Transitions alternate_ids to emory_persistent_id. (#616)

* Transitions alternate_ids to emory_persistent_id.

* Add a little spec.

* Add a little more spec.

* Rubo happiness.

---------

Co-authored-by: Alex Zotov <[email protected]>

app/controllers/catalog_controller.rb

@@ -85,7 +85,7 @@ def self.modified_field
     # Emory Note: the fields are no longer dynamically delivered via this config setting.
     #   Instead, they are hard coded in views/hyrax/base/_attribute_rows.html.erb. We, in turn,
     #   have overridden tht partial so we can institute our own choices/order.
-    config.add_show_field "alternate_ids_ssim"
+    config.add_show_field "emory_persistent_id_ssi"
     config.add_show_field "content_genre_tesi"
     config.add_show_field "title_tesim"
     config.add_show_field "creator_tesim"

app/controllers/purl_controller.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 class PurlController < ApplicationController
   def redirect_to_original
-    purl_object = Hyrax.custom_queries.find_by_alternate_id(alternate_ids: params[:alternate_ids])
+    purl_object = Hyrax.custom_queries.find_by_emory_persistent_id(emory_persistent_id: params[:emory_persistent_id])
 
     if purl_object.work?
       object_url = "#{request.base_url}/concern/publications/#{purl_object.id}"

app/indexers/collection_resource_indexer.rb

@@ -7,15 +7,8 @@ class CollectionResourceIndexer < Hyrax::PcdmCollectionIndexer
   include Hyrax::Indexer(:collection_resource)
 
   # Uncomment this block if you want to add custom indexing behavior:
-  def to_solr
-    super.tap do |index_document|
-      index_document[:alternate_ids_ssim] = find_alternate_ids
-    end
-  end
-
-  private
-
-  def find_alternate_ids
-    resource.alternate_ids.map(&:id)
-  end
+  # def to_solr
+  #   super.tap do |index_document|
+  #   end
+  # end
 end

app/indexers/publication_indexer.rb

@@ -12,7 +12,6 @@ def to_solr
       index_document[:failed_preservation_events_ssim] = failed_preservation_events
       index_document[:preservation_events_tesim] = resource&.preservation_events&.map(&:preservation_event_terms)
       index_document[:preservation_workflow_terms_tesim] = preservation_workflow_terms
-      index_document[:alternate_ids_ssim] = find_alternate_ids
       index_document[:all_text_tsimv] = resource&.primary_file_set&.extracted_text_content
       index_document[:member_of_collections_ssim] = collection_names_for_facets
     end
@@ -26,10 +25,6 @@ def failed_preservation_events
     failures.map(&:failed_event_json)
   end
 
-  def find_alternate_ids
-    resource.alternate_ids.map(&:id)
-  end
-
   def preservation_workflow_terms
     resource.preservation_workflows.map(&:preservation_terms)
   end

app/services/self_deposit/custom_queries/find_all_objects_with_alternate_ids_present.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+module SelfDeposit
+  module CustomQueries
+    # @example
+    #
+    #   Hyrax.custom_queries.find_all_objects_with_alternate_ids_present
+    #
+    # @see https://github.com/samvera/valkyrie/wiki/Queries#custom-queries
+    class FindAllObjectsWithAlternateIdsPresent
+      def self.queries
+        [:find_all_objects_with_alternate_ids_present]
+      end
+
+      def initialize(query_service:)
+        @query_service = query_service
+      end
+
+      attr_reader :query_service
+
+      ##
+      # @return enumerator of Valkyrie Fedora objects
+      def find_all_objects_with_alternate_ids_present
+        enum_for(:each)
+      end
+
+      # Queries the Solr index for parent works of the provided resource
+      # For each Document, it yields the pulled Hyrax Work object
+      # @yield [Valkyrie::Resources]
+      def each
+        objects = @query_service.find_all
+        objects.each do |obj|
+          yield obj if obj.respond_to?(:alternate_ids) && obj.alternate_ids.present?
+        end
+      end
+    end
+  end
+end

app/services/self_deposit/custom_queries/find_by_emory_persistent_id.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+class SelfDeposit::CustomQueries::FindByEmoryPersistentId < ::SelfDeposit::CustomQueries::SolrDocumentQuery
+  self.queries = [:find_by_emory_persistent_id]
+
+  def find_by_emory_persistent_id(emory_persistent_id:)
+    @emory_persistent_id = emory_persistent_id
+    raise ::Valkyrie::Persistence::ObjectNotFoundError unless resource
+    @query_service.find_by(id: resource['id'])
+  end
+
+  def query
+    "emory_persistent_id_ssi:#{@emory_persistent_id}"
+  end
+end

