Only Group Duplicates when new publication during AI import (#583)

* Tweaks to ai importer to only group duplicates when a new publication is created and tweaked the duplicate grouping task to toggle publications to not be visible if they have not already been grouped. * Niftany
psu-libraries · Oct 24, 2022 · a0e868f · a0e868f
1 parent 1eff6a6
commit a0e868f
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 8 deletions.
diff --git a/app/importers/activity_insight_importer.rb b/app/importers/activity_insight_importer.rb
@@ -167,6 +167,11 @@ def call
             update_pub_record(pub_record, pub)
           else
             pi.save!
+
+            DuplicatePublicationGroup.group_duplicates_of(pub_record)
+            if pub_record.reload.duplicate_group
+              pub_record.update!(visible: false)
+            end
           end
 
           if pub_record.updated_by_user_at.blank?
@@ -195,12 +200,6 @@ def call
               c.save!
             end
           end
-
-          DuplicatePublicationGroup.group_duplicates_of(pub_record)
-
-          if pub_record.reload.duplicate_group
-            pub_record.update!(visible: false)
-          end
         end
       rescue StandardError => e
         log_error(pub, e, u)

diff --git a/app/models/duplicate_publication_group.rb b/app/models/duplicate_publication_group.rb
@@ -8,8 +8,13 @@ def self.group_duplicates
                                  total: Publication.count)
 
     Publication.find_each do |p|
+      already_grouped = p.duplicate_group.present?
       group_duplicates_of(p)
 
+      group = p.reload.duplicate_group
+      if !already_grouped && group
+        p.update!(visible: false)
+      end
       pbar.increment
     end
     pbar.finish

diff --git a/spec/component/importers/activity_insight_importer_spec.rb b/spec/component/importers/activity_insight_importer_spec.rb
@@ -1179,7 +1179,7 @@
           end
 
           it 'groups duplicates of new publication records' do
-            expect { importer.call }.to change(DuplicatePublicationGroup, :count).by 2
+            expect { importer.call }.to change(DuplicatePublicationGroup, :count).by 1
 
             p1 = PublicationImport.find_by(source: 'Activity Insight',
                                            source_identifier: '190706413568').publication
@@ -1465,7 +1465,7 @@
           end
 
           it 'groups duplicates of new publication records' do
-            expect { importer.call }.to change(DuplicatePublicationGroup, :count).by 2
+            expect { importer.call }.to change(DuplicatePublicationGroup, :count).by 1
 
             p1 = PublicationImport.find_by(source: 'Activity Insight',
                                            source_identifier: '190706413568').publication

diff --git a/spec/component/models/duplicate_publication_group_spec.rb b/spec/component/models/duplicate_publication_group_spec.rb
@@ -400,6 +400,40 @@
         expect(p15_1.reload.duplicate_group.publications).to match_array [p15_1, p15_2]
         expect(p16_1.reload.duplicate_group.publications).to match_array [p16_1, p16_2]
       end
+
+      it "sets grouped publications' visible statuses to false when publication was not already grouped before process started" do
+        described_class.group_duplicates
+
+        expect(p1_1.reload.duplicate_group.publications.map(&:visible)).to match_array [true, false, false, false]
+        expect(p2_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p3_1.reload.visible).to be true
+        expect(p3_2.reload.visible).to be true
+        expect(p4_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p5_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p6_1.reload.visible).to be true
+        expect(p6_2.reload.visible).to be true
+        expect(p7_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p8_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p9_1.reload.visible).to be true
+        expect(p9_2.reload.visible).to be true
+        expect(p10_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p11_1.reload.visible).to be true
+        expect(p11_2.reload.visible).to be true
+        expect(p11_3.reload.visible).to be true
+        expect(p12_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false]
+        expect(p12_2.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false]
+        expect(p12_3.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false]
+        expect(p13_1.reload.visible).to be true
+        expect(p13_2.reload.visible).to be true
+        expect(p14_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false, false]
+        expect(p14_2.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false, false]
+        expect(p14_3.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false, false]
+        expect(p14_4.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false, false, false]
+        expect(p15_1.reload.duplicate_group.publications.map(&:visible)).to match_array [false, false]
+        expect(p16_1.reload.duplicate_group.publications.map(&:visible)).to match_array [true, false]
+        expect(p17_1.reload.visible).to be true
+        expect(p17_2.reload.visible).to be true
+      end
     end
   end