[ESSI-875] bulkrax 5 upgrade

Gemfile

@@ -101,8 +101,7 @@ gem 'allinson_flex', github: 'IU-Libraries-Joint-Development/allinson_flex'
 gem 'okcomputer'
 
 # Bulk Import / Export
-gem 'bulkrax', '~> 1.0.0'
-gem 'willow_sword', github: 'notch8/willow_sword', ref: '0a669d7' # deprecate willow_sword in bulkrax 5 upgrade
+gem 'bulkrax', '~> 5.0'
 gem 'webpacker'
 gem 'react-rails'
 

Gemfile.lock

@@ -1,23 +1,13 @@
 GIT
   remote: https://github.com/IU-Libraries-Joint-Development/allinson_flex.git
-  revision: 86ecb693044f451811a52033b643c91f9a241c3b
+  revision: 0fadf99b61b6567f240927e7dd54830878b315ac
   specs:
     allinson_flex (0.1.0)
       json_schemer
       rails (>= 5.1.6)
       react-rails
       webpacker
 
-GIT
-  remote: https://github.com/notch8/willow_sword.git
-  revision: 0a669d78617c6003e4aa1a46a10447be92be27d5
-  ref: 0a669d7
-  specs:
-    willow_sword (0.2.0)
-      bagit (~> 0.4.1)
-      rails (>= 5.1.6)
-      rubyzip (>= 1.0.0)
-
 GIT
   remote: https://github.com/rtomayko/posix-spawn.git
   revision: 0fce38ed5458b638eda5f3bb711903424a4366db
@@ -206,18 +196,19 @@ GEM
       sprockets (~> 3.7)
       typhoeus
     builder (3.2.4)
-    bulkrax (1.0.2)
+    bulkrax (5.5.1)
       bagit (~> 0.4)
       coderay
       iso8601 (~> 0.9.0)
       kaminari
       language_list (~> 1.2, >= 1.2.1)
-      libxml-ruby (~> 3.1.0)
+      libxml-ruby (~> 3.2.4)
       loofah (>= 2.2.3)
       oai (>= 0.4, < 2.x)
       rack (>= 2.0.6)
       rails (>= 5.1.6)
       rdf (>= 2.0.2, < 4.0)
+      rubyzip
       simple_form
     byebug (11.1.3)
     cancancan (1.17.0)
@@ -637,7 +628,7 @@ GEM
       rdf (~> 3.0)
     legato (0.7.0)
       multi_json
-    libxml-ruby (3.1.0)
+    libxml-ruby (3.2.4)
     link_header (0.0.8)
     linkeddata (3.1.1)
       equivalent-xml (~> 0.6)
@@ -1119,7 +1110,7 @@ DEPENDENCIES
   bixby
   blacklight_iiif_search
   browse-everything (= 1.1.0)
-  bulkrax (~> 1.0.0)
+  bulkrax (~> 5.0)
   byebug
   capybara
   capybara-screenshot (~> 1.0)
@@ -1178,7 +1169,6 @@ DEPENDENCIES
   webdrivers
   webmock
   webpacker
-  willow_sword!
 
 BUNDLED WITH
    2.4.22

app/assets/stylesheets/brand_tweaks.css

@@ -90,3 +90,8 @@ html {
 	height: 1rem !important;
 	width: 1rem !important;
 }
+
+/* cherry-picked from bootstrap, for bulkrax 5 implicit requirement */
+.d-none {
+  display: none !important;
+}

