summaryrefslogtreecommitdiff
path: root/rubocop
diff options
context:
space:
mode:
authorGitLab Bot <gitlab-bot@gitlab.com>2020-02-19 18:09:10 +0000
committerGitLab Bot <gitlab-bot@gitlab.com>2020-02-19 18:09:10 +0000
commit33795139ea8e72756bee3675b4e16387425e6ab1 (patch)
tree3ca568fca61482e57810ee30ad5ce4b964a82c4e /rubocop
parentc7e385e282bcb8505589bce526e692b7bb819ffa (diff)
downloadgitlab-ce-33795139ea8e72756bee3675b4e16387425e6ab1.tar.gz
Add latest changes from gitlab-org/gitlab@master
Diffstat (limited to 'rubocop')
-rw-r--r--rubocop/cop/migration/schedule_async.rb54
-rw-r--r--rubocop/cop/scalability/idempotent_worker.rb59
-rw-r--r--rubocop/migration_helpers.rb4
3 files changed, 117 insertions, 0 deletions
diff --git a/rubocop/cop/migration/schedule_async.rb b/rubocop/cop/migration/schedule_async.rb
new file mode 100644
index 00000000000..f296628c3d6
--- /dev/null
+++ b/rubocop/cop/migration/schedule_async.rb
@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+require_relative '../../migration_helpers'
+
+module RuboCop
+ module Cop
+ module Migration
+ class ScheduleAsync < RuboCop::Cop::Cop
+ include MigrationHelpers
+
+ ENFORCED_SINCE = 2020_02_12_00_00_00
+
+ MSG = <<~MSG
+ Don't call the background migration worker directly, use the `#migrate_async`,
+ `#migrate_in`, `#bulk_migrate_async` or `#bulk_migrate_in` migration helpers
+ instead.
+ MSG
+
+ def_node_matcher :calls_background_migration_worker?, <<~PATTERN
+ (send (const nil? :BackgroundMigrationWorker) {:perform_async :perform_in :bulk_perform_async :bulk_perform_in} ... )
+ PATTERN
+
+ def on_send(node)
+ return unless in_migration?(node)
+ return if version(node) < ENFORCED_SINCE
+
+ add_offense(node, location: :expression) if calls_background_migration_worker?(node)
+ end
+
+ def autocorrect(node)
+ # This gets rid of the receiver `BackgroundMigrationWorker` and
+ # replaces `perform` with `schedule`
+ schedule_method = method_name(node).to_s.sub('perform', 'migrate')
+ arguments = arguments(node).map(&:source).join(', ')
+
+ replacement = "#{schedule_method}(#{arguments})"
+ lambda do |corrector|
+ corrector.replace(node.source_range, replacement)
+ end
+ end
+
+ private
+
+ def method_name(node)
+ node.children.second
+ end
+
+ def arguments(node)
+ node.children[2..-1]
+ end
+ end
+ end
+ end
+end
diff --git a/rubocop/cop/scalability/idempotent_worker.rb b/rubocop/cop/scalability/idempotent_worker.rb
new file mode 100644
index 00000000000..a38b457b7c7
--- /dev/null
+++ b/rubocop/cop/scalability/idempotent_worker.rb
@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+require_relative '../../code_reuse_helpers.rb'
+
+module RuboCop
+ module Cop
+ module Scalability
+ # This cop checks for the `idempotent!` call in the worker scope.
+ #
+ # @example
+ #
+ # # bad
+ # class TroubleMakerWorker
+ # def perform
+ # end
+ # end
+ #
+ # # good
+ # class NiceWorker
+ # idempotent!
+ #
+ # def perform
+ # end
+ # end
+ #
+ class IdempotentWorker < RuboCop::Cop::Cop
+ include CodeReuseHelpers
+
+ HELP_LINK = 'https://github.com/mperham/sidekiq/wiki/Best-Practices#2-make-your-job-idempotent-and-transactional'
+
+ MSG = <<~MSG
+ Avoid adding not idempotent workers.
+
+ A worker is considered idempotent if:
+
+ 1. It can safely run multiple times with the same arguments
+ 2. The application side-effects are expected to happen once (or side-effects of a second run are not impactful)
+ 3. It can safely be skipped if another job with the same arguments is already in the queue
+
+ If all the above is true, make sure to mark it as so by calling the `idempotent!`
+ method in the worker scope.
+
+ See #{HELP_LINK}
+ MSG
+
+ def_node_search :idempotent?, <<~PATTERN
+ (send nil? :idempotent!)
+ PATTERN
+
+ def on_class(node)
+ return unless in_worker?(node)
+ return if idempotent?(node)
+
+ add_offense(node, location: :expression)
+ end
+ end
+ end
+ end
+end
diff --git a/rubocop/migration_helpers.rb b/rubocop/migration_helpers.rb
index 577f768da67..767cacaecaf 100644
--- a/rubocop/migration_helpers.rb
+++ b/rubocop/migration_helpers.rb
@@ -10,6 +10,10 @@ module RuboCop
dirname(node).end_with?('db/post_migrate', 'db/geo/post_migrate')
end
+ def version(node)
+ File.basename(node.location.expression.source_buffer.name).split('_').first.to_i
+ end
+
private
def dirname(node)