Merge branch 'main' into rails_hyrax_upgrade

* main:
  🎁 Add Transaction for Cleaning Up Split Pages
  ♻️ Extract direct ActiveFedora calls to adapter

README.md

@@ -115,6 +115,10 @@ uv = createUV('#uv', {
 ## Configuration to enable IiifPrint features
 **NOTE: WorkTypes and models are used synonymously here.**
 
+### Persistence Layer Adapter
+
+We created IiifPrint with an assumption of ActiveFedora.  However, as Hyrax now supports Valkyrie, we need an alternate approach.  We introduced `IiifPrint::Configuration#persistence_layer` as a configuration option.  By default it will use `ActiveFedora` methods; but you can switch adapters to use Valkyrie instead.  (See `IiifPrint::PersistentLayer` for more details).
+
 ### IIIF URL configuration
 
 If you set EXTERNAL_IIIF_URL in your environment, then IiifPrint will use that URL as the root for your IIIF URLs. It will also switch from using the file set ID to using the SHA1 of the file as the identifier. This enables using serverless_iiif or Cantaloupe (refered to as the service) by pointing the service to the same S3 bucket that FCREPO writes the uploaded files to. By setting it up that way you do not need the service to connect to FCREPO or Hyrax at all, both natively support connecting to an S3 bucket to get their data.

app/actors/iiif_print/actors/file_set_actor_decorator.rb

@@ -48,7 +48,7 @@ def destroy
         # we destroy the children before the file_set, because we need the parent relationship
         IiifPrint::SplitPdfs::DestroyPdfChildWorksService.conditionally_destroy_spawned_children_of(
           file_set: file_set,
-          work: file_set.parent
+          work: IiifPrint.parent_for(file_set)
         )
         # and now back to your regularly scheduled programming
         super

app/models/concerns/iiif_print/solr/document.rb

@@ -39,7 +39,7 @@ def digest_sha1
 
   def method_missing(method_name, *args, &block)
     super unless iiif_print_solr_field_names.include? method_name.to_s
-    self[::ActiveFedora.index_field_mapper.solr_name(method_name.to_s)]
+    self[IiifPrint.solr_name(method_name.to_s)]
   end
 
   def respond_to_missing?(method_name, include_private = false)

app/services/iiif_print/manifest_builder_service_behavior.rb

@@ -142,7 +142,7 @@ def get_solr_hits(ids)
       results = []
       ids.each_slice(SOLR_QUERY_PAGE_SIZE) do |paged_ids|
         query = "id:(#{paged_ids.join(' OR ')})"
-        results += ActiveFedora::SolrService.query(
+        results += IiifPrint.solr_query(
           query,
           { fq: "-has_model_ssim:FileSet", rows: paged_ids.size, method: :post }
         )

app/transactions/hyrax/transactions/iiif_print_container_decorator.rb

@@ -0,0 +1,32 @@
+module Hyrax
+  module Transactions
+    ##
+    # This decorator does the following:
+    #
+    # - Prepend the {ConditionallyDestroyChildrenFromSplit} transaction to the "file_set.destroy"
+    #   step.  The prependment corresponds to the behavior for
+    #   {IiifPrint::Actors::FileSetActorDecorator#destroy}
+    #
+    # For more information about adjusting transactions, see
+    # [Transitioning workshop solution for adding transaction](https://github.com/samvera-labs/transitioning-to-valkyrie-workshop/commit/bcab2bb8f65078e88395c68f72be00e7ffad57ec)
+    #
+    # @see https://github.com/samvera/hyrax/blob/f875d61dc87229cf1f05eb2bb6d414b5ef314616/lib/hyrax/transactions/container.rb
+    class IiifPrintContainerDecorator
+      extend Dry::Container::Mixin
+
+      namespace 'file_set' do |ops|
+        ops.register 'iiif_print_conditionally_destroy_spawned_children' do
+          Steps::ConditionallyDestroyChildrenFromSplit.new
+        end
+        ops.register 'destroy' do
+          Hyrax::Transactions::FileSetDestroy.new(
+            steps: (['file_set.iiif_print_conditionally_destroy_spawned_children'] +
+              Hyrax::Transactions::FileSetDestroy::DEFAULT_STEPS)
+          )
+        end
+      end
+    end
+  end
+end
+
+Hyrax::Transactions::Container.merge(Hyrax::Transactions::IiifPrintContainerDecorator)

app/transactions/hyrax/transactions/steps/conditionally_destroy_children_from_split.rb

