iiif_print 1.0.0 → 2.0.0

data/app/jobs/iiif_print/jobs/create_relationships_job.rb

@@ -0,0 +1,117 @@
+# rubocop:disable Metrics/ClassLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+module IiifPrint
+  module Jobs
+    # Link newly created child works to the parent
+    class CreateRelationshipsJob < IiifPrint::Jobs::ApplicationJob
+      include Hyrax::Lockable
+
+      RETRY_MAX = 10
+
+      # @param parent_id: [<String>] parent work id
+      # @param parent_model: [<String>] parent model
+      # @param child_model: [<String>] child model
+      # @param retries: [<Integer>] count used during rescheduling to prevent infinite retries
+      def perform(parent_id:, parent_model:, child_model:, retries: 0, **)
+        @parent_id = parent_id
+        @parent_model = parent_model
+        @child_model = child_model
+        @retries = retries + 1
+
+        @number_of_successes = 0
+        @number_of_failures = 0
+        @parent_record_members_added = false
+        @errors = []
+
+        # Because we need our children in the correct order, we can't create any
+        # relationships until all child works have been created.
+        if completed_child_data
+          # add the members
+          add_children_to_parent
+          if @number_of_failures.zero? && @number_of_successes == @pending_children.count
+            # remove pending relationships upon valid completion
+            @pending_children.each(&:destroy)
+          elsif @number_of_failures.zero? && @number_of_successes > @pending_children.count
+            # remove pending relationships but raise error that too many relationships formed
+            @pending_children.each(&:destroy)
+            raise "CreateRelationshipsJob for parent id: #{@parent_id} " \
+                  "added #{@number_of_successes} children, " \
+                  "expected #{@pending_children.count} children."
+          else
+            # report failures & keep pending relationships
+            raise "CreateRelationshipsJob failed for parent id: #{@parent_id} " \
+                  "had #{@number_of_successes} successes & #{@number_of_failures} failures, " \
+                  "with errors: #{@errors}. Wanted #{@pending_children.count} children."
+          end
+        else
+          # if we aren't ready yet, reschedule the job and end this one normally
+          reschedule_job
+        end
+      end
+
+      private
+
+      # load @child_works and @pending children, and
+      # return boolean indicating whether all chilren are present
+      def completed_child_data
+        @child_works = []
+        found_all_children = true
+
+        # find and sequence all pending children
+        @pending_children = IiifPrint::PendingRelationship.where(parent_id: @parent_id).order('child_order asc')
+
+        # find child works (skip out if any haven't yet been created)
+        @pending_children.each do |child|
+          # find by title... if any aren't found, the child works are not yet ready
+          found_children = IiifPrint.find_by_title_for(title: child.child_title, model: @child_model)
+          found_all_children = false if found_children.empty?
+          break unless found_all_children == true
+          @child_works += found_children
+        end
+        # return boolean
+        found_all_children
+      end
+
+      def add_children_to_parent
+        parent_work = IiifPrint.find_by(id: @parent_id)
+        create_relationships(parent: parent_work, ordered_children: @child_works)
+      end
+
+      def reschedule_job
+        return if @retries > RETRY_MAX
+        CreateRelationshipsJob.set(wait: 10.minutes).perform_later(
+          parent_id: @parent_id,
+          parent_model: @parent_model,
+          child_model: @child_model,
+          retries: @retries
+        )
+      end
+
+      def create_relationships(parent:, ordered_children:)
+        acquire_lock_for(parent.id) do
+          # Not sure uncached is needed here, but kept
+          # for consistency with Bulkrax's relationships logic
+          ActiveRecord::Base.uncached do
+            ordered_children.each do |child|
+              add_to_work(child_record: child, parent_record: parent)
+              @number_of_successes += 1
+            rescue => e
+              @number_of_failures += 1
+              @errors << e
+            end
+          end
+
+          IiifPrint.save(object: parent) if @parent_record_members_added && @number_of_failures.zero?
+        end
+
+        # Bulkrax no longer reindexes file_sets, but IiifPrint needs both to add is_page_of_ssim for universal viewer.
+        # This is where child works need to be indexed (AFTER the parent save), as opposed to how Bulkrax does it.
+        IiifPrint.index_works(objects: ordered_children)
+      end
+
+      def add_to_work(child_record:, parent_record:)
+        @parent_record_members_added = IiifPrint.create_relationship_between(child_record: child_record, parent_record: parent_record)
+      end
+    end
+  end
+  # rubocop:enable Metrics/ClassLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+end

