SeedMigrator

The problem of seed data can get annoying once your Rails app is in production. Ordinarily, you would place your seeds data in seeds.rb. Unfortunately, once your application goes live, you will likely not be in a position to reload your seeds. This leaves you with Rails migrations as the likely choice, but that now makes development harder, because things seed out of order (migrations run before seeds). This gem solves these problems.

Installation

In your Gemfile:

gem 'seed_migrator'

Usage

Data updates are defined similar to migrations. Each file contains a class definition which should ideally extend SeedMigrator::Updater but doesn't have to as long as it implements apply_update and revert_update. They need to follow the default Rails naming convention; a file called update_order_types.rb should contain the class UpdateOrderTypes. It is highly recommended that each file have a prefix that defines its order. The format is pretty flexible, but the prefix must start with a number and not have any underscores. So 01_update_order_types.rb is fine, so is 1A5_update_order_types.rb. In each of these cases, the name of the class is still UpdateOrderTypes. If you extend SeedMigrator::Updater you only need to override revert_update if you need your migration to be reversible, otherwise it's not necessary.

Here's an example data update class definition:

class UpdateOrderTypes < SeedMigrators::Updater
  def perform_update
    OrderType.create :type_code => 'very_shiny'
  end

  # Overriden in case we need to roll back this migration.
  def undo_update
    OrderType.where(:type_code => 'very_shiny').first.delete
  end
end

Migrations

Assuming we have created the update file above in db/data_updates in a file named 01_update_order_types.rb

root_updates_path and optionally should_run?(update_name) must be defined in every migration where we intend to do data updates. Realistically, our app should extend SeedMigrator and then include the new module in each migration where needed.

module CoreDataUpdate
  include SeedMigrator

  def root_updates_path
    Rails.root.join('db','data_updates')
  end

  def should_run?(update_name)
    Rails.env == 'production'
  end
end

Now, define a migration that will perform our data update.

class CreateVeryShinyOrderType < ActiveRecord::Migration
  include CoreDataUpdate

  def up
    apply_update :01_update_order_types
  end

  def down
    revert_update :01_update_order_types
  end
end

Note that the update prefix is optional. The following will also work.

class CreateVeryShinyOrderType < ActiveRecord::Migration
  include CoreDataUpdate

  def up
    apply_update :update_order_types
  end

  def down
    revert_update :update_order_types
  end
end

Old style migrations, i.e. def self.up are not supported.

Seeds

At the bottom of your seeds.rb, include the following:

include SeedMigrator::Seeds
apply_updates Rails.root.join('db','data_updates')

Generators

Two generators are added for your convenience:

install: Updates the seeds file as specified above and creates a custom data update module for engines. See USAGE for examples
create: Creates new data_update and migration files as specified above. See USAGE for examples

NOTE: If you're using this from a Rails engine with Rails 3.2 and you're using RSpec to test, invoking the generator will likely put the generated files in spec/dummy/db instead of db. Just something to be aware of.

Testing

Install all the gem dependencies.

bundle install

Run tests

rake spec