perform_later(file_set, repository_file.id, pathhint || io.path)
+        end
+
+        def perform_ingest_file_through_valkyrie(io)
+          # Skip versioning because versions will be minted by VersionCommitter as necessary during save_characterize_and_record_committer.
+          storage_adapter = Valkyrie.config.storage_adapter
+          persister = Valkyrie.config.metadata_adapter.persister # TODO: Explore why valkyrie6 branch used indexing_persister adapter for this
+          node_builder = Wings::FileNodeBuilder.new(storage_adapter: storage_adapter,
+                                                    persister: persister)
+          unsaved_node = io.to_file_node
+          unsaved_node.use = relation
+          begin
+            saved_node = node_builder.create(file: io.file, node: unsaved_node, file_set: file_set)
+          rescue StandardError => e # Handle error persisting file node
+            Rails.logger.error("Failed to save file_node through valkyrie: #{e.message}")
+            return false
+          end
+          Hyrax::VersioningService.create(saved_node, user)
+          saved_node
+        end
+
+        def normalize_relation(relation, use_valkyrie: false)
+          use_valkyrie ? normalize_relation_for_valkyrie(relation) : normalize_relation_for_active_fedora(relation)
+        end
+
+        def normalize_relation_for_active_fedora(relation)
+          return relation if relation.is_a? Symbol
+          return relation.to_sym if relation.respond_to? :to_sym
+
+          # TODO: whereever these are set, they should use Valkyrie::Vocab::PCDMUse... making the casecmp unnecessary
+          return :original_file if relation.to_s.casecmp(Valkyrie::Vocab::PCDMUse.original_file.to_s)
+          return :extracted_file if relation.to_s.casecmp(Valkyrie::Vocab::PCDMUse.extracted_file.to_s)
+          return :thumbnail_file if relation.to_s.casecmp(Valkyrie::Vocab::PCDMUse.thumbnail_file.to_s)
+          :original_file # TODO: This should never happen.  What should be done if none of the other conditions are met?
+        end
+
+        def normalize_relation_for_valkyrie(relation)
+          return relation if relation.is_a? RDF::URI
+
+          relation = relation.to_sym
+          return Valkyrie::Vocab::PCDMUse.original_file if relation == :original_file
+          return Valkyrie::Vocab::PCDMUse.extracted_file if relation == :extracted_file
+          return Valkyrie::Vocab::PCDMUse.thumbnail_file if relation == :thumbnail_file
+          Valkyrie::Vocab::PCDMUse.original_file # TODO: This should never happen.  What should be done if none of the other conditions are met?
         end
     end
   end

app/models/job_io_wrapper.rb

@@ -1,3 +1,6 @@
+require 'wings/models/file_node'
+require 'wings/valkyrie/query_service'
+
 # Primarily for jobs like IngestJob to revivify an equivalent FileActor to one that existed on
 # the caller's side of an asynchronous Job invocation.  This involves providing slots
 # for the metadata that might travel w/ the actor's various supported types of @file.
@@ -56,18 +59,36 @@ def mime_type
     super || extracted_mime_type
   end
 
-  def file_set
-    FileSet.find(file_set_id)
+  def file_set(use_valkyrie: false)
+    return FileSet.find(file_set_id) unless use_valkyrie
+    adapter = Valkyrie.config.metadata_adapter
+    query_service = Wings::Valkyrie::QueryService.new(adapter: adapter)
+    query_service.find_by(id: Valkyrie::ID.new(file_set_id))
+    # TODO: At least temporarily, should this return the valkyrie resource version of the fileset or the active fedora fileset?
   end
 
   def file_actor
     Hyrax::Actors::FileActor.new(file_set, relation.to_sym, user)
   end
 
+  # @return [FileNode, FalseClass] the created file node on success, false on failure
   def ingest_file
     file_actor.ingest_file(self)
   end
 
