Merge pull request #4215 from samvera/valk-char

Create `FileSetDescription` to find characterization for FileSets
samvera · Jan 30, 2020 · 1d5c2f2 · 1d5c2f2
2 parents 226e010 + a749f23
commit 1d5c2f2
Show file tree

Hide file tree

Showing 36 changed files with 233 additions and 118 deletions.
diff --git a/app/actors/hyrax/actors/file_actor.rb b/app/actors/hyrax/actors/file_actor.rb
@@ -100,29 +100,28 @@ def normalize_relation(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 FileSet.*_use... making the casecmp unnecessary
-          return :original_file if relation.to_s.casecmp(Hyrax::FileSet::ORIGINAL_FILE_USE.to_s)
-          return :extracted_file if relation.to_s.casecmp(Hyrax::FileSet::EXTRACTED_TEXT_USE.to_s)
-          return :thumbnail_file if relation.to_s.casecmp(Hyrax::FileSet::THUMBNAIL_USE.to_s)
-          :original_file
+          case relation
+          when Hyrax::FileMetadata::Use::ORIGINAL_FILE
+            :original_file
+          when Hyrax::FileMetadata::Use::EXTRACTED_TEXT
+            :extracted_file
+          when Hyrax::FileMetadata::Use::THUMBNAIL
+            :thumbnail_file
+          else
+            :original_file
+          end
         end
 
+        ##
+        # @return [RDF::URI]
         def normalize_relation_for_valkyrie(relation)
-          # TODO: When this is fully switched to valkyrie, this should probably be removed and relation should always be passed
-          #       in as a valid URI already set to the file's use
-          case relation.to_s.to_sym
-          when :original_file
-            Hyrax::FileSet::ORIGINAL_FILE_USE
-          when :extracted_file
-            Hyrax::FileSet.EXTRACTED_TEXT_USE
-          when :thumbnail_file
-            Hyrax::FileSet::THUMBNAIL_USE
-          else
-            Hyrax::FileSet::ORIGINAL_FILE_USE
-          end
+          return relation if relation.is_a?(RDF::URI)
+
+          Hyrax::FileMetadata::Use.uri_for(use: relation.to_sym)
+        rescue ArgumentError
+          Hyrax::FileMetadata::Use::ORIGINAL_FILE
         end
     end
   end

diff --git a/app/controllers/concerns/hyrax/embargoes_controller_behavior.rb b/app/controllers/concerns/hyrax/embargoes_controller_behavior.rb
@@ -27,7 +27,7 @@ def update
       filter_docs_with_edit_access!
       copy_visibility = []
       copy_visibility = params[:embargoes].values.map { |h| h[:copy_visibility] } if params[:embargoes]
-      af_objects = Hyrax.query_service.custom_queries.find_many_by_alternate_ids(alternate_ids: batch, use_valkyrie: false)
+      af_objects = Hyrax.custom_queries.find_many_by_alternate_ids(alternate_ids: batch, use_valkyrie: false)
       af_objects.each do |curation_concern|
         Hyrax::Actors::EmbargoActor.new(curation_concern).destroy
         # if the concern is a FileSet, set its visibility and visibility propagation

diff --git a/app/controllers/concerns/hyrax/leases_controller_behavior.rb b/app/controllers/concerns/hyrax/leases_controller_behavior.rb
@@ -26,7 +26,7 @@ def update
       filter_docs_with_edit_access!
       copy_visibility = []
       copy_visibility = params[:leases].values.map { |h| h[:copy_visibility] } if params[:leases]
-      af_objects = Hyrax.query_service.custom_queries.find_many_by_alternate_ids(alternate_ids: batch, use_valkyrie: false)
+      af_objects = Hyrax.custom_queries.find_many_by_alternate_ids(alternate_ids: batch, use_valkyrie: false)
       af_objects.each do |curation_concern|
         Hyrax::Actors::LeaseActor.new(curation_concern).destroy
         Hyrax::VisibilityPropagator.for(source: curation_concern).propagate if

