From 012f2630894da56d0efdbd250c8f5fedbedb4992 Mon Sep 17 00:00:00 2001
From: Kat <katrina@kosiak.co.uk>
Date: Tue, 17 Sep 2024 12:09:02 +0100
Subject: [PATCH] Add rake task to write duplicate scheme sets

---
 lib/tasks/count_duplicates.rake         | 23 ++++++++++++
 spec/lib/tasks/count_duplicates_spec.rb | 49 +++++++++++++++++++++++++
 2 files changed, 72 insertions(+)
 create mode 100644 lib/tasks/count_duplicates.rake
 create mode 100644 spec/lib/tasks/count_duplicates_spec.rb

diff --git a/lib/tasks/count_duplicates.rake b/lib/tasks/count_duplicates.rake
new file mode 100644
index 000000000..1ce82975a
--- /dev/null
+++ b/lib/tasks/count_duplicates.rake
@@ -0,0 +1,23 @@
+namespace :count_duplicates do
+  desc "Count the number of duplicate schemes per organisation"
+  task scheme_duplicates_per_org: :environment do
+    duplicates_csv = CSV.generate(headers: true) do |csv|
+      csv << ["Organisation id", "Number of duplicate sets", "Total duplicate schemes"]
+
+      Organisation.visible.each do |organisation|
+        if organisation.owned_schemes.duplicate_sets.count.positive?
+          csv << [organisation.id, organisation.owned_schemes.duplicate_sets.count, organisation.owned_schemes.duplicate_sets.sum(&:size)]
+        end
+      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)
+
+    url = storage_service.get_presigned_url(filename, EXPIRATION_TIME)
+    Rails.logger.info("Download URL: #{url}")
+  end
+end
diff --git a/spec/lib/tasks/count_duplicates_spec.rb b/spec/lib/tasks/count_duplicates_spec.rb
new file mode 100644
index 000000000..b939c05e4
--- /dev/null
+++ b/spec/lib/tasks/count_duplicates_spec.rb
@@ -0,0 +1,49 @@
+require "rails_helper"
+require "rake"
+
+RSpec.describe "count_duplicates" do
+  describe "count_duplicates:scheme_duplicates_per_org", type: :task do
+    subject(:task) { Rake::Task["count_duplicates:scheme_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 schemes" do
+        let!(:organisation) { create(:organisation) }
+
+        it "creates a csv with headers only" do
+          expect(storage_service).to receive(:write_file).with(/scheme-duplicates-.*\.csv/, "\uFEFFOrganisation id,Number of duplicate sets,Total duplicate schemes\n")
+          expect(Rails.logger).to receive(:info).with("Download URL: #{test_url}")
+          task.invoke
+        end
+      end
+
+      context "and there are duplicate schemes" do
+        let(:organisation) { create(:organisation) }
+        let(:organisation2) { create(:organisation) }
+
+        before do
+          create_list(:scheme, 2, :duplicate, owning_organisation: organisation)
+          create_list(:scheme, 3, :duplicate, primary_client_group: "I", owning_organisation: organisation)
+          create_list(:scheme, 5, :duplicate, owning_organisation: organisation2)
+        end
+
+        it "creates a csv with correct duplicate numbers" do
+          expect(storage_service).to receive(:write_file).with(/scheme-duplicates-.*\.csv/, "\uFEFFOrganisation id,Number of duplicate sets,Total duplicate schemes\n#{organisation.id},2,5\n#{organisation2.id},1,5\n")
+          expect(Rails.logger).to receive(:info).with("Download URL: #{test_url}")
+          task.invoke
+        end
+      end
+    end
+  end
+end