data/app/jobs/iiif_print/jobs/request_split_pdf_job.rb

@@ -0,0 +1,31 @@
+module IiifPrint
+  module Jobs
+    ##
+    # Encapsulates logic for cleanup when the PDF is destroyed after pdf splitting into child works
+    class RequestSplitPdfJob < IiifPrint::Jobs::ApplicationJob
+      ##
+      # @param file_set [FileSet]
+      # @param user [User]
+      # rubocop:disable Metrics/MethodLength
+      def perform(file_set:, user:)
+        return true unless file_set.pdf?
+
+        work = IiifPrint.parent_for(file_set)
+
+        # Woe is ye who changes the configuration of the model, thus removing the splitting.
+        raise WorkNotConfiguredToSplitFileSetError.new(work: work, file_set: file_set) unless work&.iiif_print_config&.pdf_splitter_job&.presence
+
+        # clean up any existing spawned child works of this file_set
+        IiifPrint::SplitPdfs::DestroyPdfChildWorksService.conditionally_destroy_spawned_children_of(
+          file_set: file_set,
+          work: work
+        )
+
+        location = Hyrax::WorkingDirectory.find_or_retrieve(file_set.files.first.id, file_set.id)
+
+        IiifPrint.conditionally_submit_split_for(work: work, file_set: file_set, locations: [location], user: user)
+      end
+      # rubocop:enable Metrics/MethodLength
+    end
+  end
+end

data/app/listeners/iiif_print/listener.rb

@@ -0,0 +1,31 @@
+##
+# @see https://github.com/samvera/hyrax/wiki/Hyrax's-Event-Bus-(Hyrax::Publisher)
+# @see https://www.rubydoc.info/gems/hyrax/Hyrax/Publisher
+# @see https://dry-rb.org/gems/dry-events
+module IiifPrint
+  class Listener
+    ##
+    # Responsible for conditionally enqueuing the creation of child works from a PDF.
+    #
+    # @param event [#[]] a hash like construct with keys :user and :file_set
+    # @param service [#conditionally_enqueue]
+    #
+    # @see Hyrax::WorkUploadsHandler
+    def on_file_characterized(event, service: IiifPrint::SplitPdfs::ChildWorkCreationFromPdfService)
+      file_set = event[:file_set]
+      return false unless file_set
+      return false unless file_set.file_set?
+      return false unless file_set.original_file.pdf?
+
+      work = IiifPrint.parent_for(file_set)
+      # A short-circuit to avoid fetching the underlying file.
+      return false unless work
+
+      user = work.depositor
+      # TODO: Verify that this is the correct thing to be sending off for conditional enquing.  That
+      # will require a more involved integration test.
+      file = file_set.original_file
+      service.conditionally_enqueue(file_set: file_set, work: work, file: file, user: user)
+    end
+  end
+end

data/app/models/concerns/iiif_print/set_child_flag.rb

@@ -4,6 +4,10 @@ module RDF
   class CustomIsChildTerm < Vocabulary('http://id.loc.gov/vocabulary/identifiers/')
     property 'is_child'
   end
+
+  class FromPdfIdTerm < Vocabulary('http://id.loc.gov/vocabulary/identifiers/')
+    property 'split_from_pdf_id'
+  end
 end
 
 module IiifPrint
@@ -18,10 +22,15 @@ module IiifPrint
               multiple: false do |index|
                 index.as :stored_searchable
               end
+      property :split_from_pdf_id,
+              predicate: ::RDF::FromPdfIdTerm.split_from_pdf_id,
+              multiple: false do |index|
+                index.as :stored_searchable
+              end
     end
 
     def set_children
-      ordered_works.each do |child_work|
+      IiifPrint.object_ordered_works(self).each do |child_work|
         child_work.update(is_child: true) unless child_work.is_child
       end
     end

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

@@ -19,6 +19,8 @@ module IiifPrint::Solr::Document
   def self.decorate(base)
     base.prepend(self)
     base.send(:attribute, :is_child, Hyrax::SolrDocument::Metadata::Solr::String, 'is_child_bsi')
+    base.send(:attribute, :split_from_pdf_id, Hyrax::SolrDocument::Metadata::Solr::String, 'split_from_pdf_id_ssi')
+    base.send(:attribute, :digest, Hyrax::SolrDocument::Metadata::Solr::String, 'digest_ssim')
 
