CLDC-2713 Add a rake task to reimport addresses from csv (#1890)

* Add a rake task to reimport addresses from csv

* Add empty line

* Update postcode known

* Update values updated at column

lib/tasks/import_address_from_csv.rake

@@ -0,0 +1,42 @@
+namespace :data_import do
+  desc "Import address data from a csv file"
+  task :import_address_from_csv, %i[file_name] => :environment do |_task, args|
+    file_name = args[:file_name]
+
+    raise "Usage: rake data_import:import_address_from_csv['csv_file_name']" if file_name.blank?
+
+    s3_service = Storage::S3Service.new(PlatformHelper.is_paas? ? Configuration::PaasConfigurationService.new : Configuration::EnvConfigurationService.new, ENV["IMPORT_PAAS_INSTANCE"])
+    addresses_csv = CSV.parse(s3_service.get_file_io(file_name), headers: true)
+
+    addresses_csv.each do |row|
+      lettings_log_id = row[0]
+
+      if lettings_log_id.blank?
+        Rails.logger.info("Lettings log ID not provided for address: #{[row[1], row[2], row[3], row[4]].join(', ')}")
+        next
+      end
+
+      lettings_log = LettingsLog.find_by(id: lettings_log_id)
+      if lettings_log.blank?
+        Rails.logger.info("Could not find a lettings log with id #{lettings_log_id}")
+        next
+      end
+
+      lettings_log.uprn_known = 0
+      lettings_log.uprn = nil
+      lettings_log.uprn_confirmed = nil
+      lettings_log.address_line1 = row[1]
+      lettings_log.address_line2 = row[2]
+      lettings_log.town_or_city = row[3]
+      lettings_log.postcode_full = row[4]
+      lettings_log.postcode_known = lettings_log.postcode_full.present? ? 1 : nil
+      lettings_log.county = nil
+      lettings_log.is_la_inferred = nil
+      lettings_log.la = nil
+      lettings_log.values_updated_at = Time.zone.now
+
+      lettings_log.save!
+      Rails.logger.info("Updated lettings log #{lettings_log_id}, with address: #{[lettings_log.address_line1, lettings_log.address_line2, lettings_log.town_or_city, lettings_log.postcode_full].join(', ')}")
+    end
+  end
+end

spec/fixtures/files/addresses_reimport.csv

@@ -0,0 +1,6 @@
+lettings_log_id,address_line1,address_line2,town_or_city,postcode_full
+{id},address 1,address 2,town,B1 1BB
+{id2},address 3,address 4,city,B1 1BB
+{id3},,,,C1 1CC
+,address 5,address 6,city,D1 1DD
+fake_id,address 7,address 8,city,D1 1DD

