id

simple, immutable models based on hashes

JSON is a great way to transfer data between systems, and it's easy to parse into a Ruby hash. But sometimes it's nice to have actual methods to call when you want to get attributes from your data, rather than coupling your entire codebase to the hash representation by littering it with calls to fetch or []. The same goes for BSON documents stored in Mongo.

That's where id (as in Freud) comes in. You define your model classes using syntax that should look pretty familiar if you've used any popular Ruby ORMs - but id is not an ORM. Model objects defined with id have a constructor that accepts a hash, and you define the values of this hash that are made readable as fields, but that hash can come from any source.

Defining a model

To define a model, include the Id::Model module in your class.

class Cat
  include Id::Model
end

Defining fields

At its most basic, you can define a field like this:

class Cat
  field :name
end

This gives you the equivalent of an attr_reader for that field, as well as a predicate method which checks if it has been set (in the case of boolean fields, this predicate method also checks the value of that field - nil is interpreted as false).

cat = Cat.new(name: "Travis")
cat.name  # => "Travis"
cat.name? # => true

cat = Cat.new
cat.name? # => false

What happens if you try to access a field that hasn't been set?

irb(main):007:0> Cat.new.name
  Id::MissingAttributeError: Cat had a nil value for 'name'.

Id is allergic to nil, and refuses to return it. If you want to access a field you need to be sure it's there, or you'll get an error. This means if you have a bug in your code and a field isn't set, you'll find out as soon as possible, rather that letting a nil leak out and cause an undefined method 'foo' for nil:NilClass at some unspecified future point, from where it might be hard to track down the source of the problem.

What about optional fields?

Sometimes fields really are optional. In this case you can either test for their presence using the predicate methods (which is a bit ugly, but at least forces you to deal with the empty case), or you can mark the field as optional: true, which will make them return an Option type rather than a raw value.

Option types are a pattern found in many functional languages (the Option type in Scala, the Maybe monad in Haskell) and the Ruby implementation used by id can be found here.

Can I set default values?

Yes. Default values are specified like this:

class Cat
  ...
  field :paws, default: 4
end

Cat.new.paws # => 4

You can also specify lambda defaults. These will be run on initialization of an instance of your model.

class Cat
  ...
  field :birthday, default: -> { Date.new }
end

Cat.new.birthday # => #<Date: 2013-10-21 ((2456587j,0s,0n),+0s,2299161j)>

I don't like the key names in my data-source :-(

We don't always get what we want. But don't despair! If the hash you're using to create your models has badly named keys - e.g., horror of horrors, keys in camelCase - you can use key aliases to convert them to nice, succinct, Rubyish identifiers:

class Camel
  include Id::Model

  field :name,  key: 'camelName'
  field :humps, key: 'camelHumps'
end

Camel.new('camelName' => 'Terry').name # => "Terry"

Type Coercion

Id can coerce your fields into a number of basic types. Just specify the type you want as part of the field definition.

class Cat
  ...
  field :paws, type: Integer
end

Cat.new(paws: "3").paws => 3
Cat.new(paws: "3").paws.class => Integer

You can typecast array elements like this:

field :counts, type: Array[Integer]

Because Ruby doesn't have a Boolean type (weird, right?), if you want to coerce something to either true or false, you need to use Id::Boolean, like this:

field :admin, type: Id::Boolean

And if you need to coerce a type id doesn't yet support, such as a custom type of your own, you can register new coercions by passing the "from" type, "to" type, and a block to perform the conversion to Id::Coercion.register.

Id::Coercion.register String, Money do |value|
  value.to_money
end

or more succinctly:

Id::Coercion.register String, Money, &:to_money

Associations

If you have nested arrays or hashes in your source, you can define id models for them in turn, and treat them much like you would associations in an ORM, by defining them as has_one or has_many associations on the parent model.

class Lion
  include Id::Model
  field :name
end

class Person
  include Id::Model
  field :name
end

class Zoo
  include Id::Model

  has_many :lions
  has_one :zookeeper, type: Person
end

zoo = Zoo.new(lions: [{name: 'Hetty'}],
              zookeeper: {name: 'Russell'})

zoo.lions.first.class # => Lion
zoo.lions.first.name  # => "Hetty"
zoo.zookeeper.class   # => Person
zoo.zookeeper.name    # => "Russell"

Types are inferred from the association name unless specified.

Designed for immutability

id models provide accessor methods, but no mutator methods, because they are designed for immutability. How do immutable models work? When you need to change some field of a model object, a new copy of the object is created with the field changed as required. This is handled for you by id's set method:

person1 = Person.new(name: 'Russell', job: 'programmer')
person2 = person1.set(name: 'Radek')
person1.name # => 'Russell'
person2.name # => 'Radek'
person2.job  # => 'programmer'

Obviously, this is Ruby, and if you really want to mutate some of the internal state of an id model you will be able to find a way to do it. Don't do it!

Id and Rails

We might not love everything about Rails, but we probably use it at least some of the time. So does id play nicely with it?

With a few modifications, yes. Models that blow up at the sight of nil don't play happy with Rails' nil-happy forms, but you can make your id models active model compliant in forms, while otherwise retaining the nil-allergy in the rest of your code, by doing two things.

Include the Id::Form module in your model.

class Cat
  include Id::Model
  include Id::Form
end

Add the following line to config/application.rb:

config.action_view.default_form_builder = Id::FormBuilder

With Id::Form included you can use Active Model validations as normal, and override methods like to_partial_path, self.model_name, and persisted? right there in your model.

Timestamps

And finally, it's reasonably common to want to know when a particular model was created and/or updated. Id provides you this functionality out of the box through the Id::Timestamps module. Just include it to your model to get created_at and updated_at fields that behave as you would expect.