@@ -0,0 +1,31 @@
+module Hyrax
+  module Transactions
+    module Steps
+      ##
+      # For a FileSet that is a PDF, we need to delete any works and file_sets that are the result of
+      # splitting that PDF into constituent images of each page of the PDF.  This is responsible for
+      # that work.
+      class ConditionallyDestroyChildrenFromSplit
+        include Dry::Monads[:result]
+
+        ##
+        # @param resource [Hyrax::FileSet]
+        def call(resource)
+          return Failure(:resource_not_persisted) unless resource.persisted?
+
+          parent = IiifPrint.persistence_adapter.parent_for(resource)
+          return Success(true) unless parent
+
+          # We do not care about the results of this call; as it is conditionally looking for things
+          # to destroy.
+          IiifPrint::SplitPdfs::DestroyPdfChildWorksService.conditionally_destroy_spawned_children_of(
+            file_set: resource,
+            work: parent
+          )
+
+          Success(true)
+        end
+      end
+    end
+  end
+end

lib/iiif_print.rb

@@ -44,37 +44,22 @@ def self.config(&block)
   end
 
   class << self
-    delegate :skip_splitting_pdf_files_that_end_with_these_texts, to: :config
-  end
-
-  ##
-  # Return the immediate parent of the given :file_set.
-  #
-  # @param file_set [FileSet]
-  # @return [#work?, Hydra::PCDM::Work]
-  # @return [NilClass] when no parent is found.
-  def self.parent_for(file_set)
-    # fallback to Fedora-stored relationships if work's aggregation of
-    #   file set is not indexed in Solr
-    file_set.parent || file_set.member_of.find(&:work?)
-  end
+    delegate(
+      :persistence_adapter,
+      :skip_splitting_pdf_files_that_end_with_these_texts,
+      to: :config
+    )
 
-  ##
-  # Return the parent's parent of the given :file_set.
-  #
-  # @param file_set [FileSet]
-  # @return [#work?, Hydra::PCDM::Work]
-  # @return [NilClass] when no grand parent is found.
-  def self.grandparent_for(file_set)
-    parent_of_file_set = parent_for(file_set)
-    # HACK: This is an assumption about the file_set structure, namely that an image page split from
-    # a PDF is part of a file set that is a child of a work that is a child of a single work.  That
-    # is, it only has one grand parent.  Which is a reasonable assumption for IIIF Print but is not
-    # valid when extended beyond IIIF Print.  That is GenericWork does not have a parent method but
-    # does have a parents method.
-    parent_of_file_set.try(:parent_works).try(:first) ||
-      parent_of_file_set.try(:parents).try(:first) ||
-      parent_of_file_set&.member_of&.find(&:work?)
+    delegate(
+      :clean_for_tests!,
+      :destroy_children_split_from,
+      :grandparent_for,
+      :solr_construct_query,
+      :solr_name,
+      :solr_query,
+      :parent_for,
+      to: :persistence_adapter
+    )
   end
 
   def self.use_valkyrie?(obj)

lib/iiif_print/catalog_search_builder.rb

@@ -25,9 +25,9 @@ class CatalogSearchBuilder < Hyrax::CatalogSearchBuilder
     # rubocop:enable Naming/PredicateName
     def show_parents_only(solr_parameters)
       query = if blacklight_params["include_child_works"] == 'true'
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: 'true')
+                IiifPrint.solr_construct_query(is_child_bsi: 'true')
               else
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: nil)
+                IiifPrint.solr_construct_query(is_child_bsi: nil)
               end
       solr_parameters[:fq] += [query]
     end

lib/iiif_print/configuration.rb

@@ -3,6 +3,22 @@ module IiifPrint
   class Configuration
     attr_writer :after_create_fileset_handler
 
+    attr_writer :persistence_adapter
+    def persistence_adapter
+      @persistent_adapter || default_persistence_adapter
+    end
+
+    def default_persistence_adapter
+      # There's probably some configuration of Hyrax we could use to better refine this; but it's
+      # likely a reasonable guess.  The main goal is to not break existing implementations and
+      # maintain an upgrade path.
+      if Gem::Version.new(Hyrax::VERSION) >= Gem::Version.new('6.0.0')
+        IiifPrint::PersistenceLayer::ValkyrieAdapter
+      else
+        IiifPrint::PersistenceLayer::ActiveFedoraAdapter
+      end
+    end
+
     # @param file_set [FileSet]
     # @param user [User]
     def handle_after_create_fileset(file_set, user)

lib/iiif_print/data/work_derivatives.rb

@@ -239,7 +239,7 @@ def primary_file_path
           #   of the first assigned file path for single-file work.
           work_file = parent
           return if work_file.nil?
-          work_files = work_file.parent
+          work_files = IiifPrint.parent_for(work_file)
           return if work_files.nil?
           work_files.assigned[0]
         else

lib/iiif_print/engine.rb

@@ -11,6 +11,15 @@ module IiifPrint
   class Engine < ::Rails::Engine
     isolate_namespace IiifPrint
 