app/services/self_deposit/valkyrie_object_remediation_service.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module SelfDeposit
+  class ValkyrieObjectRemediationService
+    def self.migrate_alternate_ids_to_emory_persistent_id
+      operating_ids = pull_operating_ids
+      process_objects_with_alternate_ids
+      operating_ids
+    end
+
+    class << self
+      private
+
+      def objects_with_alternate_ids
+        Hyrax.custom_queries.find_all_objects_with_alternate_ids_present
+      end
+
+      def pull_operating_ids
+        objects_with_alternate_ids.map(&:id).join(', ')
+      end
+
+      def process_objects_with_alternate_ids
+        objects_with_alternate_ids.each do |obj|
+          value_in_alternate_ids = obj.alternate_ids.map(&:to_s).first
+          obj.emory_persistent_id = value_in_alternate_ids if obj.emory_persistent_id.blank?
+          obj.alternate_ids = []
+          Hyrax.persister.save(resource: obj)
+        end
+      end
+    end
+  end
+end

app/views/hyrax/base/_purl.html.erb

@@ -1,8 +1,8 @@
 <dt>Persistent URL</dt>
 <dd>
   <ul class="tabular">
-    <% if presenter.solr_document['alternate_ids_ssim'].present? %>
-        <% url = "#{request.base_url}/purl/#{presenter.solr_document['alternate_ids_ssim'].first}" %>
+    <% if presenter.solr_document['emory_persistent_id_ssi'].present? %>
+        <% url = "#{request.base_url}/purl/#{presenter.solr_document['emory_persistent_id_ssi']}" %>
         <li><a href="<%= url %>"><%= url %></a></li>
     <% else %>
         <li>No alternate ID available</li>

app/views/hyrax/collections/_show_descriptions.html.erb

@@ -4,8 +4,8 @@
   <div class='metadata-line'>
     <div class='col-xs-12 col-sm-4'><dt >Persistent URL</dt></div>
     <div class='col-xs-12 col-sm-8'><dd>
-    <% if @presenter.solr_document['alternate_ids_ssim'].present? %>
-        <% url = "#{request.base_url}/purl/#{@presenter.solr_document['alternate_ids_ssim'].first}" %>
+    <% if @presenter.solr_document['emory_persistent_id_ssi'].present? %>
+        <% url = "#{request.base_url}/purl/#{@presenter.solr_document['emory_persistent_id_ssi']}" %>
         <a href="<%= url %>"><%= url %></a>
     <% else %>
         No alternate ID available
@@ -19,4 +19,4 @@
       <div class='col-xs-12 col-sm-8'><dd><%= collection_metadata_value(@presenter, field_name) %></dd></div>
     </div>
   <% end %>
-</dl>
+</dl>

app/views/hyrax/dashboard/collections/_show_descriptions.html.erb

@@ -5,8 +5,8 @@
   <div class='metadata-line'>
     <div class='col-xs-12 col-sm-4'><dt >Persistent URL</dt></div>
     <div class='col-xs-12 col-sm-8'><dd>
-    <% if @presenter.solr_document['alternate_ids_ssim'].present? %>
-        <% url = "#{request.base_url}/purl/#{@presenter.solr_document['alternate_ids_ssim'].first}" %>
+    <% if @presenter.solr_document['emory_persistent_id_ssi'].present? %>
+        <% url = "#{request.base_url}/purl/#{@presenter.solr_document['emory_persistent_id_ssi']}" %>
         <a href="<%= url %>"><%= url %></a>
     <% else %>
         No alternate ID available
@@ -21,4 +21,4 @@
       <dd class="col-7"><%= collection_metadata_value(@presenter, field_name) %></dd>
     </div>
   <% end %>
-</dl>
+</dl>

config/initializers/hyrax.rb

@@ -338,9 +338,10 @@
                   Hyrax::CustomQueries::FindModelsByAccess,
                   Hyrax::CustomQueries::FindCountBy,
                   Hyrax::CustomQueries::FindByDateRange,
+                  SelfDeposit::CustomQueries::FindAllObjectsWithAlternateIdsPresent,
                   SelfDeposit::CustomQueries::FindPublicationByDeduplicationKey,
                   SelfDeposit::CustomQueries::FindParentWorks,
-                  SelfDeposit::CustomQueries::FindByAlternateId,
+                  SelfDeposit::CustomQueries::FindByEmoryPersistentId,
                   SelfDeposit::CustomQueries::FindByCollectionTitle,
                   SelfDeposit::CustomQueries::FindBySourceIdentifier]
 custom_queries.each do |handler|

