Merge pull request #4343 from sul-dlss/indexing-integration-tests

add some Indexing and search behavior integration tests

app/helpers/blacklight_config_helper.rb

@@ -2,6 +2,8 @@
 
 # Used in CatalogController.  This module is split out because catalog_controller.rb
 #  contains all the required blacklight configurations and is plenty large enough.
+#
+# See spec/features/indexing_xxx_spec.rb for relevancy tests of Solr search results
 module BlacklightConfigHelper
   # sending these values to Solr as arguments with search requests will override
   #   the default params configured for Solr searching via solrconfig.xml

solr_conf/conf/solrconfig.xml

@@ -21,11 +21,11 @@
       <str name="dir">${solr.core1.data.dir:}</str>
     </updateLog>
     <autoCommit>
-      <maxTime>${solr.autoCommit.maxTime:600000}</maxTime><!-- 10 minutes -->
+      <maxTime>${solr.autoCommit.maxTime:500}</maxTime><!-- 0.5 seconds (for local tests) -->
       <openSearcher>false</openSearcher>
     </autoCommit>
     <autoSoftCommit>
-      <maxTime>${solr.autoSoftCommit.maxTime:5000}</maxTime><!-- 5 seconds -->
+      <maxTime>${solr.autoSoftCommit.maxTime:100}</maxTime><!-- 0.1 seconds (for local tests) -->
     </autoSoftCommit>
   </updateHandler>
 

spec/factories/items.rb

@@ -1,5 +1,8 @@
 # frozen_string_literal: true
 
+# Because we are instantiating an immutable cocina-model rather than a database record,
+#   nested hash structure values (e.g. a title value in a description)
+#   may require a setter method in ItemMethodSender before they can be passed in
 FactoryBot.define do
   factory :persisted_item, class: 'Cocina::Models::RequestDRO' do
     initialize_with do
@@ -8,9 +11,7 @@
                                        'type' => type,
                                        'label' => label,
                                        'version' => 1,
-                                       'identification' => {
-                                         'sourceId' => "sul:#{SecureRandom.uuid}"
-                                       },
+                                       'identification' => identification,
                                        'administrative' => {
                                          'hasAdminPolicy' => admin_policy_id
                                        },
@@ -30,6 +31,12 @@
     admin_policy_id { 'druid:hv992ry2431' }
     label { 'test object' }
     type { Cocina::Models::ObjectType.object }
+    source_id { "sul:#{SecureRandom.uuid}" }
+    identification do
+      {
+        'sourceId' => source_id
+      }
+    end
 
