From b4f995b048905567dc4a7da0e22341e56484437b Mon Sep 17 00:00:00 2001
From: Naomi Dushay <ndushay@stanford.edu>
Date: Wed, 13 Mar 2024 11:40:08 -0700
Subject: [PATCH 1/2] indexing_identifiers_spec less noisy

---
 spec/features/indexing_identifiers_spec.rb | 213 ++++++++++-----------
 1 file changed, 100 insertions(+), 113 deletions(-)

diff --git a/spec/features/indexing_identifiers_spec.rb b/spec/features/indexing_identifiers_spec.rb
index 6a2d27240..c3a0aa6cd 100644
--- a/spec/features/indexing_identifiers_spec.rb
+++ b/spec/features/indexing_identifiers_spec.rb
@@ -16,6 +16,7 @@
     sign_in create(:user), groups: ['sdr:administrator-role']
     solr_conn.commit # ensure no deletes are pending
     visit '/'
+    item.identification # ensure item is created before searching
   end
 
   after do
@@ -23,148 +24,134 @@
     solr_conn.commit
   end
 
-  describe 'identifier searching' do
-    context 'for druids' do
-      let(:prefixed_druid) { item.externalIdentifier }
+  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
+    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:) }
+  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:) }
+
+    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
 
-      before do
-        item.identification.sourceId # ensure item is created before searching
-      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 whole string, including prefix before first colon' do
-        fill_in 'q', with: source_id
+    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 '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 '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
 
-      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
+    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 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
+    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
 
-      it 'is not case sensitive' do
-        fill_in 'q', with: 'm2549 STERTZER'
+    it 'matches query with bare and prefixed barcode' do
+      [barcode, "barcode:#{barcode}"].each do |query|
+        fill_in 'q', with: query
         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_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
+  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
+  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
 
-    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
+    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?
+  context 'for DOIs' do
+    # is there a reason to tokenize DOIs?
 
-      it 'matches bare and "doi:" prefixed DOIs' do
-        skip('write this test')
-      end
+    it 'matches bare and "doi:" prefixed DOIs' do
+      skip('write this test')
     end
   end
 end

From dc2842b3bd82a2275ff61f0d8ba3fea624015ecd Mon Sep 17 00:00:00 2001
From: Naomi Dushay <ndushay@stanford.edu>
Date: Wed, 13 Mar 2024 11:44:12 -0700
Subject: [PATCH 2/2] indexing_tags_spec added

---
 spec/features/indexing_tags_spec.rb | 120 ++++++++++++++++++++++++++++
 1 file changed, 120 insertions(+)
 create mode 100644 spec/features/indexing_tags_spec.rb

