Cleanup the method to reduce AbcSize

3466b832 · Tetiana Chupryna · bbf692ba · 3466b832 · 3466b832
--- a/ee/lib/gitlab/license_scanning/package_licenses.rb
+++ b/ee/lib/gitlab/license_scanning/package_licenses.rb
@@ -10,72 +10,19 @@ def initialize(components:)
        @components = components
      end

-      # rubocop: disable Metrics/AbcSize
      def fetch
        use_replica_if_available do
-          first_query_search_fields = [
-            ::PackageMetadata::PackageVersion.arel_table[:pm_package_id],
-            ::PackageMetadata::PackageVersion.arel_table[:version]
-          ]
-
-          search_fields = [
-            ::PackageMetadata::Package.arel_table[:purl_type],
-            ::PackageMetadata::Package.arel_table[:name],
-            ::PackageMetadata::PackageVersion.arel_table[:version]
-          ]
-
          records_with_licenses = {}

          components.each_slice(BATCH_SIZE).each do |_components_batch|
-            # rubocop: disable CodeReuse/ActiveRecord
-
-            component_tuples = components.map do |component|
-              purl_type_int = ::Enums::Sbom.purl_types[component.purl_type]
-              [purl_type_int, component.name, component.version]
-            end
-
-            component_values_list = Arel::Nodes::Grouping.new(Arel::Nodes::ValuesList.new(component_tuples))
-
-            pm_packages_table = ::PackageMetadata::Package.arel_table
-
-            needed_package_versions_table = Arel::Table.new(:needed_package_versions)
-
-            component_values = Arel::Nodes::As.new(
-              Hashie::Mash.new(
-                name: Arel.sql('needed_package_versions(purl_type, name, version)')), component_values_list
-            )
-
-            query = needed_package_versions_table
-                      .project(pm_packages_table[:id], needed_package_versions_table[:version])
-                      .with(component_values)
-                      .distinct
-                      .join(pm_packages_table)
-                      .on(
-                        pm_packages_table[:purl_type].eq(needed_package_versions_table[:purl_type])
-                          .and(
-                            pm_packages_table[:name].eq(needed_package_versions_table[:name]))
-                      )
-
-            query_results = ApplicationRecord.connection.execute(query.to_sql)
+            query_results = ApplicationRecord.connection.execute(query(components).to_sql)

            package_ids_and_versions = query_results.map do |result|
              Arel::Nodes::Grouping.new([Arel::Nodes.build_quoted(result["id"]),
                Arel::Nodes.build_quoted(result["version"])])
            end

-            records_for_batch = ::PackageMetadata::Package
-                        .joins(package_versions: :licenses)
-                        .where(
-                          Arel::Nodes::In.new(
-                            Arel::Nodes::Grouping.new(first_query_search_fields),
-                            package_ids_and_versions
-                          )
-                        )
-                        .group(search_fields)
-                        .pluck(*search_fields, "ARRAY_AGG(pm_licenses.spdx_identifier)")
-            # rubocop: enable CodeReuse/ActiveRecord
-
-            records_for_batch.each do |purl_type, name, version, licenses|
+            records_for_batch(package_ids_and_versions).each do |purl_type, name, version, licenses|
              records_with_licenses[File.join(name, version, purl_type)] =
                Hashie::Mash.new(purl_type: purl_type, name: name, version: version, licenses: licenses)
            end
@@ -85,7 +32,6 @@ def fetch
          records_with_licenses.values
        end
      end
-      # rubocop: enable Metrics/AbcSize

      private

@@ -102,6 +48,61 @@ def add_records_with_unknown_licenses(records_with_licenses)
        end
      end

+      # rubocop: disable CodeReuse/ActiveRecord
+      def query(components)
+        component_tuples = components.map do |component|
+          purl_type_int = ::Enums::Sbom.purl_types[component.purl_type]
+          [purl_type_int, component.name, component.version]
+        end
+
+        component_values_list = Arel::Nodes::Grouping.new(Arel::Nodes::ValuesList.new(component_tuples))
+
+        pm_packages_table = ::PackageMetadata::Package.arel_table
+
+        needed_package_versions_table = Arel::Table.new(:needed_package_versions)
+
+        component_values = Arel::Nodes::As.new(
+          Hashie::Mash.new(
+            name: Arel.sql('needed_package_versions(purl_type, name, version)')), component_values_list
+        )
+
+        needed_package_versions_table
+          .project(pm_packages_table[:id], needed_package_versions_table[:version])
+          .with(component_values)
+          .distinct
+          .join(pm_packages_table)
+          .on(
+            pm_packages_table[:purl_type].eq(needed_package_versions_table[:purl_type])
+                                         .and(
+                                           pm_packages_table[:name].eq(needed_package_versions_table[:name]))
+          )
+      end
+
+      def records_for_batch(package_ids_and_versions)
+        first_query_search_fields = [
+          ::PackageMetadata::PackageVersion.arel_table[:pm_package_id],
+          ::PackageMetadata::PackageVersion.arel_table[:version]
+        ]
+
+        search_fields = [
+          ::PackageMetadata::Package.arel_table[:purl_type],
+          ::PackageMetadata::Package.arel_table[:name],
+          ::PackageMetadata::PackageVersion.arel_table[:version]
+        ]
+
+        ::PackageMetadata::Package
+          .joins(package_versions: :licenses)
+          .where(
+            Arel::Nodes::In.new(
+              Arel::Nodes::Grouping.new(first_query_search_fields),
+              package_ids_and_versions
+            )
+          )
+          .group(search_fields)
+          .pluck(*search_fields, "ARRAY_AGG(pm_licenses.spdx_identifier)")
+      end
+      # rubocop: enable CodeReuse/ActiveRecord
+
      def use_replica_if_available(&block)
        ::Gitlab::Database::LoadBalancing::Session.current.use_replicas_for_read_queries(&block)
      end

--- a/ee/spec/lib/gitlab/license_scanning/package_licenses_spec.rb
+++ b/ee/spec/lib/gitlab/license_scanning/package_licenses_spec.rb
@@ -67,7 +67,7 @@
        it 'fetches in batches of BATCH_SIZE' do
          control = ActiveRecord::QueryRecorder.new { fetch }

-          # when the batch size is 1, we execute 2 queries for each component
+          # we execute 2 queries for each batch
          expect(control.count).to be(components_to_fetch.count * 2)
        end