+    config.eager_load_paths += %W[#{config.root}/app/transactions]
+
+    initializer 'requires' do
+      require 'hyrax/transactions/iiif_print_container_decorator'
+      require 'iiif_print/persistence_layer'
+      require 'iiif_print/persistence_layer/active_fedora_adapter' if defined?(ActiveFedora)
+      require 'iiif_print/persistence_layer/valkyrie_adapter' if defined?(Valkyrie)
+    end
+
     # rubocop:disable Metrics/BlockLength
     config.to_prepare do
       require "iiif_print/jobs/create_relationships_job"

lib/iiif_print/homepage_search_builder.rb

@@ -7,9 +7,9 @@ class HomepageSearchBuilder < Hyrax::HomepageSearchBuilder
 
     def show_parents_only(solr_parameters)
       query = if blacklight_params["include_child_works"] == 'true'
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: 'true')
+                IiifPrint.solr_construct_query(is_child_bsi: 'true')
               else
-                ActiveFedora::SolrQueryBuilder.construct_query(is_child_bsi: nil)
+                IiifPrint.solr_construct_query(is_child_bsi: nil)
               end
       solr_parameters[:fq] += [query]
     end

lib/iiif_print/persistence_layer.rb

@@ -0,0 +1,58 @@
+module IiifPrint
+  ##
+  # The PersistenceLayer module provides the namespace for other adapters:
+  #
+  # - {IiifPrint::PersistenceLayer::ActiveFedoraAdapter}
+  # - {IiifPrint::PersistenceLayer::ValkyrieAdapter}
+  #
+  # And the defining interface in the {IiifPrint::PersistenceLayer::AbstractAdapter}
+  module PersistenceLayer
+    # @abstract
+    class AbstractAdapter
+      ##
+      # @param file_set [Object]
+      # @param work [Object]
+      # @param model [Class] The class name for which we'll split children.
+      def self.destroy_children_split_from(file_set:, work:, model:)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+
+      ##
+      # @abstract
+      def self.parent_for(*)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+
+      ##
+      # @abstract
+      def self.grandparent_for(*)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+
+      ##
+      # @abstract
+      def self.solr_field_query(*)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+
+      ##
+      # @abstract
+      def self.clean_for_tests!
+        return false unless Rails.env.test?
+        yield
+      end
+
+      ##
+      # @abstract
+      def self.solr_query(*args)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+
+      ##
+      # @abstract
+      def self.solr_name(*args)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
+    end
+  end
+end

lib/iiif_print/persistence_layer/active_fedora_adapter.rb

@@ -0,0 +1,83 @@
+module IiifPrint
+  module PersistenceLayer
+    class ActiveFedoraAdapter < AbstractAdapter
+      ##
+      # Return the immediate parent of the given :file_set.
+      #
+      # @param file_set [FileSet]
+      # @return [#work?, Hydra::PCDM::Work]
+      # @return [NilClass] when no parent is found.
+      def self.parent_for(file_set)
+        # fallback to Fedora-stored relationships if work's aggregation of
+        #   file set is not indexed in Solr
+        file_set.parent || file_set.member_of.find(&:work?)
+      end
+
+      ##
+      # Return the parent's parent of the given :file_set.
+      #
+      # @param file_set [FileSet]
+      # @return [#work?, Hydra::PCDM::Work]
+      # @return [NilClass] when no grand parent is found.
+      def self.grandparent_for(file_set)
+        parent_of_file_set = parent_for(file_set)
+        # HACK: This is an assumption about the file_set structure, namely that an image page split from
+        # a PDF is part of a file set that is a child of a work that is a child of a single work.  That
+        # is, it only has one grand parent.  Which is a reasonable assumption for IIIF Print but is not
+        # valid when extended beyond IIIF Print.  That is GenericWork does not have a parent method but
+        # does have a parents method.
+        parent_of_file_set.try(:parent_works).try(:first) ||
+          parent_of_file_set.try(:parents).try(:first) ||
+          parent_of_file_set&.member_of&.find(&:work?)
+      end
+
+      def self.solr_construct_query(*args)
+        if defined?(Hyrax::SolrQueryBuilderService)
+          Hyrax::SolrQueryBuilderService.construct_query(*args)
+        else
+          ActiveFedora::SolrQueryBuilderService.construct_query(*args)
+        end
+      end
+
+      def self.clean_for_tests!
+        super do
+          ActiveFedora::Cleaner.clean!
+        end
+      end
+
+      def self.solr_query(*args)
+        if defined?(Hyrax::SolrService)
+          Hyrax::SolrService.query(*args)
+        else
+          ActiveFedora::SolrService.query(*args)
+        end
+      end
+
+      def self.solr_name(field_name)
+        if defined?(Hyrax) && Hyrax.config.respond_to?(:index_field_mapper)
+          Hyrax.config.index_field_mapper.solr_name(field_name.to_s)
+        else
+          ::ActiveFedora.index_field_mapper.solr_name(field_name.to_s)
+        end
+      end
+
+      ##
+      # @param file_set [Object]
+      # @param work [Object]
+      # @param model [Class] The class name for which we'll split children.
+      def self.destroy_children_split_from(file_set:, work:, model:)
+        # look first for children by the file set id they were split from
+        children = model.where(split_from_pdf_id: file_set.id)
+        if children.blank?
+          # find works where file name and work `to_param` are both in the title
+          children = model.where(title: file_set.label).where(title: work.to_param)
+        end
+        return if children.blank?
+        children.each do |rcd|
+          rcd.destroy(eradicate: true)
+        end
+        true
+      end
+    end
+  end
+end