diff --git a/app/jobs/concerns/hyrax/members_permission_job_behavior.rb b/app/jobs/concerns/hyrax/members_permission_job_behavior.rb
@@ -16,7 +16,7 @@ def file_set_ids(work)
         when ActiveFedora::Base
           ::FileSet.search_with_conditions(id: work.member_ids).map(&:id)
         when Valkyrie::Resource
-          Hyrax.query_service.custom_queries.find_child_fileset_ids(resource: work)
+          Hyrax.custom_queries.find_child_fileset_ids(resource: work)
         end
       end
 

diff --git a/app/models/concerns/hyrax/collection_behavior.rb b/app/models/concerns/hyrax/collection_behavior.rb
@@ -55,7 +55,7 @@ def collection_type=(new_collection_type)
     #       add_member_objects using the member_of_collections relationship.  Deprecate?
     def add_members(new_member_ids)
       return if new_member_ids.blank?
-      members << Hyrax.query_service.custom_queries.find_many_by_alternate_ids(alternate_ids: new_member_ids, use_valkyrie: false)
+      members << Hyrax.custom_queries.find_many_by_alternate_ids(alternate_ids: new_member_ids, use_valkyrie: false)
     end
 
     # Add member objects by adding this collection to the objects' member_of_collection association.

diff --git a/app/models/hyrax/file_metadata.rb b/app/models/hyrax/file_metadata.rb
@@ -2,15 +2,45 @@
 
 module Hyrax
   class FileMetadata < Valkyrie::Resource
+    GENERIC_MIME_TYPE = 'application/octet-stream'
+
+    ##
+    # Constants for PCDM Use URIs; use these constants in place of hard-coded
+    # URIs in the `::Valkyrie::Vocab::PCDMUse` vocabulary.
+    module Use
+      ORIGINAL_FILE = ::Valkyrie::Vocab::PCDMUse.OriginalFile
+      EXTRACTED_TEXT = ::Valkyrie::Vocab::PCDMUse.ExtractedText
+      THUMBNAIL = ::Valkyrie::Vocab::PCDMUse.ThumbnailImage
+
+      ##
+      # @param use [Symbol]
+      #
+      # @return [RDF::URI]
+      # @raise [ArgumentError] if no use is known for the argument
+      def uri_for(use:)
+        case use
+        when :original_file
+          ORIGINAL_FILE
+        when :extracted_file
+          EXTRACTED_TEXT
+        when :thumbnail_file
+          THUMBNAIL
+        else
+          raise ArgumentError, "No PCDM use is recognized for #{use}"
+        end
+      end
+      module_function :uri_for
+    end
+
     attribute :file_identifiers, ::Valkyrie::Types::Set # id of the file stored by the storage adapter
     attribute :alternate_ids, Valkyrie::Types::Set.of(Valkyrie::Types::ID) # id of the Hydra::PCDM::File which holds metadata and the file in ActiveFedora
     attribute :file_set_id, ::Valkyrie::Types::ID # id of parent file set resource
 
     # all remaining attributes are on AF::File metadata_node unless otherwise noted
     attribute :label, ::Valkyrie::Types::Set
     attribute :original_filename, ::Valkyrie::Types::Set
-    attribute :mime_type, ::Valkyrie::Types::Set
-    attribute :type, ::Valkyrie::Types::Set # AF::File type
+    attribute :mime_type, ::Valkyrie::Types::String.default(GENERIC_MIME_TYPE)
+    attribute :type, ::Valkyrie::Types::Set.default([Use::ORIGINAL_FILE])
     attribute :content, ::Valkyrie::Types::Set
 
     # attributes set by fits
@@ -75,20 +105,25 @@ class FileMetadata < Valkyrie::Resource
     def self.for(file:)
       new(label: file.original_filename,
           original_filename: file.original_filename,
-          mime_type: file.content_type,
-          type: file.try(:type) || [Hyrax::FileSet::ORIGINAL_FILE_USE])
+          mime_type: file.content_type)
     end
 
+    ##
+    # @return [Boolean]
     def original_file?
-      type.include?(Hyrax::FileSet::ORIGINAL_FILE_USE)
+      type.include?(Use::ORIGINAL_FILE)
     end
 
+    ##
+    # @return [Boolean]
     def thumbnail_file?
-      type.include?(Hyrax::FileSet::THUMBNAIL_USE)
+      type.include?(Use::THUMBNAIL)
     end
 