     # @note These properties came from the newspaper_works gem.  They are configurable.
     base.class_attribute :iiif_print_solr_field_names, default: %w[alternative_title genre
@@ -31,17 +33,31 @@ module IiifPrint::Solr::Document
     base
   end
 
+  def digest_sha1
+    digest[/urn:sha1:([\w]+)/, 1]
+  end
+
   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)
     iiif_print_solr_field_names.include?(method_name.to_s) || super
   end
 
-  # TODO: consider configuring this field name; we use the magic field in lots of places.
+  # @see https://github.com/samvera/hyrax/commit/7108409c619cd2ba4ae8c836b9f3b429a7e9837b
   def file_set_ids
-    self['file_set_ids_ssim']
+    # Yes, this looks a little odd.  But the truth is the prior key (e.g. `file_set_ids_ssim`) was
+    # an alias of `member_ids_ssim`.
+    self['member_ids_ssim']
+  end
+
+  def any_highlighting?
+    response&.[]('highlighting')&.[](id)&.present?
+  end
+
+  def solr_document
+    self
   end
 end

data/app/models/iiif_print/iiif_search_decorator.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+# OVERRIDE: Blacklight IIIF Search v1.0.0
+# IiifSearchDecorator module extends the functionality of the BlacklightIiifSearch::IiifSearch class
+# by overriding the solr_params method to modify the search query to include the parent's metadata.
+module IiifPrint
+  module IiifSearchDecorator
+    ##
+    # Overrides the solr_params method from BlacklightIiifSearch::IiifSearch to modify the search query.
+    # The method adds an additional filter to the query to include either the object_relation_field OR the
+    # parent document's id and removes the :f parameter from the query.
+    # :object_relation_field refers to the CatalogController's configuration which is typically set to
+    # 'is_page_of_ssim' in the host application which only searches child works by default.
+    #
+    #   config.iiif_search = {
+    #     full_text_field: 'all_text_tsimv',
+    #     object_relation_field: 'is_page_of_ssim',
+    #     supported_params: %w[q page],
+    #     autocomplete_handler: 'iiif_suggest',
+    #     suggester_name: 'iiifSuggester'
+    #   }
+    #
+    # @return [Hash] A hash containing the modified Solr search parameters
+    #
+    def solr_params
+      return { q: 'nil:nil' } unless q
+
+      {
+        q: "#{q} AND (#{iiif_config[:object_relation_field]}:\"#{parent_document.id}\" OR id:\"#{parent_document.id}\")",
+        rows: rows,
+        page: page
+      }
+    end
+  end
+end

data/app/models/iiif_print/iiif_search_response_decorator.rb

@@ -4,14 +4,37 @@ module IiifPrint
     # @see https://github.com/scientist-softserv/louisville-hyku/commit/67467e5cf9fdb755f54419f17d3c24c87032d0af
     def annotation_list
       json_results = super
-      json_results&.[]('resources')&.each do |result_hit|
+
+      # Check and process invalid hit
+      if json_results&.[]('resources')
+        remove_invalid_hit(json_results)
+        add_metadata_match(json_results)
+      end
+
+      json_results
+    end
+
+    def remove_invalid_hit(json_results)
+      resources = json_results['resources']
+      invalid_hit = resources.detect { |resource| resource["on"].include?(IiifPrint::BlacklightIiifSearch::AnnotationDecorator::INVALID_MATCH_TEXT) }
+      return unless invalid_hit
+
+      # Delete invalid hit from resources, remove first hit (which is from the invalid hit), decrement total within
+      resources.delete(invalid_hit)
+      json_results['hits'].shift
+      json_results['within']['total'] -= 1
+    end
+
+    def add_metadata_match(json_results)
+      json_results['resources'].each do |result_hit|
         next if result_hit['resource'].present?
+
+        # Add resource details if not present
         result_hit['resource'] = {
           "@type": "cnt:ContentAsText",
           "chars": "Metadata match, see sidebar for details"
         }
       end
-      json_results
     end
   end
 end

data/app/models/iiif_print/pending_relationship.rb

@@ -3,5 +3,8 @@ module IiifPrint
     validates :parent_id, presence: true
     validates :child_title, presence: true
     validates :child_order, presence: true
+    validates :parent_model, presence: true
+    validates :child_model, presence: true
+    validates :file_id, presence: true
   end
 end

data/app/presenters/iiif_print/file_set_presenter_decorator.rb

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+
+module IiifPrint
+  module FileSetPresenterDecorator
+    # uses Hyku's TenantConfig to determine whether to allow PDF splitting button
+    def show_split_button?
+      return parent.try(:split_pdfs?) if parent.respond_to?(:split_pdfs?)
+      true
+    end
+  end
+end

data/app/presenters/iiif_print/iiif_manifest_presenter_behavior.rb

@@ -3,8 +3,128 @@ module IiifPrint
   module IiifManifestPresenterBehavior
     extend ActiveSupport::Concern
 
