feat: add rake task to clean unused data to improve performance

Currently, it deletes more verifications than expected, but if you are building the provider often, and you are not using can-i-deploy, this should be ok.

pact-foundation#203

Author: bethesque

lib/pact_broker/db/clean.rb

@@ -0,0 +1,49 @@
+require 'sequel'
+require 'pact_broker/project_root'
+
+module PactBroker
+  module DB
+    class Clean
+      def self.call database_connection, options = {}
+        new(database_connection, options).call
+      end
+
+      def initialize database_connection, options = {}
+        @db = database_connection
+        @options = options
+      end
+
+      def call
+        db[:verifications].where(id: db[:materialized_head_matrix].select(:verification_id)).invert.delete
+        pp_ids = db[:materialized_head_matrix].select(:pact_publication_id)
+
+        triggered_webhook_ids = db[:triggered_webhooks].where(pact_publication_id: pp_ids).invert.select(:id)
+        db[:webhook_executions].where(triggered_webhook_id: triggered_webhook_ids).delete
+        db[:triggered_webhooks].where(id: triggered_webhook_ids).delete
+        db[:webhook_executions].where(pact_publication_id: pp_ids).invert.delete
+
+        db[:pact_publications].where(id: pp_ids).invert.delete
+
+        referenced_pact_version_ids = db[:pact_publications].select(:pact_version_id).collect{ | h| h[:pact_version_id] } +
+          db[:verifications].select(:pact_version_id).collect{ | h| h[:pact_version_id] }
+        db[:pact_versions].where(id: referenced_pact_version_ids).invert.delete
+
+        referenced_version_ids = db[:pact_publications].select(:consumer_version_id).collect{ | h| h[:consumer_version_id] } +
+          db[:verifications].select(:provider_version_id).collect{ | h| h[:provider_version_id] }
+
+        db[:tags].where(version_id: referenced_version_ids).invert.delete
+        db[:versions].where(id: referenced_version_ids).invert.delete
+
+        db[:materialized_matrix].delete
+        db[:materialized_matrix].insert(db[:matrix].select_all)
+        db[:materialized_head_matrix].delete
+        db[:materialized_head_matrix].insert(db[:head_matrix].select_all)
+      end
+
+      private
+
+      attr_reader :db
+
+    end
+  end
+end

lib/pact_broker/tasks.rb

@@ -1,2 +1,3 @@
 require 'pact_broker/tasks/migration_task'
 require 'pact_broker/tasks/version_task'
+require 'pact_broker/tasks/clean_task'

lib/pact_broker/tasks/clean_task.rb

@@ -0,0 +1,25 @@
+module PactBroker
+  module DB
+    class CleanTask < ::Rake::TaskLib
+
+      attr_accessor :database_connection
+
+      def initialize &block
+        rake_task &block
+      end
+
+      def rake_task &block
+        namespace :pact_broker do
+          namespace :db do
+            desc "Clean unused pacts and verifications from database"
+            task :clean do | t, args |
+              require 'pact_broker/db/clean'
+              instance_eval(&block)
+              PactBroker::DB::Clean.call(database_connection)
+            end
+          end
+        end
+      end
+    end
+  end
+end

spec/lib/pact_broker/db/clean_spec.rb

@@ -0,0 +1,62 @@
+require 'pact_broker/db/clean'
+
+module PactBroker
+  module DB
+    describe Clean do
+      describe ".call" do
+
+        let(:td) { TestDataBuilder.new }
+        before do
+          td.create_pact_with_hierarchy("Foo", "0", "Bar")
+            .create_consumer_version_tag("prod")
+            .create_consumer_version("1")
+            .create_pact
+            .create_consumer_version_tag("prod")
+            .comment("keep")
+            .create_verification(provider_version: "20")
+            .create_consumer_version("2")
+            .create_pact
+            .comment("don't keep")
+            .create_webhook
+            .create_triggered_webhook
+            .create_webhook_execution
+            .create_deprecated_webhook_execution
+            .create_verification(provider_version: "30")
+            .create_consumer_version("3")
+            .create_pact
+            .comment("keep")
+            .create_verification(provider_version: "40")
+            .create_verification(provider_version: "50", number: 2)
+        end
+
+        subject { Clean.call(PactBroker::DB.connection) }
+        let(:db) { PactBroker::DB.connection }
+
+        it "does not delete any rows in the head matrix" do
+          head_matrix_before = db[:head_matrix].select_all
+          subject
+          head_matrix_after = db[:head_matrix].select_all
+          expect(head_matrix_before).to eq head_matrix_after
+        end
+
+        it "deletes rows that aren't the latest or latest tagged" do
+          subject
+          expect(db[:materialized_matrix].where(consumer_version_number: "2").count).to eq 0
+        end
+
+        it "deletes orphan pact_versions" do
+          subject
+          expect(db[:pact_versions].count).to eq 2
+        end
+
+        it "deletes orphan versions" do
+          subject
+          expect(db[:versions].where(number: "20").count).to be 1
+          expect(db[:versions].where(number: "30").count).to be 0
+          expect(db[:versions].where(number: "40").count).to be 0
+          expect(db[:versions].where(number: "50").count).to be 1
+        end
+      end
+    end
+  end
+end