+    ##
+    # @return [Boolean]
     def extracted_file?
-      type.include?(Hyrax::FileSet::EXTRACTED_TEXT_USE)
+      type.include?(Use::EXTRACTED_TEXT)
     end
 
     def title

diff --git a/app/models/hyrax/file_set.rb b/app/models/hyrax/file_set.rb
@@ -8,10 +8,6 @@ module Hyrax
   class FileSet < Hyrax::Resource
     include Hyrax::Schema(:core_metadata)
 
-    ORIGINAL_FILE_USE = ::Valkyrie::Vocab::PCDMUse.OriginalFile
-    EXTRACTED_TEXT_USE = ::Valkyrie::Vocab::PCDMUse.ExtractedText
-    THUMBNAIL_USE = ::Valkyrie::Vocab::PCDMUse.Thumbnail
-
     attribute :file_ids, Valkyrie::Types::Array.of(Valkyrie::Types::ID) # id for FileMetadata resources
     attribute :original_file_id, Valkyrie::Types::ID # id for FileMetadata resource
     attribute :thumbnail_id, Valkyrie::Types::ID # id for FileMetadata resource
@@ -28,36 +24,5 @@ def pcdm_object?
     def file_set?
       true
     end
-
-    ##
-    # Gives file metadata for the file filling the http://pcdm.org/OriginalFile use
-    # @return [FileMetadata] the FileMetadata resource of the original file
-    def original_file
-      filter_files_by_type(Hyrax::FileSet::ORIGINAL_FILE_USE).first
-    end
-
-    ##
-    # Gives file metadata for the file filling the http://pcdm.org/ExtractedText use
-    # @return [FileMetadata] the FileMetadata resource of the extracted text
-    def extracted_text
-      filter_files_by_type(Hyrax::FileSet::EXTRACTED_TEXT_USE).first
-    end
-
-    ##
-    # Gives file metadata for the file filling the http://pcdm.org/Thumbnail use
-    # @return [FileMetadata] the FileMetadata resource of the thumbnail
-    def thumbnail
-      filter_files_by_type(Hyrax::FileSet::THUMBNAIL_USE).first
-    end
-
-    ##
-    # Gives file metadata for files that have the requested RDF Type for use
-    # @param [RDF::URI] uri for the desired Type
-    # @return [Enumerable<FileMetadata>] the FileMetadata resources
-    # @example
-    #   filter_files_by_type(::RDF::URI("http://pcdm.org/ExtractedText"))
-    def filter_files_by_type(uri)
-      Hyrax.query_service.custom_queries.find_many_file_metadata_by_use(resource: self, use: uri)
-    end
   end
 end
diff --git a/app/models/job_io_wrapper.rb b/app/models/job_io_wrapper.rb
@@ -80,7 +80,7 @@ def to_file_metadata
     Hyrax::FileMetadata.new(label: original_name,
                             original_filename: original_name,
                             mime_type: mime_type,
-                            use: [Hyrax::FileSet::ORIGINAL_FILE_USE])
+                            use: [Hyrax::FileMetadata::Use::ORIGINAL_FILE])
   end
 
   # The magic that switches *once* between local filepath and CarrierWave file

diff --git a/app/services/hyrax/characterization/file_set_description.rb b/app/services/hyrax/characterization/file_set_description.rb
@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+module Hyrax
+  module Characterization
+    ##
+    # @api public
+    class FileSetDescription
+      include Hydra::Works::MimeTypes
+
+      ##
+      # @!attribute [rw] file_set
+      #   @return [Hyrax::FileSet]
+      attr_accessor :file_set
+
+      delegate :mime_type, to: :primary_file
+
+      ##
+      # @param [Hyrax::FileSet] file_set
+      # @param [Symbol] primary_file  a symbol mapping to the file_set member
+      #   used for characterization
+      def initialize(file_set:, primary_file: :original_file)
+        self.file_set = file_set
+
+        @primary_file_type_uri =
+          Hyrax::FileMetadata::Use.uri_for(use: primary_file)
+      end
+
+      ##
+      # @api public
+      # @return [Hyrax::FileMetadata] the member file to use for characterization
+      def primary_file
+        queries.find_many_file_metadata_by_use(resource: file_set, use: @primary_file_type_uri).first ||
+          Hyrax::FileMetadata.new
+      end
+
+      private
+
+        ##
+        # @api private
+        def queries
+          Hyrax.custom_queries
+        end
+    end
+  end
+end
diff --git a/app/services/hyrax/characterization/model_wrapper.rb b/app/services/hyrax/characterization/model_wrapper.rb
diff --git a/app/services/hyrax/custom_queries/find_access_control.rb b/app/services/hyrax/custom_queries/find_access_control.rb
@@ -1,7 +1,7 @@
 module Hyrax
   module CustomQueries
     # @example
-    #   Hyrax.query_service.custom_queries.find_access_control_for(resource: resource)
+    #   Hyrax.custom_queries.find_access_control_for(resource: resource)
     class FindAccessControl
       def self.queries
         [:find_access_control_for]