+    # Extending the presenter to the base url which includes the protocol.
+    # We need the base url to render the facet links and normalize the interface.
+    attr_accessor :base_url
+
+    def manifest_metadata
+      # ensure we are using a SolrDocument
+      @manifest_metadata ||= IiifPrint.manifest_metadata_from(work: model.solr_document, presenter: self)
+    end
+
     def search_service
       Rails.application.routes.url_helpers.solr_document_iiif_search_url(id, host: hostname)
     end
+
+    # OVERRIDE: Hyrax 3x, avoid nil returning to IIIF Manifest gem
+    # @see https://github.com/samvera/iiif_manifest/blob/c408f90eba11bef908796c7236ba6bcf8d687acc/lib/iiif_manifest/v3/manifest_builder/record_property_builder.rb#L28
+    ##
+    # @return [Array<Hash{String => String}>]
+    def sequence_rendering
+      Array(try(:rendering_ids)).map do |file_set_id|
+        rendering = file_set_presenters.find { |p| p.id == file_set_id }
+        return [] unless rendering
+
+        { '@id' => Hyrax::Engine.routes.url_helpers.download_url(rendering.id, host: hostname),
+          'format' => rendering.mime_type.presence || I18n.t("hyrax.manifest.unknown_mime_text"),
+          'label' => I18n.t("hyrax.manifest.download_text") + (rendering.label || '') }
+      end.flatten
+    end
+
+    # OVERRIDE: Hyrax v3.x
+    module DisplayImagePresenterBehavior
+      # Extending the presenter to the base url which includes the protocol.
+      # We need the base url to render the facet links and normalize the interface.
+      attr_accessor :base_url
+
+      # Extending this class because there is an #ability= but not #ability and this definition
+      # mirrors the Hyrax::IiifManifestPresenter#ability.
+      def ability
+        @ability ||= NullAbility.new
+      end
+
+      def display_image
+        return nil unless latest_file_id
+        return nil unless model.image?
+        return nil unless IiifPrint.config.default_iiif_manifest_version == 2
+
+        IIIFManifest::DisplayImage
+          .new(display_image_url(hostname),
+            format: image_format(alpha_channels),
+            width: width,
+            height: height,
+            iiif_endpoint: iiif_endpoint(latest_file_id, base_url: hostname))
+      end
+
+      # OVERRIDE: IIIF Hyrax AV v0.2 #display_content for prez 3 manifests
+      def display_content
+        return nil unless latest_file_id
+        return super unless model.image?
+
+        IIIFManifest::V3::DisplayContent
+          .new(display_image_url(hostname),
+            format: image_format(alpha_channels),
+            width: width,
+            height: height,
+            type: 'Image',
+            iiif_endpoint: iiif_endpoint(latest_file_id, base_url: hostname))
+      end
+
+      def display_image_url(base_url)
+        if ENV['EXTERNAL_IIIF_URL'].present?
+          # At the moment we are only concerned about Hyrax's default image url builder
+          iiif_image_url_builder(url_builder: Hyrax.config.iiif_image_url_builder)
+        else
+          super
+        end
+      end
+
+      def iiif_endpoint(file_id, base_url: request.base_url)
+        if ENV['EXTERNAL_IIIF_URL'].present?
+          IIIFManifest::IIIFEndpoint.new(
+            File.join(ENV['EXTERNAL_IIIF_URL'], file_id),
+            profile: Hyrax.config.iiif_image_compliance_level_uri
+          )
+        else
+          super
+        end
+      end
+
+      def hostname
+        @hostname || 'localhost'
+      end
+
+      ##
+      # @return [Boolean] false
+      def work?
+        false
+      end
+
+      private
+
+      def latest_file_id
+        if ENV['EXTERNAL_IIIF_URL'].present?
+          external_latest_file_id
+        else
+          super
+        end
+      end
+
+      def external_latest_file_id
+        @latest_file_id ||= digest_sha1
+      end
+
+      def iiif_image_url_builder(url_builder:)
+        args = [
+          latest_file_id,
+          ENV['EXTERNAL_IIIF_URL'],
+          Hyrax.config.iiif_image_size_default
+        ]
+        # In Hyrax 3, Hyrax.config.iiif_image_url_builder takes an additional argument
+        args << image_format(alpha_channels) if url_builder.arity == 4
+
+        url_builder.call(*args).gsub(%r{images/}, '')
+      end
+    end
   end
 end

