Add rake task to write duplicate locations

lib/tasks/count_duplicates.rake

@@ -11,13 +11,38 @@ namespace :count_duplicates do
       end
     end
 
-    BYTE_ORDER_MARK = "\uFEFF".freeze
-    EXPIRATION_TIME = 72.hours.to_i
     filename = "scheme-duplicates-#{Time.zone.now}.csv"
     storage_service = Storage::S3Service.new(Configuration::EnvConfigurationService.new, ENV["BULK_UPLOAD_BUCKET"])
-    storage_service.write_file(filename, BYTE_ORDER_MARK + duplicates_csv)
+    storage_service.write_file(filename, "#{duplicates_csv}")
 
-    url = storage_service.get_presigned_url(filename, EXPIRATION_TIME)
+    url = storage_service.get_presigned_url(filename, 72.hours.to_i)
+    Rails.logger.info("Download URL: #{url}")
+  end
+
+  desc "Count the number of duplicate locations per organisation"
+  task location_duplicates_per_org: :environment do
+    duplicates_csv = CSV.generate(headers: true) do |csv|
+      csv << ["Organisation id", "Number of duplicate sets", "Total duplicate locations"]
+
+      Organisation.visible.each do |organisation|
+        duplicate_sets_count = 0
+        total_duplicate_locations = 0
+        organisation.owned_schemes.each do |scheme|
+          duplicate_sets_count += scheme.locations.duplicate_sets.count
+          total_duplicate_locations += scheme.locations.duplicate_sets.sum(&:size)
+        end
+
+        if duplicate_sets_count.positive?
+          csv << [organisation.id, duplicate_sets_count, total_duplicate_locations]
+        end
+      end
+    end
+
+    filename = "location-duplicates-#{Time.zone.now}.csv"
+    storage_service = Storage::S3Service.new(Configuration::EnvConfigurationService.new, ENV["BULK_UPLOAD_BUCKET"])
+    storage_service.write_file(filename, "#{duplicates_csv}")
+
+    url = storage_service.get_presigned_url(filename, 72.hours.to_i)
     Rails.logger.info("Download URL: #{url}")
   end
 end

spec/lib/tasks/count_duplicates_spec.rb

@@ -46,4 +46,53 @@ RSpec.describe "count_duplicates" do
       end
     end
   end
+
+  describe "count_duplicates:location_duplicates_per_org", type: :task do
+    subject(:task) { Rake::Task["count_duplicates:location_duplicates_per_org"] }
+
+    let(:storage_service) { instance_double(Storage::S3Service) }
+    let(:test_url) { "test_url" }
+
+    before do
+      Rake.application.rake_require("tasks/count_duplicates")
+      Rake::Task.define_task(:environment)
+      task.reenable
+      allow(Storage::S3Service).to receive(:new).and_return(storage_service)
+      allow(storage_service).to receive(:write_file)
+      allow(storage_service).to receive(:get_presigned_url).and_return(test_url)
+    end
+
+    context "when the rake task is run" do
+      context "and there are no duplicate locations" do
+        let!(:organisation) { create(:organisation) }
+
+        it "creates a csv with headers only" do
+          expect(storage_service).to receive(:write_file).with(/location-duplicates-.*\.csv/, "\uFEFFOrganisation id,Number of duplicate sets,Total duplicate locations\n")
+          expect(Rails.logger).to receive(:info).with("Download URL: #{test_url}")
+          task.invoke
+        end
+      end
+
+      context "and there are duplicate locations" do
+        let(:organisation) { create(:organisation) }
+        let(:scheme) { create(:scheme, owning_organisation: organisation) }
+        let(:organisation2) { create(:organisation) }
+        let(:scheme2) { create(:scheme, owning_organisation: organisation2) }
+        let(:scheme3) { create(:scheme, owning_organisation: organisation2) }
+
+        before do
+          create_list(:location, 2, postcode: "A1 1AB", mobility_type: "M", scheme:)
+          create_list(:location, 3, postcode: "A1 1AB", mobility_type: "A", scheme:)
+          create_list(:location, 5, postcode: "A1 1AB", mobility_type: "M", scheme: scheme2)
+          create_list(:location, 2, postcode: "A1 1AB", mobility_type: "M", scheme: scheme3)
+        end
+
+        it "creates a csv with correct duplicate numbers" do
+          expect(storage_service).to receive(:write_file).with(/location-duplicates-.*\.csv/, "\uFEFFOrganisation id,Number of duplicate sets,Total duplicate locations\n#{organisation.id},2,5\n#{organisation2.id},2,7\n")
+          expect(Rails.logger).to receive(:info).with("Download URL: #{test_url}")
+          task.invoke
+        end
+      end
+    end
+  end
 end