diff --git a/app/services/hyrax/custom_queries/find_file_metadata.rb b/app/services/hyrax/custom_queries/find_file_metadata.rb
@@ -1,15 +1,16 @@
 # Provide custom queries for finding Hyrax::FileMetadata
 # @example
-#   Hyrax.query_service.custom_queries.find_file_metadata_by(id: valkyrie_id)
-#   Hyrax.query_service.custom_queries.find_file_metadata_by_alternate_identifier(alternate_identifier: alt_id)
-#   Hyrax.query_service.custom_queries.find_many_file_metadata_by_ids(ids: [valkyrie_id, valkyrie_id])
+#   Hyrax.custom_queries.find_file_metadata_by(id: valkyrie_id)
+#   Hyrax.custom_queries.find_file_metadata_by_alternate_identifier(alternate_identifier: alt_id)
+#   Hyrax.custom_queries.find_many_file_metadata_by_ids(ids: [valkyrie_id, valkyrie_id])
 module Hyrax
   module CustomQueries
     class FindFileMetadata
       def self.queries
         [:find_file_metadata_by,
          :find_file_metadata_by_alternate_identifier,
-         :find_many_file_metadata_by_ids]
+         :find_many_file_metadata_by_ids,
+         :find_many_file_metadata_by_use]
       end
 
       def initialize(query_service:)

diff --git a/app/services/hyrax/custom_queries/find_many_by_alternate_ids.rb b/app/services/hyrax/custom_queries/find_many_by_alternate_ids.rb
@@ -2,7 +2,7 @@ module Hyrax
   module CustomQueries
     class FindManyByAlternateIds
       # Use:
-      #   Hyrax.query_service.custom_queries.find_many_by_alternate_ids(alternate_ids: ids)
+      #   Hyrax.custom_queries.find_many_by_alternate_ids(alternate_ids: ids)
 
       def self.queries
         [:find_many_by_alternate_ids]

diff --git a/app/services/hyrax/custom_queries/navigators/find_files.rb b/app/services/hyrax/custom_queries/navigators/find_files.rb
@@ -3,10 +3,10 @@ module CustomQueries
     module Navigators
       class FindFiles
         # @example
-        #   Hyrax.query_service.custom_queries.find_files(file_set: file_set_resource)
-        #   Hyrax.query_service.custom_queries.find_original_file(file_set: file_set_resource)
-        #   Hyrax.query_service.custom_queries.find_extracted_text(file_set: file_set_resource)
-        #   Hyrax.query_service.custom_queries.find_thumbnail(file_set: file_set_resource)
+        #   Hyrax.custom_queries.find_files(file_set: file_set_resource)
+        #   Hyrax.custom_queries.find_original_file(file_set: file_set_resource)
+        #   Hyrax.custom_queries.find_extracted_text(file_set: file_set_resource)
+        #   Hyrax.custom_queries.find_thumbnail(file_set: file_set_resource)
 
         def self.queries
           [:find_files,

diff --git a/app/services/hyrax/resource_visibility_propagator.rb b/app/services/hyrax/resource_visibility_propagator.rb
@@ -26,7 +26,7 @@ def initialize(source:,
                    embargo_manager: Hyrax::EmbargoManager,
                    lease_manager:   Hyrax::LeaseManager,
                    persister:       Hyrax.persister,
-                   queries:         Hyrax.query_service.custom_queries)
+                   queries:         Hyrax.custom_queries)
       @persister       = persister
       @queries         = queries
       self.source      = source