data/app/presenters/iiif_print/iiif_manifest_presenter_factory_behavior.rb

@@ -14,7 +14,7 @@ module IiifPrint
           presenter_class.for(solr_doc)
         elsif Hyrax.config.curation_concerns.include?(solr_doc.hydra_model)
           # look up file set ids and loop through those
-          file_set_docs = load_file_set_docs(solr_doc.file_set_ids)
+          file_set_docs = load_file_set_docs(solr_doc.try(:member_ids) || solr_doc.try(:[], 'member_ids_ssim'))
           file_set_docs.map { |doc| presenter_class.for(doc) } if file_set_docs.length
         end
       end.flatten.compact

data/app/presenters/iiif_print/work_show_presenter_decorator.rb

@@ -2,28 +2,40 @@
 
 module IiifPrint
   module WorkShowPresenterDecorator
-    delegate :file_set_ids, to: :solr_document
+    delegate :member_ids, to: :solr_document
+    alias file_set_ids member_ids
 
-    # OVERRIDE Hyrax 2.9.6 to remove check for representative_presenter.image? and allow
-    # a fallback to check for images on the child works
+    # OVERRIDE Hyrax 2.9.6 to remove check for representative_presenter.image?
     # @return [Boolean] render a IIIF viewer
     def iiif_viewer?
-      parent_work_has_files? || child_work_has_files?
+      Hyrax.config.iiif_image_server? &&
+        representative_id.present? &&
+        representative_presenter.present? &&
+        members_include_viewable_image?
     end
 
     alias universal_viewer? iiif_viewer?
 
     private
 
-    def parent_work_has_files?
-      Hyrax.config.iiif_image_server? &&
-        representative_id.present? &&
-        representative_presenter.present? &&
-        members_include_viewable_image?
+    # overriding Hyrax to include file sets for both work and child works (file set ids include both)
+    # process each id, short-circuiting the loop once one true value is found. This speeds up the test
+    # by not loading more member_presenters than needed.
+    #
+    # @todo Review if this is necessary for Hyrax 5.
+    def members_include_viewable_image?
+      all_member_ids = solr_document.try(:member_ids) || solr_document.try(:[], 'member_ids_ssim')
+      Array.wrap(all_member_ids).each do |id|
+        return true if file_type_and_permissions_valid?(member_presenters_for([id]).first)
+      end
+      false
     end
 
-    def child_work_has_files?
-      file_set_ids.present?
+    # This method allows for overriding to add additional file types to mix in with IiifAv
+    # TODO: add configuration setting for file types to loop through so an override is unneeded.
+    def file_type_and_permissions_valid?(presenter)
+      current_ability.can?(:read, presenter.id) &&
+        (presenter.try(:image?) || presenter.try(:solr_document).try(:image?))
     end
   end
 end

data/app/search_builders/concerns/iiif_print/allinson_flex_fields.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module IiifPrint
+  module AllinsonFlexFields
+    def include_allinson_flex_fields(solr_parameters)
+      return unless defined?(AllinsonFlex)
+
+      query_fields = solr_parameters[:qf].split(' ') + IiifPrint.allinson_flex_fields
+                                                                .each_with_object([]) do |field, arr|
+                                                         arr << (field.name + '_tesim') if field.is_a?(AllinsonFlex::ProfileProperty)
+                                                       end
+      solr_parameters[:qf] = query_fields.uniq.join(' ')
+    end
+  end
+end

data/app/search_builders/concerns/iiif_print/highlight_search_params.rb

@@ -6,9 +6,10 @@ module IiifPrint
     def highlight_search_params(solr_parameters = {})
       return unless solr_parameters[:q] || solr_parameters[:all_fields]
       solr_parameters[:hl] = true
-      solr_parameters[:'hl.fl'] = 'all_text_tsimv'
+      solr_parameters[:'hl.fl'] = '*'
       solr_parameters[:'hl.fragsize'] = 100
       solr_parameters[:'hl.snippets'] = 5
+      solr_parameters[:'hl.requiredFieldMatch'] = true
     end
   end
 end