[RUBY] Rails: How to write a rake task nicely

Good feeling = the following two points

1. Easy to test

• To reduce man-hours

1. (To some extent) the writing style can be unified among engineers

• For example, if the method of specifying dryrun is different depending on the writer, an accident may occur when executing a task in commercial use, so it is better if it can be unified.
• If the log format is not unified, work efficiency will be poor.

etc

code

lib/tasks/issue_6885.rake

require_relative 'helpers/all_user_name_update_helper.rb'

namespace :issue_6885 do

  desc 'This is a sample'
  task all_user_name_update: :environment do
    helper = AllUserNameUpdateHelper.new

    helper.main
  end
end

lib/tasks/helpers/all_user_name_update_helper.rb

require_relative 'rake_helper_template'

class AllUserNameUpdateHelper < RakeHelperTemplate
  NEW_NAME = 'bar'

  def main
    template do |logger|
      all_users = get_all_user
      logger.info("Number of target users: #{all_users.size}")

      if all_users.blank?
        logger.info('The process ends because the number of target users was 0.')
        return
      end

      all_users.each { |user| update_name(user, NEW_NAME) }
    end
  end

  private

  def get_all_user
    User.all
  end

  def update_name(user, new_name)
    user.update!(name: new_name)
    user
  end
end

lib/tasks/helpers/rake_helper_template.rb

class RakeHelperTemplate
  #Also performs standard output
  def make_logger(log_file_path)
    logger = ActiveSupport::Logger.new(log_file_path)
    stdout_logger = ActiveSupport::Logger.new(STDOUT)
    broadcast_logger = ActiveSupport::Logger.broadcast(stdout_logger)
    logger.extend(broadcast_logger)
    logger.formatter = Logger::Formatter.new
    logger
  end

  #Make the log file name the same as the task name (replace it because it is better not to use a colon)
  #Cases called from other than the execution of the rake task (eg:spec) should be taken into consideration
  def make_log_file_path
    task_name = Rake.try(:application)&.top_level_tasks&.[](0)&.gsub(':','_') || Rails.env
    log_file_name = "#{task_name}.log"
    Rails.root.join('log', log_file_name)
  end

  def template
    log_file_path = make_log_file_path

    #Be sure to dryrun unless you explicitly pass the string false
    is_dryrun = ENV['is_dryrun'] != 'false'

    logger = make_logger(log_file_path)
    logger.info("Start. is_dryrun: #{is_dryrun}")

    ActiveRecord::Base.transaction do
      yield(logger)
      raise ActiveRecord::Rollback if is_dryrun
    end

    logger.info("Finish. log_file_path: #{log_file_path}")
  end
end

Execution example

When making a dry run

$ is_dryrun=false bundle exec rake issue_6885:all_user_name_update
I, [2020-09-27T13:33:27.229764 #13040]  INFO -- : Start. is_dryrun: true
I, [2020-09-27T13:33:27.556890 #13040]  INFO -- :Number of target users: 1
I, [2020-09-27T13:33:27.755933 #13040]  INFO -- : Finish. log_file_path: /app/log/issue_6885_all_user_name_update.log 

point

1. The .rake file only calls the helper class, rake's DSL? I tried to test the method of the class that I am accustomed to writing without worrying about the way of doing things

• Reference: Qiita-A simple and effective way to test Rake tasks with Rails

2. Use yield to standardize dryrun and logger (part of) so that you don't have to define them every time you create a rake task.

• In the main method (any name can be used), define only the processing flow of the task as much as possible, and cut out individual specific processing into another method with a particle size that is easy to test.

Once you have a template, you can concentrate on what you want to do like all_user_name_update_helper.rb! (^ ○ ^)

Impressions

• I forgot how to write the rake task after a long time, so this article actually has a memorial meaning.
• Yield is honestly difficult to understand, so I don't want to use it, but I think it's okay if it's something that you don't write or read every time like a template.
• If you have any other good writing styles, please let me know!