diff --git a/lib/hyrax.rb b/lib/hyrax.rb
@@ -104,4 +104,10 @@ def self.storage_adapter
   def self.query_service
     metadata_adapter.query_service
   end
+
+  ##
+  # The custom queries common to Hyrax
+  def self.custom_queries
+    query_service.custom_queries
+  end
 end
diff --git a/lib/wings/hydra/pcdm/models/concerns/pcdm_valkyrie_behavior.rb b/lib/wings/hydra/pcdm/models/concerns/pcdm_valkyrie_behavior.rb
@@ -44,7 +44,7 @@ def parent_collection_ids(valkyrie: false)
       # @return [Enumerable<ActiveFedora::Base> | Enumerable<Valkyrie::Resource>] an enumerable over the child collections
       # @todo There is no guarantee to collection ordering until Hyrax is fully valkyrie-native, see issue 3784
       def child_collections(valkyrie: false)
-        resources = Hyrax.query_service.custom_queries.find_child_collections(resource: self)
+        resources = Hyrax.custom_queries.find_child_collections(resource: self)
         return resources if valkyrie
         resources.map { |r| Wings::ActiveFedoraConverter.new(resource: r).convert }
       end

diff --git a/lib/wings/hydra/works/services/add_file_to_file_set.rb b/lib/wings/hydra/works/services/add_file_to_file_set.rb
@@ -37,9 +37,14 @@ def association_type(type)
         end
 
         def type_to_association_type(type)
-          return :original_file if type.to_s.casecmp?(Hyrax::FileSet::ORIGINAL_FILE_USE.to_s)
-          return :extracted_text if type.to_s.casecmp?(Hyrax::FileSet::EXTRACTED_TEXT_USE.to_s)
-          return :thumbnail if type.to_s.casecmp?(Hyrax::FileSet::THUMBNAIL_USE.to_s)
+          case type
+          when Hyrax::FileMetadata::Use::ORIGINAL_FILE
+            :original_file
+          when Hyrax::FileMetadata::Use::EXTRACTED_TEXT
+            :extracted_text
+          when Hyrax::FileMetadata::Use::THUMBNAIL
+            :thumbnail
+          end
         end
 
         def type_to_rdf_uri(type)

diff --git a/lib/wings/services/custom_queries/find_access_control.rb b/lib/wings/services/custom_queries/find_access_control.rb
@@ -3,7 +3,7 @@ module CustomQueries
     class FindAccessControl
       # Custom query override specific to Wings
       # Use:
-      #   Hyrax.query_service.custom_queries.find_access_control_for(resource: resource)
+      #   Hyrax.custom_queries.find_access_control_for(resource: resource)
 
       def self.queries
         [:find_access_control_for]

diff --git a/lib/wings/services/custom_queries/find_file_metadata.rb b/lib/wings/services/custom_queries/find_file_metadata.rb
@@ -1,7 +1,7 @@
 # Custom query override specific to Wings for finding Hydra::PCDM::File and converting to Hyrax::FileMetadata.
 # @example
-#   Hyrax.query_service.custom_queries.find_file_metadata_by(id: valkyrie_id, use_valkyrie: true)
-#   Hyrax.query_service.custom_queries.find_file_metadata_by_alternate_identifier(alternate_identifier: id, use_valkyrie: true)
+#   Hyrax.custom_queries.find_file_metadata_by(id: valkyrie_id, use_valkyrie: true)
+#   Hyrax.custom_queries.find_file_metadata_by_alternate_identifier(alternate_identifier: id, use_valkyrie: true)
 require 'wings/services/file_converter_service'
 module Wings
   module CustomQueries

diff --git a/lib/wings/services/custom_queries/find_many_by_alternate_ids.rb b/lib/wings/services/custom_queries/find_many_by_alternate_ids.rb
@@ -3,7 +3,7 @@ module CustomQueries
     class FindManyByAlternateIds
       # Custom query override specific to Wings
       # Use:
-      #   Hyrax.query_service.custom_queries.find_many_by_alternate_ids(alternate_ids: ids, use_valkyrie: true)
+      #   Hyrax.custom_queries.find_many_by_alternate_ids(alternate_ids: ids, use_valkyrie: true)
 
       def self.queries
         [:find_many_by_alternate_ids]