Dataset hack

dataset_hack.rb

# Based on the discussion here: https://github.com/aiwilliams/dataset/issues/3
# Used to optimized the datasets being created in your tests.
# Currently works in a test suite for a production app with Postgresql
#
# TODO 
# - gemify
# - add tests
# - be able to reuse datasets accross files (declare them in a shared context)
# - support other databases than postgres
# - file access optimization (only read once during a test run)
# - query optimization (change INSERTS into COPY statements for Postgres) - not sure what the gain will be
# - Remember what is being inserted so it is easier to refer to the created objects (no manual queries)
#
module DatasetHack
  FIXTURE_DIR = Rails.root.join("tmp/datasets")

  module Helpers
    # Capture the sql using active support's notifications
    def capture_sql(filter = nil)
      capture = []
      subscription = ActiveSupport::Notifications.subscribe('sql.active_record') do |name, start, finish, id, payload|
        if filter && payload[:sql] =~ Regexp.new(filter)
          capture << payload[:sql]
        end
      end
      yield
      ActiveSupport::Notifications.unsubscribe(subscription)
      capture
    end
  end
  
  include Helpers
  extend Helpers

  # Hook to reset the database sequences (for Postgres at least)
  # This is needed because otherwise id's will change and then the dataset is possibly not compatible
  # anymore
  # Database cleaner does this by default on truncation
  # 
  # Use this in e.g. your spec_helper  
  #
  #   config.after(:each) do
  #     DatasetHack.reset!
  #   end
  #
  def self.reset!
    capture_sql(/^ALTER/) do    
      (ActiveRecord::Base.connection.tables - ["schema_migrations"]).each do |table|
        ActiveRecord::Base.connection.execute("TRUNCATE TABLE #{table} RESTART IDENTITY")
      end
    end
  end
  
  # In order to not get stuck with old fixtures it is often nice to clean up all 
  # the fixtures e.g. in your spec_helper
  #
  #   config.after(:suite) do
  #     DatasetHack.clean!
  #   end
  # 
  def self.clean!
    Dir[FIXTURE_DIR.join("*")].each { |f| File.delete(f) if File.file?(f) }
  end

  def self.included(klass)
    klass.extend(ClassMethods)
  end
  
  # Add dataset as a macro for rspec
  module ClassMethods
    def dataset(label, &block)
      before(:each) do
        dataset(label) { instance_eval(&block) }
      end
    end
  end

  # Call dataset with a label and a block to cache the generate queries e.g.
  #
  #   dataset :very_expensive_operation do
  #     1000.times { User.create! }
  #  end
  #
  def dataset(label)
    sql_file = FIXTURE_DIR.join("data_#{label}.sql")
    if File.exist?(sql_file)
      ActiveRecord::Base.connection.execute(File.read(sql_file))
    else
      captured_sql = capture_sql(/^INSERT/) { yield }
      transformed_sql = captured_sql.join(';')
      File.open(sql_file, 'w') {|f| f.write(transformed_sql) }
    end
  end
end