app/jobs/bulkrax/import_file_set_job.rb

@@ -0,0 +1,76 @@
+# frozen_string_literal: true
+# @todo drop this direct import after upgrading to bulkrax commit 0de8ee06 or later
+
+module Bulkrax
+  class MissingParentError < ::StandardError; end
+
+  class ImportFileSetJob < ApplicationJob
+    include DynamicRecordLookup
+
+    queue_as Bulkrax.config.ingest_queue_name
+
+    attr_reader :importer_run_id
+
+    def perform(entry_id, importer_run_id)
+      @importer_run_id = importer_run_id
+      entry = Entry.find(entry_id)
+      # e.g. "parents" or "parents_1"
+      parent_identifier = (entry.raw_metadata[entry.related_parents_raw_mapping] || entry.raw_metadata["#{entry.related_parents_raw_mapping}_1"])&.strip
+
+      begin
+        validate_parent!(parent_identifier)
+      rescue MissingParentError => e
+        handle_retry(entry, importer_run_id, e)
+        return
+      end
+
+      entry.build
+      if entry.succeeded?
+        # rubocop:disable Rails/SkipsModelValidations
+        ImporterRun.increment_counter(:processed_records, importer_run_id)
+        ImporterRun.increment_counter(:processed_file_sets, importer_run_id)
+      else
+        ImporterRun.increment_counter(:failed_records, importer_run_id)
+        ImporterRun.increment_counter(:failed_file_sets, importer_run_id)
+        # rubocop:enable Rails/SkipsModelValidations
+      end
+      ImporterRun.decrement_counter(:enqueued_records, importer_run_id) unless ImporterRun.find(importer_run_id).enqueued_records <= 0 # rubocop:disable Rails/SkipsModelValidations
+      entry.save!
+      entry.importer.current_run = ImporterRun.find(importer_run_id)
+      entry.importer.record_status
+    end
+
+    private
+
+    attr_reader :parent_record
+
+    def validate_parent!(parent_identifier)
+      # if parent_identifier is missing, it will be caught by #validate_presence_of_parent!
+      return if parent_identifier.blank?
+
+      find_parent_record(parent_identifier)
+      check_parent_is_a_work!(parent_identifier)
+    end
+
+    def check_parent_is_a_work!(parent_identifier)
+      error_msg = %(A record with the ID "#{parent_identifier}" was found, but it was a #{parent_record.class}, which is not an valid/available work type)
+      raise ::StandardError, error_msg unless curation_concern?(parent_record)
+    end
+
+    def find_parent_record(parent_identifier)
+      _, @parent_record = find_record(parent_identifier, importer_run_id)
+      raise MissingParentError, %(Unable to find a record with the identifier "#{parent_identifier}") unless parent_record
+    end
+
+    def handle_retry(entry, importer_run_id, e)
+      entry.import_attempts += 1
+      entry.save!
+      if entry.import_attempts < 5
+        ImportFileSetJob.set(wait: (entry.import_attempts + 1).minutes).perform_later(entry.id, importer_run_id)
+      else
+        ImporterRun.decrement_counter(:enqueued_records, importer_run_id) # rubocop:disable Rails/SkipsModelValidations
+        entry.set_status_info(e)
+      end
+    end
+  end
+end

app/models/ability.rb

@@ -24,6 +24,27 @@ def custom_permissions
     end
   end
 
+  # bulkrax import
+  def can_import_works?
+    can_create_any_work?
+  end
+
+  # bulkrax export
+  def can_export_works?
+    current_user.admin? || can_manage_works?
+  end
+
+  def can_manage_works?
+    @can_manage_works ||= begin
+      managing_role = Sipity::Role.find_by(name: Hyrax::RoleRegistry::MANAGING)
+      return false unless managing_role
+      Hyrax::Workflow::PermissionQuery.scope_processing_agents_for(user: current_user).any? do |agent|
+        agent.workflow_responsibilities.joins(:workflow_role)
+             .where('sipity_workflow_roles.role_id' => managing_role.id).any?
+      end
+    end
+  end
+
   # Modified method from blacklight-access_controls Blacklight::AccessControls::Ability
   # Grants registered status for authenticated visibility ("Institution") by ldap group membership, if so configured, and admins
   def user_groups

app/parsers/bulkrax/mets_xml_parser.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+# @todo update for bulkrax 5+, where XML import isn't fully supported; currently broken
 module Bulkrax
   class MetsXmlParser < ApplicationParser
     def entry_class
@@ -15,6 +16,9 @@ def create_collections; end
     # @todo not yet supported
     def import_fields; end
 
+    # @todo not yet supported
+    def file_set_entry_class; end
+
     def valid_import?
       raise StandardError, 'No metadata files found' if metadata_paths.blank?
       raise StandardError, 'No records found' if records.blank?