spec/lib/tasks/correct_address_from_csv_spec.rb

@@ -0,0 +1,154 @@
+require "rails_helper"
+require "rake"
+
+RSpec.describe "data_import" do
+  def replace_entity_ids(lettings_log, second_lettings_log, third_lettings_log, export_template)
+    export_template.sub!(/\{id\}/, lettings_log.id.to_s)
+    export_template.sub!(/\{id2\}/, second_lettings_log.id.to_s)
+    export_template.sub!(/\{id3\}/, third_lettings_log.id.to_s)
+  end
+
+  describe ":import_address_from_csv", type: :task do
+    subject(:task) { Rake::Task["data_import:import_address_from_csv"] }
+
+    let(:instance_name) { "paas_import_instance" }
+    let(:storage_service) { instance_double(Storage::S3Service) }
+    let(:env_config_service) { instance_double(Configuration::EnvConfigurationService) }
+    let(:paas_config_service) { instance_double(Configuration::PaasConfigurationService) }
+
+    before do
+      allow(Storage::S3Service).to receive(:new).and_return(storage_service)
+      allow(Configuration::EnvConfigurationService).to receive(:new).and_return(env_config_service)
+      allow(Configuration::PaasConfigurationService).to receive(:new).and_return(paas_config_service)
+      allow(ENV).to receive(:[])
+      allow(ENV).to receive(:[]).with("IMPORT_PAAS_INSTANCE").and_return(instance_name)
+      allow(ENV).to receive(:[]).with("VCAP_SERVICES").and_return("dummy")
+
+      Rake.application.rake_require("tasks/import_address_from_csv")
+      Rake::Task.define_task(:environment)
+      task.reenable
+
+      WebMock.stub_request(:get, /api.postcodes.io/)
+        .to_return(status: 200, body: "{\"status\":404,\"error\":\"Postcode not found\"}", headers: {})
+      WebMock.stub_request(:get, /api.postcodes.io\/postcodes\/B11BB/)
+        .to_return(status: 200, body: '{"status":200,"result":{"admin_district":"Westminster","codes":{"admin_district":"E08000035"}}}', headers: {})
+    end
+
+    context "when the rake task is run" do
+      let(:addresses_csv_path) { "addresses_reimport_123.csv" }
+      let(:wrong_file_path) { "/test/no_csv_here.csv" }
+      let!(:lettings_log) do
+        create(:lettings_log,
+               uprn_known: nil,
+               uprn: nil,
+               uprn_confirmed: nil,
+               address_line1: nil,
+               address_line2: nil,
+               town_or_city: nil,
+               county: nil,
+               postcode_known: 1,
+               postcode_full: "A1 1AA",
+               la: "E06000064",
+               is_la_inferred: true)
+      end
+
+      let!(:second_lettings_log) do
+        create(:lettings_log,
+               uprn_known: 1,
+               uprn: "1",
+               uprn_confirmed: nil,
+               address_line1: "wrong address line1",
+               address_line2: "wrong address 2",
+               town_or_city: "wrong town",
+               county: "wrong city",
+               postcode_known: 1,
+               postcode_full: "A1 1AA",
+               la: "E06000064",
+               is_la_inferred: true)
+      end
+
+      let!(:third_lettings_log) do
+        create(:lettings_log,
+               uprn_known: 1,
+               uprn: "1",
+               uprn_confirmed: nil,
+               address_line1: "wrong address line1",
+               address_line2: "wrong address 2",
+               town_or_city: "wrong town",
+               county: "wrong city",
+               postcode_known: 1,
+               postcode_full: "A1 1AA",
+               la: "E06000064",
+               is_la_inferred: true)
+      end
+
+      before do
+        allow(storage_service).to receive(:get_file_io)
+        .with("addresses_reimport_123.csv")
+        .and_return(replace_entity_ids(lettings_log, second_lettings_log, third_lettings_log, File.open("./spec/fixtures/files/addresses_reimport.csv").read))
+      end
+
+      it "updates the log address when old address was not given" do
+        task.invoke(addresses_csv_path)
+        lettings_log.reload
+        expect(lettings_log.uprn_known).to eq(0)
+        expect(lettings_log.uprn).to eq(nil)
+        expect(lettings_log.uprn_confirmed).to eq(nil)
+        expect(lettings_log.address_line1).to eq("address 1")
+        expect(lettings_log.address_line2).to eq("address 2")
+        expect(lettings_log.town_or_city).to eq("town")
+        expect(lettings_log.county).to eq(nil)
+        expect(lettings_log.postcode_known).to eq(1)
+        expect(lettings_log.postcode_full).to eq("B1 1BB")
+        expect(lettings_log.la).to eq("E08000035")
+        expect(lettings_log.is_la_inferred).to eq(true)
+      end
+
+      it "updates the log address when old address was given" do
+        task.invoke(addresses_csv_path)
+        second_lettings_log.reload
+        expect(second_lettings_log.uprn_known).to eq(0)
+        expect(second_lettings_log.uprn).to eq(nil)
+        expect(second_lettings_log.uprn_confirmed).to eq(nil)
+        expect(second_lettings_log.address_line1).to eq("address 3")
+        expect(second_lettings_log.address_line2).to eq("address 4")
+        expect(second_lettings_log.town_or_city).to eq("city")
+        expect(second_lettings_log.county).to eq(nil)
+        expect(second_lettings_log.postcode_known).to eq(1)
+        expect(second_lettings_log.postcode_full).to eq("B1 1BB")
+        expect(second_lettings_log.la).to eq("E08000035")
+        expect(second_lettings_log.is_la_inferred).to eq(true)
+      end
+
+      it "updates the log address when new address is not given" do
+        task.invoke(addresses_csv_path)
+        third_lettings_log.reload
+        expect(third_lettings_log.uprn_known).to eq(0)
+        expect(third_lettings_log.uprn).to eq(nil)
+        expect(third_lettings_log.uprn_confirmed).to eq(nil)
+        expect(third_lettings_log.address_line1).to eq(nil)
+        expect(third_lettings_log.address_line2).to eq(nil)
+        expect(third_lettings_log.town_or_city).to eq(nil)
+        expect(third_lettings_log.county).to eq(nil)
+        expect(third_lettings_log.postcode_known).to eq(1)
+        expect(third_lettings_log.postcode_full).to eq("C11CC")
+        expect(third_lettings_log.la).to eq(nil)
+        expect(third_lettings_log.is_la_inferred).to eq(false)
+      end
+
+      it "logs the progress of the update" do
+        expect(Rails.logger).to receive(:info).with("Updated lettings log #{lettings_log.id}, with address: address 1, address 2, town, B1 1BB")
+        expect(Rails.logger).to receive(:info).with("Updated lettings log #{second_lettings_log.id}, with address: address 3, address 4, city, B1 1BB")
+        expect(Rails.logger).to receive(:info).with("Updated lettings log #{third_lettings_log.id}, with address: , , , C11CC")
+        expect(Rails.logger).to receive(:info).with("Lettings log ID not provided for address: address 5, address 6, city, D1 1DD")
+        expect(Rails.logger).to receive(:info).with("Could not find a lettings log with id fake_id")
+
+        task.invoke(addresses_csv_path)
+      end
+
+      it "raises an error when no path is given" do
+        expect { task.invoke(nil) }.to raise_error(RuntimeError, "Usage: rake data_import:import_address_from_csv['csv_file_name']")
+      end
+    end
+  end
+end