diff --git a/spec/features/indexing_tags_spec.rb b/spec/features/indexing_tags_spec.rb
new file mode 100644
index 000000000..3dc9e4ecf
--- /dev/null
+++ b/spec/features/indexing_tags_spec.rb
@@ -0,0 +1,120 @@
+# 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.
+#
+# tag tests need javascript for facet testing because they are so slow to load in production
+#   javascript possibly also needed for the edit tags modal
+#
+# rubocop:disable Capybara/ClickLinkOrButtonStyle
+RSpec.describe 'Indexing and search results for tags', :js do
+  let(:item) { FactoryBot.create_for_repository(:persisted_item) }
+  let(:solr_id) { item.externalIdentifier }
+  let(:project_tag) { 'Project : ARS 78s : broken' }
+  let(:non_project_tag) { 'willet : murder of crows : curlew' }
+  let(:project_tag2) { 'Project : jira-517' }
+  let(:blacklight_config) { CatalogController.blacklight_config }
+  let(:solr_conn) { blacklight_config.repository_class.new(blacklight_config).connection }
+
+  # I wanted to do this as a before(:context) but after much teeth gnashing, I gave up
+  #   That would have facilitated a single setup done once that each test could utilize.
+  before do
+    sign_in create(:user), groups: ['sdr:administrator-role']
+    solr_conn.commit # ensure no deletes are pending
+    visit solr_document_path(item.externalIdentifier)
+    find("a[aria-label='Edit tags']").click
+    within('#edit-modal') do
+      click_button '+ Add another tag'
+      fill_in currently_with: '', with: project_tag
+      click_button '+ Add another tag'
+      fill_in currently_with: '', with: non_project_tag
+      click_button '+ Add another tag'
+      fill_in currently_with: '', with: project_tag2
+      click_button 'Save'
+    end
+    click_link_or_button 'Reindex'
+    # wait for indexing
+    expect(page).to have_text('Successfully updated index') # rubocop:disable RSpec/ExpectInHook
+    visit '/'
+    item.description # ensure item is created before searching
+  end
+
+  after do
+    solr_conn.delete_by_id(solr_id)
+    solr_conn.commit
+  end
+
+  # one giant it block to reduce the time to run the tests; this is because I
+  #   fussed with before(:context) for some hours and then gave up on it.
+  it 'searches and facets behave as expected' do
+    # ------- search behavior --------
+
+    # project tags values include "Project" in searchable value
+    fill_in 'q', with: 'Project'
+    click_button 'search'
+    expect(page).to have_content('1 entry found')
+    expect(page).to have_css('dd.blacklight-id', text: solr_id)
+
+    # project tags are searchable, tokenized
+    ['ARS', '78s', 'ARS 78s', 'broken', '"78s broken"'].each do |token|
+      fill_in 'q', with: token
+      click_button 'search'
+      expect(page).to have_content('1 entry found')
+      expect(page).to have_css('dd.blacklight-id', text: solr_id)
+    end
+
+    # non-project tags tags are searchable, tokenized
+    ['willet', 'murder of crows', 'murder', 'of', 'crows', 'curlew', '"crows curlew"'].each do |token|
+      fill_in 'q', with: token
+      click_button 'search'
+      expect(page).to have_content('1 entry found')
+      expect(page).to have_css('dd.blacklight-id', text: solr_id)
+    end
+
+    # Some Argo accessioneers have developed systems to work around problematic
+    #   searching existing in Argo for years.  In some cases there are giant spreadsheets
+    #   with links that depend on old things working.
+    #
+    # project tags with spaces around the colon work (searching)' do
+    # Tag of “Project : jira-517” should match search term of “jira-517”
+    # We do not need “jira” or “517” to match, but … it can?
+    ['jira-517', 'jira', '517', '"jira 517"'].each do |token|
+      fill_in 'q', with: token
+      click_button 'search'
+      expect(page).to have_content('1 entry found')
+      expect(page).to have_css('dd.blacklight-id', text: solr_id)
+    end
+
+    # ------- facet behavior --------
+
+    # project tags are a hierarchical facet
+    fill_in 'q', with: solr_id
+    click_button 'search'
+    click_link_or_button 'Project'
+    # ensure facet has been expanded by javascript
+    expect(page).to have_css('#facet-exploded_project_tag_ssim')
+    # Note that "Project" is not indexed as part of facet
+    click_link_or_button 'ARS 78s'
+    click_link_or_button 'broken'
+    expect(page).to have_content('1 entry found')
+    expect(page).to have_css('dd.blacklight-id', text: solr_id)
+
+    # non-project tags are a hierarchical facet
+    fill_in 'q', with: solr_id
+    click_button 'search'
+    click_link_or_button 'Tag'
+    # ensure facet has been expanded by javascript
+    expect(page).to have_css('#facet-exploded_nonproject_tag_ssim')
+    click_link_or_button 'willet'
+    skip 'FIXME: is this failing on spaces in nonproject tag values?'
+    click_link_or_button 'murder of crows'
+    click_link_or_button 'curlew'
+    expect(page).to have_content('1 entry found')
+    expect(page).to have_css('dd.blacklight-id', text: solr_id)
+  end
+end
+# rubocop:enable Capybara/ClickLinkOrButtonStyle