+  def to_file_node
+    Wings::FileNode.new(label: original_name,
+                        original_filename: original_name,
+                        mime_type: mime_type,
+                        use: [Valkyrie::Vocab::PCDMUse.OriginalFile])
+  end
+
+  # The magic that switches *once* between local filepath and CarrierWave file
+  # @return [File, StringIO, #read] File-like object ready to #read
+  def file
+    @file ||= (file_from_path || file_from_uploaded_file!)
+  end
+
   private
 
     def extracted_original_name
@@ -80,12 +101,6 @@ def extracted_mime_type
       uploaded_file ? uploaded_file.uploader.content_type : Hydra::PCDM::GetMimeTypeForFile.call(original_name)
     end
 
-    # The magic that switches *once* between local filepath and CarrierWave file
-    # @return [File, StringIO, #read] File-like object ready to #read
-    def file
-      @file ||= (file_from_path || file_from_uploaded_file!)
-    end
-
     # @return [File, StringIO] depending on CarrierWave configuration
     # @raise when uploaded_file *becomes* required but is missing
     def file_from_uploaded_file!

app/services/hyrax/versioning_service.rb

@@ -1,26 +1,69 @@
+require 'wings/models/file_node'
+require 'wings/services/file_node_builder'
+
 module Hyrax
   class VersioningService
-    # Make a version and record the version committer
-    # @param [ActiveFedora::File] content
-    # @param [User, String] user
-    def self.create(content, user = nil)
-      content.create_version
-      record_committer(content, user) if user
-    end
+    class << self
+      # Make a version and record the version committer
+      # @param [ActiveFedora::File | Wings::FileNode] content
+      # @param [User, String] user
+      def create(content, user = nil)
+        use_valkyrie = content.is_a? Wings::FileNode
+        perform_create(content, user, use_valkyrie)
+      end
 
-    # @param [ActiveFedora::File] file
-    def self.latest_version_of(file)
-      file.versions.last
-    end
+      # @param [ActiveFedora::File | Wings::FileNode] content
+      def latest_version_of(file)
+        file.versions.last
+      end
+
+      # Record the version committer of the last version
+      # @param [ActiveFedora::File | Wings::FileNode] content
+      # @param [User, String] user_key
+      def record_committer(content, user_key)
+        user_key = user_key.user_key if user_key.respond_to?(:user_key)
+        version = latest_version_of(content)
+        return if version.nil?
+        version_id = content.is_a?(Wings::FileNode) ? version.id.to_s : version.uri
+        Hyrax::VersionCommitter.create(version_id: version_id, committer_login: user_key)
+      end
+
+      # TODO: Copied from valkyrie6 branch.  Need to explore whether this is needed?
+      # # @param [FileSet] file_set
+      # # @param [Wings::FileNode] content
+      # # @param [String] revision_id
+      # # @param [User, String] user
+      # def restore_version(file_set, content, revision_id, user = nil)
+      #   found_version = content.versions.find { |x| x.label == Array.wrap(revision_id) }
+      #   return unless found_version
+      #   node = Wings::FileNodeBuilder.new(storage_adapter: nil, persister: indexing_adapter.persister).attach_file_node(node: found_version, file_set: file_set)
+      #   create(node, user)
+      # end
+
+      private
+
+        # # TODO: Should we create and use indexing adapter for persistence?  This is what was used in branch valkyrie6.
+        # def indexing_adapter
+        #   Valkyrie::MetadataAdapter.find(:indexing_persister)
+        # end
+
+        def perform_create(content, user, use_valkyrie)
+          use_valkyrie ? perform_create_through_valkyrie(content, user) : perform_create_through_active_fedora(content, user)
+        end
+
+        def perform_create_through_active_fedora(content, user)
+          content.create_version
+          record_committer(content, user) if user
+        end
 
-    # Record the version committer of the last version
-    # @param [ActiveFedora::File] content
-    # @param [User, String] user_key
-    def self.record_committer(content, user_key)
-      user_key = user_key.user_key if user_key.respond_to?(:user_key)
-      version = latest_version_of(content)
-      return if version.nil?
-      VersionCommitter.create(version_id: version.uri, committer_login: user_key)
+        def perform_create_through_valkyrie(content, user)
+          new_version = content.new(id: nil)
+          new_version.label = "version#{content.member_ids.length + 1}"
+          # new_version = indexing_adapter.persister.save(resource: new_version)
+          content.member_ids = content.member_ids + [new_version.id]
+          content = indexing_adapter.persister.save(resource: content)
+          record_committer(content, user) if user
+        end
     end
   end
 end

lib/wings.rb

@@ -36,9 +36,11 @@ module Wings; end
 Valkyrie::MetadataAdapter.register(
   Wings::Valkyrie::MetadataAdapter.new, :wings_adapter
 )
+Valkyrie.config.metadata_adapter = :wings_adapter
 