config/initializers/hyrax_work_uploads_handler_override.rb

@@ -77,7 +77,7 @@ def file_set_args(file)
       { depositor: file.user.user_key,
         creator: file.user.user_key,
         date_uploaded: file.created_at,
-        alternate_ids: ["#{Noid::Rails::Service.new.mint}-emory"],
+        emory_persistent_id: "#{Noid::Rails::Service.new.mint}-emory",
         date_modified: Hyrax::TimeService.time_in_utc,
         label: file.fileset_name.presence || file.uploader.filename,
         title: file.fileset_name.presence || file.uploader.filename,

config/metadata/emory_basic_metadata.yaml

@@ -97,6 +97,12 @@ attributes:
     index_keys:
       - 'emory_ark_tesim'
     predicate: http://id.loc.gov/vocabulary/identifiers/local#ark
+  emory_persistent_id:
+    type: string
+    multiple: false
+    index_keys:
+      - 'emory_persistent_id_ssi'
+    predicate: http://metadata.emory.edu/vocab/cor-terms#emoryPersistentId
   holding_repository:
     type: string
     multiple: false

config/metadata/emory_file_set_metadata.yaml

@@ -1,4 +1,10 @@
 attributes:
+  emory_persistent_id:
+    type: string
+    multiple: false
+    index_keys:
+      - 'emory_persistent_id_ssi'
+    predicate: http://metadata.emory.edu/vocab/cor-terms#emoryPersistentId
   file_set_use:
     type: string
     multiple: false

config/routes.rb

@@ -23,7 +23,7 @@
   mount Hyrax::Engine, at: '/'
   resources :welcome, only: 'index'
   root 'hyrax/homepage#index'
-  get 'purl/:alternate_ids', to: 'purl#redirect_to_original'
+  get 'purl/:emory_persistent_id', to: 'purl#redirect_to_original'
   curation_concerns_basic_routes
   concern :exportable, Blacklight::Routes::Exportable.new
 

lib/hyrax/transactions/steps/set_noid_id.rb

@@ -5,22 +5,22 @@ module Hyrax
   module Transactions
     module Steps
       ##
-      # A step that sets the `#collection_type_gid` in the change set.
+      # A step that sets the `#emory_persistent_id` in the change set.
       #
       # @since 3.2.0
       class SetNoidId
         include Dry::Monads[:result]
 
         ##
         # @param [Hyrax::ChangeSet] change_set
-        # @param alternate_ids for the collection, file_set, publication
-        def call(change_set, alternate_ids: [new_noid_id])
-          return Failure[:no_alternate_ids, change_set] unless
-            change_set.respond_to?(:alternate_ids=)
+        # @param emory_persistent_id for the collection, file_set, publication
+        def call(change_set, emory_persistent_id: new_noid_id)
+          return Failure[:no_emory_persistent_id, change_set] unless
+            change_set.respond_to?(:emory_persistent_id=)
           return Success(change_set) if
-            change_set.alternate_ids.present?
+            change_set.emory_persistent_id.present?
 
-          change_set.alternate_ids = alternate_ids
+          change_set.emory_persistent_id = emory_persistent_id
           Hyrax.persister.save(resource: change_set)
           Success(change_set)
         end

lib/tasks/valkyrie_object_remediation.rake

@@ -0,0 +1,11 @@
+# frozen_string_literal: true
+namespace :selfdeposit do
+  namespace :valkyrie_object_remediation do
+    desc "Moves 'alternate_ids' values to 'emory_persistent_id'."
+    task migrate_alternate_ids_to_emory_persistent_id: :environment do
+      operating_ids = ::SelfDeposit::ValkyrieObjectRemediationService.migrate_alternate_ids_to_emory_persistent_id
+
+      puts "Valkyrie object IDs remediated: #{operating_ids}"
+    end
+  end
+end

spec/controllers/catalog_controller_spec.rb

@@ -4,7 +4,7 @@
 RSpec.describe CatalogController, type: :controller do
   include Devise::Test::ControllerHelpers
 
