¶ ↑
TruckerTrucker is a gem that helps migrate legacy data into a Rails app.
¶ ↑
Installation-
Install the trucker gem
sudo gem install trucker
-
Add trucker to your config.gem block in environment.rb
config.gem "trucker"
-
Generate the basic trucker files
rails generate truck This will do the following things: * Add legacy adapter to database.yml * Add legacy base class * Add legacy sub classes for all existing models * Add app/models/legacy to autoload_paths in Rails Initializer config block * Generate sample migration task (using pluralized model names)
-
Update the legacy database adapter in database.yml with your legacy database info
legacy: adapter: mysql encoding: utf8 database: app_legacy username: root password: (By convention, we recommend naming your legacy database APP_NAME_legacy.)
-
If the legacy database doesn’t already exist, add it.
rake db:create:all
-
Import your legacy data into the legacy database
mysql -u root app_legacy < old_database.sql
-
Update set_table_name in each of your legacy models as needed
class LegacyPost < LegacyBase set_table_name "LEGACY_TABLE_NAME_GOES_HERE" end
-
Update legacy model field mappings as needed
class LegacyPost < LegacyBase set_table_name "YOUR LEGACY TABLE NAME GOES HERE" def map { :headline => self.title.squish, :body => self.long_text.squish } end end New model attributes on the left side. Legacy model attributes on the right side. (aka :new_field => legacy_field)
-
Need to tweak some data? Just add some core ruby methods or add a helper method.
class LegacyPost < LegacyBase set_table_name "YOUR LEGACY TABLE NAME GOES HERE" def map { :headline => self.title.squish.capitalize, # <= Added capitalize method :body => tweak_body(self.long_text.squish) # <= Added tweak_body method } end # Insert helper methods as needed def tweak_body(body) body = body.gsub(/<br \//,"\n") # <= Convert break tags into normal line breaks body = body.gsub(/teh/, "the") # <= Fix common typos end end
-
Start migrating!
rake db:migrate:posts
¶ ↑
Migration command line optionsTrucker supports a few command line options when migrating records:
rake db:migrate:posts limit=100 # migrates 100 records rake db:migrate:posts limit=100 offset=100 # migrates 100 records, but skips first 100 records rake db:migrate:users limit=1 where=":username => 'gilesgoatboy'" # migrates just one record
You can pass just about anything to a where clause, but checking it in the console ahead of time is strongly recommended. Most obvious use case: migrating just one record to verify migrating happens the way you want it to.
¶ ↑
Custom migration labelsYou can tweak the default migration output generated by Trucker by using the :label option.
rake db:migrate:posts => Migrating posts rake db:migrate:posts, :label => "blog posts" => Migrating blog posts
¶ ↑
Custom helpersTrucker is intended for migrating data from fairly simple web apps that started life on PHP, Perl, etc. So, if you’re migrating data from an enterprise system, this may not be your best choice.
That said, if you need to pull off a complex migration for a model, you can use a custom helper method to override Trucker’s default migrate method in your rake task.
namespace :db do namespace :migrate do ... desc 'Migrate pain_in_the_ass model' task :pain_in_the_ass => :environment do Trucker.migrate :pain_in_the_ass, :helper => pain_in_the_ass_migration end end end def pain_in_the_ass_migration # Custom code goes here end
Then just copy the migrate method from lib/trucker.rb and tweak accordingly.
As an example, here’s a custom helper used to migrate join tables on a bunch of models.
namespace :db do namespace :migrate do desc 'Migrates join tables' task :joins => :environment do migrate :joins, :helper => :migrate_joins end end end def migrate_joins puts "Migrating #{number_of_records || "all"} joins #{"after #{offset_for_records}" if offset_for_records}" ["chain", "firm", "function", "style", "website"].each do |model| # Start migration puts "Migrating theaters_#{model.pluralize}" # Delete existing joins ActiveRecord::Base.connection.execute("TRUNCATE table theaters_#{model.pluralize}") # Tweak model ids and foreign keys to match model syntax if model == 'website' model_id = "url_id" send_foreign_key = "url_id".to_sym else model_id = "#{model}_id" send_foreign_key = "#{model}_id".to_sym end # Create join object class join = Object.const_set("Theaters#{model.classify}", Class.new(ActiveRecord::Base)) # Set model foreign key model_foreign_key = "#{model}_id".to_sym # Migrate join (unless duplicate) "LegacyTheater#{model.classify}".constantize.find(:all, with(:order => model_id)).each do |record| unless join.find(:first, :conditions => {:theater_id => record.theater_id, model_foreign_key => record.send(send_foreign_key)}) attributes = { model_foreign_key => record.send(send_foreign_key), :theater_id => record.theater_id } # Check if theater chain is current attributes[:is_current] = {'Yes' => 1, 'No' => 0, '' => 0}[record.current] if model == 'chain' # Migrate join join.create(attributes) end end end end
¶ ↑
ProTip: Catching Validation ErrorsHere’s the standard ‘migrate` method in `LegacyBase`:
def migrate new_record = self.class.to_s.gsub(/Legacy/,'::').constantize.new(map) new_record[:id] = self.id new_record.save end
And here’s a handy variant:
def migrate new_record = self.class.to_s.gsub(/Legacy/,'::').constantize.new(map) new_record[:id] = self.id begin new_record.save! rescue Exception => e puts "error saving #{new_record.class} #{new_record.id}!" puts e.inspect end end
Note that this version responds to failed saves by simply logging it to ‘stderr` and going on to the next one like Jay-Z. In practice, this might not be what you want. Your mileage may vary, so season to taste.
¶ ↑
Sample applicationCheck out the Trucker sample app for a working example of Trucker-based legacy data migration.
¶ ↑
Rails 3.1 CompatibilityAdapted Trucker itself, but not its generators, to Rails 3.1 compatibility. Further work remains to bring the generators up to date. If you use Trucker with the current (June 2011) Rails 3.1 release candidate and MySQL, make sure you use the latest mysql2 gem (currently 0.3.2), as other versions, or the original mysql gem, will cause explosive pants-shitting failure. (Postgres and SQLite are fine.)
Rails 3.1 compatible code lives on the rails31 branch. The functionality’s refactored to an object- oriented model in the refactor branch.
¶ ↑
BackgroundTrucker is based on a migration technique using legacy models first pioneered by Dave Thomas: pragdave.blogs.pragprog.com/pragdave/2006/01/sharing_externa.html
¶ ↑
Note on patches/pull requests-
Fork the project.
-
Make your feature addition or bug fix.
-
Add tests for it. This is important so we don’t break a future version unintentionally.
-
Commit your changes, but do not mess with the rakefile, version, or history. (if you want to have your own version, that is fine but bump version in a commit by itself so we can ignore when we pull)
-
Send a pull request. Bonus points for topic branches.
¶ ↑
Contributors-
Patrick Crowley / mokolabs
-
Rob Kaufman / notch8
-
Jordan Fowler / TheBreeze
-
Roel Bondoc / roelbondoc
-
Giles Bowkett / gilesbowkett (GitHub) / gilesgoatboy (Twitter)
¶ ↑
CopyrightCopyright © 2010 Patrick Crowley and Rob Kaufman. See LICENSE for details.