 Valkyrie::StorageAdapter.register(
   Valkyrie::Storage::Fedora
     .new(connection: Ldp::Client.new(ActiveFedora.fedora.host)),
   :fedora
 )
+Valkyrie.config.storage_adapter = :fedora

lib/wings/models/file_node.rb

@@ -0,0 +1,82 @@
+# frozen_string_literal: true
+
+module Wings
+  class FileNode < ::Valkyrie::Resource
+    # TODO: Branch valkyrie6 included the valkyrie resource access controls.  Including this now causes an exception.
+    #       Need to explore whether this line should be uncommented.
+    # include ::Valkyrie::Resource::AccessControls
+    attribute :id, ::Valkyrie::Types::ID.optional
+    attribute :label, ::Valkyrie::Types::Set
+    attribute :mime_type, ::Valkyrie::Types::Set
+    attribute :format_label, ::Valkyrie::Types::Set # e.g. "JPEG Image"
+    attribute :height, ::Valkyrie::Types::Set
+    attribute :width, ::Valkyrie::Types::Set
+    attribute :checksum, ::Valkyrie::Types::Set
+    attribute :size, ::Valkyrie::Types::Set
+    attribute :original_filename, ::Valkyrie::Types::Set
+    attribute :file_identifiers, ::Valkyrie::Types::Set
+    attribute :use, ::Valkyrie::Types::Set
+    attribute :member_ids, ::Valkyrie::Types::Set
+
+    # @param [ActionDispatch::Http::UploadedFile] file
+    def self.for(file:)
+      new(label: file.original_filename,
+          original_filename: file.original_filename,
+          mime_type: file.content_type,
+          use: file.try(:use) || [::Valkyrie::Vocab::PCDMUse.OriginalFile])
+    end
+
+    def original_file?
+      use.include?(::Valkyrie::Vocab::PCDMUse.OriginalFile)
+    end
+
+    def thumbnail_file?
+      use.include?(::Valkyrie::Vocab::PCDMUse.ThumbnailImage)
+    end
+
+    def extracted_file?
+      use.include?(::Valkyrie::Vocab::PCDMUse.ExtractedImage)
+    end
+
+    def title
+      label
+    end
+
+    def download_id
+      id
+    end
+
+    # @return [Boolean] whether this instance is a Wings::FileNode.
+    def file_node?
+      true
+    end
+
+    # @return [Boolean] whether this instance is a Hydra::Works FileSet.
+    def file_set?
+      false
+    end
+
+    # @return [Boolean] whether this instance is a Hydra::Works Generic Work.
+    def work?
+      false
+    end
+
+    # @return [Boolean] whether this instance is a Hydra::Works Collection.
+    def collection?
+      false
+    end
+
+    def valid?
+      file.valid?(size: size.first, digests: { sha256: checksum.first.sha256 })
+    end
+
+    def file
+      ::Valkyrie::StorageAdapter.find_by(id: file_identifiers.first)
+    end
+
+    def versions
+      query_service = Wings::Valkyrie::QueryService.new(adapter: ::Valkyrie.config.metadata_adapter)
+      query_service.find_members(resource: self, model: Wings::FileNode).to_a
+    end
+  end
+end

lib/wings/models/multi_checksum.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Wings
+  class MultiChecksum < ::Valkyrie::Resource
+    attribute :sha256, ::Valkyrie::Types::SingleValuedString
+    attribute :md5, ::Valkyrie::Types::SingleValuedString
+    attribute :sha1, ::Valkyrie::Types::SingleValuedString
+
+    def self.for(file_object)
+      digests = file_object.checksum(digests: [::Digest::MD5.new, ::Digest::SHA256.new, ::Digest::SHA1.new])
+      MultiChecksum.new(
+        md5: digests.shift,
+        sha256: digests.shift,
+        sha1: digests.shift
+      )
+    end
+  end
+end