     factory :agreement do
       type { Cocina::Models::ObjectType.agreement }

spec/features/indexing_identifiers_spec.rb

@@ -0,0 +1,170 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+# Integration tests for expected behaviors of our Solr indexing choices, through
+#   our whole stack: tests create cocina objects with factories, write them
+#   to dor-services-app, index the new objects via dor-indexing-app and then use
+#   the Argo UI to test Solr behavior such as search results and facet values.
+RSpec.describe 'Indexing and search results for identifiers' do
+  let(:item) { FactoryBot.create_for_repository(:persisted_item) }
+  let(:blacklight_config) { CatalogController.blacklight_config }
+  let(:solr_conn) { blacklight_config.repository_class.new(blacklight_config).connection }
+  let(:solr_id) { item.externalIdentifier }
+
+  before do
+    sign_in create(:user), groups: ['sdr:administrator-role']
+    solr_conn.commit # ensure no deletes are pending
+    visit '/'
+  end
+
+  after do
+    solr_conn.delete_by_id(solr_id)
+    solr_conn.commit
+  end
+
+  describe 'identifier searching' do
+    context 'for druids' do
+      let(:prefixed_druid) { item.externalIdentifier }
+
+      it 'matches query with bare and prefixed druid' do
+        [prefixed_druid, prefixed_druid.split(':').last].each do |query|
+          fill_in 'q', with: query
+          click_button 'search'
+          expect(page).to have_content('1 entry found')
+          expect(page).to have_css('dd.blacklight-id', text: solr_id)
+        end
+      end
+    end
+
+    context 'for sourceids' do
+      # SPEC: Source ID: M2549_2022-259_stertzer, where M2549 is the collection number and 2022-259 is the accession number
+      # sul:M0997_S1_B473_021_0001 (S is for series, B is for box, F is for folder ...)
+      let(:source_id) { "sul:M2549_2022-259_stertzer_#{SecureRandom.alphanumeric(12)}" }
+      let(:item) { FactoryBot.create_for_repository(:persisted_item, source_id:) }
+
+      before do
+        item.identification.sourceId # ensure item is created before searching
+      end
+
+      it 'matches whole string, including prefix before first colon' do
+        fill_in 'q', with: source_id
+        click_button 'search'
+        # expect a single result, but Solr may not finish commit for previous test delete in time
+        # expect(page).to have_content('1 entry found')
+        expect(page).to have_css('dd.blacklight-id', text: solr_id)
+      end
+
+      it 'matches without prefix before the first colon' do
+        fill_in 'q', with: source_id.split(':').last
+        click_button 'search'
+        # expect a single result, but Solr may not finish commit for previous test delete in time
+        # expect(page).to have_content('1 entry found')
+        expect(page).to have_css('dd.blacklight-id', text: solr_id)
+      end
+
+      it 'matches source_id fragments' do
+        fragments = [
+          'M2549',
+          '2022-259', # accession number
+          'M2549_2022-259',
+          'M2549 2022 259',
+          'stertzer',
+          '259_stertzer',
+          '259-stertzer',
+          '259 stertzer'
+        ]
+        fragments.each do |fragment|
+          fill_in 'q', with: fragment
+          click_button 'search'
+          # expect a single result, but Solr may not finish commit for previous test delete in time
+          # expect(page).to have_content('1 entry found')
+          expect(page).to have_css('dd.blacklight-id', text: solr_id)
+        end
+      end
+
+      it 'is not case sensitive' do
+        fill_in 'q', with: 'm2549 STERTZER'
+        click_button 'search'
+        # expect a single result, but Solr may not finish commit for previous test delete in time
+        # expect(page).to have_content('1 entry found')
+        expect(page).to have_css('dd.blacklight-id', text: solr_id)
+      end
+
+      punctuation_source_ids = [
+        'sulcons:8552-RB_Miscellanies_agabory,Before treatment photos',
+        'Archiginnasio:Bassi_Box10_Folder2_Item3.14',
+        'Revs:2012-015GHEW-CO-1980-b1_1.16_0007'
+      ]
+      punctuation_source_ids.each do |punctuation_source_id|
+        context "when punctuation in #{punctuation_source_id}" do
+          let(:source_id) { "#{punctuation_source_id}.#{SecureRandom.alphanumeric(4)}" }
+
+          it 'matches without punctuation' do
+            fill_in 'q', with: source_id.gsub(/[_\-:.,]/, ' ')
+            click_button 'search'
+            expect(page).to have_css('dd.blacklight-id', text: solr_id)
+          end
+        end
+      end
+    end
+
+    context 'for barcodes' do
+      let(:barcode) { '20503740296' }
+      let(:item) do
+        FactoryBot.create_for_repository(:persisted_item, identification: {
+                                           sourceId: "sul:#{SecureRandom.uuid}",
+                                           barcode:
+                                         })
+      end
+
+      before do
+        item.identification.barcode # ensure item is created before searching
+      end
+
+      it 'matches query with bare and prefixed barcode' do
+        [barcode, "barcode:#{barcode}"].each do |query|
+          fill_in 'q', with: query
+          click_button 'search'
+          expect(page).to have_content('1 entry found')
+          expect(page).to have_css('dd.blacklight-id', text: solr_id)
+        end
+      end
+    end
+
+    context 'for ILS (folio) identifiers' do
+      let(:catalog_id) { 'a11403803' }
+      let(:item) do
+        FactoryBot.create_for_repository(:persisted_item, identification: {
+                                           sourceId: "sul:#{SecureRandom.uuid}",
+                                           catalogLinks: [{
+                                             catalog: 'folio',
+                                             refresh: false,
+                                             catalogRecordId: catalog_id
+                                           }]
+                                         })
+      end
+
+      before do
+        item.identification.catalogLinks # ensure item is created before searching
+      end
+
+      it 'matches catalog identifier with and without folio prefix' do
+        [catalog_id, "folio:#{catalog_id}"].each do |query|
+          fill_in 'q', with: query
+          click_button 'search'
+          expect(page).to have_content('1 entry found')
+          expect(page).to have_css('dd.blacklight-id', text: solr_id)
+        end
+      end
+    end
+
+    context 'for DOIs' do
+      # is there a reason to tokenize DOIs?
+
+      it 'matches bare and "doi:" prefixed DOIs' do
+        skip('write this test')
+      end
+    end
+  end
+end

spec/support/item_method_sender.rb

@@ -1,20 +1,35 @@
 # frozen_string_literal: true
 
+# Used in conjunction with factories/items.rb to create a Cocina::Models::RequestDRO,
+#  allowing for customization of fields/values in the immutable cocina-model.
 class ItemMethodSender
   def initialize(cocina_request)
     @cocina_model = cocina_request
   end
 
   attr_reader :cocina_model
 
+  # @example
+  #   item = create(:persisted_item, title: 'simple title value')
   def title=(title)
     @cocina_model = @cocina_model.new(description: { title: [{ value: title }] })
   end
 
+  # @param [Array<Hash>] title_values - an array of arbitrarily complex cocina title values
+  # @example
+  #   item = create(:persisted_item, title_values: [{ value: 'simple title value' }, { value: 'another title' }])
+  def title_values=(title_values)
+    @cocina_model = @cocina_model.new(description: { title: title_values })
+  end
+
+  # @example
+  #   item = create(:persisted_item, source_id: 'sul:M932_1623_B1_F1_001')
   def source_id=(source_id)
     @cocina_model = @cocina_model.new(identification: @cocina_model.identification.new(sourceId: source_id))
   end
 
+  # @example
+  #   item = create(:persisted_item,  collection_id: 'druid:rt923rd3423')
   def collection_id=(collection_id)
     @cocina_model = @cocina_model.new(structural: Cocina::Models::DROStructural.new(isMemberOf: [collection_id]))
   end