-  expected_show_fields = ["alternate_ids_ssim", "content_genre_tesi", "title_tesim", "creator_tesim",
+  expected_show_fields = ["emory_persistent_id_ssi", "content_genre_tesi", "title_tesim", "creator_tesim",
                           "language_tesim", "date_issued_tesi", "publisher_tesim", "publisher_version_tesi",
                           "parent_title_tesi", "conference_name_tesi", "issn_tesi", "isbn_tesi",
                           "series_title_tesi", "grant_agencies_tesim", "grant_information_tesim",

spec/controllers/purl_controller_spec.rb

@@ -12,10 +12,10 @@
     let(:admin_set) { FactoryBot.valkyrie_create(:hyrax_admin_set) }
     let(:permission_template) { FactoryBot.create(:permission_template, source_id: admin_set.id) }
     let(:workflow) { FactoryBot.create(:workflow, allows_access_grant: true, active: true, permission_template_id: permission_template.id) }
-    let(:publication_alt_ids) { ['1234567-emory'] }
-    let!(:publication) { FactoryBot.valkyrie_create(:publication, with_index: true, admin_set_id: admin_set.id, depositor: user.user_key, alternate_ids: publication_alt_ids) }
-    let(:collection_alt_ids) { ['89abcef-emory'] }
-    let!(:collection) { FactoryBot.valkyrie_create(:hyrax_collection, alternate_ids: collection_alt_ids) }
+    let(:publication_alt_id) { '1234567-emory' }
+    let!(:publication) { FactoryBot.valkyrie_create(:publication, with_index: true, admin_set_id: admin_set.id, depositor: user.user_key, emory_persistent_id: publication_alt_id) }
+    let(:collection_alt_id) { '89abcef-emory' }
+    let!(:collection) { FactoryBot.valkyrie_create(:hyrax_collection, emory_persistent_id: collection_alt_id) }
 
     before do
       Hyrax.index_adapter.wipe!
@@ -25,7 +25,7 @@
     context 'when purl_object is a Publication' do
       it 'redirects to the publication URL' do
         Hyrax.index_adapter.save(resource: publication)
-        get :redirect_to_original, params: { alternate_ids: publication_alt_ids.first }
+        get :redirect_to_original, params: { emory_persistent_id: publication_alt_id }
 
         expect(response).to redirect_to("http://test.host/concern/publications/#{publication.id}")
       end
@@ -34,15 +34,15 @@
     context 'when purl_object is a Collection' do
       it 'redirects to the collection URL' do
         Hyrax.index_adapter.save(resource: collection)
-        get :redirect_to_original, params: { alternate_ids: collection_alt_ids.first }
+        get :redirect_to_original, params: { emory_persistent_id: collection_alt_id }
 
         expect(response).to redirect_to("http://test.host/collections/#{collection.id}")
       end
     end
 
     context 'when purl_object is not found' do
       before do
-        get :redirect_to_original, params: { alternate_ids: admin_set.id }
+        get :redirect_to_original, params: { emory_persistent_id: admin_set.id }
       end
 
       it 'returns a 404 not found status' do

spec/hyrax/transactions/steps/set_noid_id_spec.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+require 'rails_helper'
+require 'hyrax/transactions'
+
+RSpec.describe Hyrax::Transactions::Steps::SetNoidId do
+  subject(:step) { described_class.new }
+  let(:publication) { FactoryBot.valkyrie_create(:publication) }
+
+  describe '#call' do
+    context 'without a passed ID' do
+      it 'is a success' do
+        expect(step.call(publication)).to be_success
+        expect(publication.emory_persistent_id).to be_present
+      end
+    end
+
+    context 'with a passed ID' do
+      let(:fake_id) { '123jkdcjh-emory' }
+      it 'is a success' do
+        expect(step.call(publication, emory_persistent_id: fake_id)).to be_success
+        expect(publication.emory_persistent_id).to eq(fake_id)
+      end
+    end
+  end
+end

spec/indexers/collection_resource_indexer_spec.rb

@@ -13,15 +13,15 @@
   it_behaves_like 'a Hyrax::Resource indexer'
   it_behaves_like 'a Basic metadata indexer'
 
-  context 'alternate_ids_ssim' do
-    let(:alternate_ids) { [double(id: 'alt_id_1'), double(id: 'alt_id_2')] }
+  context 'emory_persistent_id_ssi' do
+    let(:emory_persistent_id) { 'alt_id_1' }
 
     before do
-      allow(resource).to receive(:alternate_ids).and_return(alternate_ids)
+      allow(resource).to receive(:emory_persistent_id).and_return(emory_persistent_id)
     end
 
     it 'contains an array of alternate IDs' do
-      expect(indexer.to_solr['alternate_ids_ssim']).to eq(['alt_id_1', 'alt_id_2'])
+      expect(indexer.to_solr['emory_persistent_id_ssi']).to eq('alt_id_1')
     end
   end
 end