The purpose of ActiveModel::Serializers is to provide an object to encapsulate serialization of ActiveModel objects, including ActiveRecord objects.

Serializers know about both a model and the current_user, so you can customize serialization based upon whether a user is authorized to see the content.

In short, serializers replaces hash-driven development with object-oriented development.

For now, the easiest way to install ActiveModel::Serializers is to add this to your Gemfile:

gem "active_model_serializers", :git => "git://github.com/josevalim/active_model_serializers.git"

Then, install it on the command line:

$ bundle install
$ rails g serializers:install

The installation will generate an app/serializers directory and create an ApplicationSerializer inside it. You can use the ApplicationSerializer for common functionality, and to customize the behavior of embedded associations. See more on this later.

The easiest way to create a new serializer is to generate a new resource, which will generate a serializer at the same time:

$ rails g resource post title:string body:string 

This will generate a serializer in app/serializers/post_serializer.rb for your new model. You can also generate a serializer for an existing model with the serializer generator:

$ rails g serializer post

In your controllers, when you use render :json, Rails will now first search for a serializer for the object and use it if available.

class PostController < ApplicationController
	def show
		@post = Post.find(params[:id])
		render :json => @post
	end
end

In this case, Rails will look for a serializer named PostSerializer, and if it exists, use it to serialize the Post. If it does not exist, Rails will simply call to_json on the @post object.

This also works with render_with, which uses to_json under the hood. Also note that any options passed to render :json will be passed to your serializer and available as @options inside.

Once you have a serializer, you can specify which attributes and associations you would like to include in the serialized form.

class PostSerializer < ApplicationSerializer
	attributes :id, :title, :body
	has_many :comments
end

For specified attributes, the serializer will look up the attribute on the object you passed to render :json. It uses read_attribute_for_serialization, which ActiveRecord objects implement as a regular attribute lookup.

If you would like the key in the outputted JSON to be different from its name in ActiveRecord, you can use the :key option to customize it:

class PostSerializer < ApplicationSerializer
	attributes :id, :body

	# look up :subject on the model, but use +title+ in the JSON
	attribute :subject, :key => :title
	has_many :comments
end

For specified associations, the serializer will look up the association and then serialize each element of the association. For instance, a has_many :comments association will create a new CommentSerializer for each comment and use it to serialize the comment.

By default, serializers simply look up the association on the original object. You can customize this behavior by implementing a method with the name of the association and returning a different Array. Often, you will do this to customize the objects returned based on the current user.

class PostSerializer < ApplicationSerializer
	attributes :id, :title, :body
	has_many :comments

	# only let the user see comments he created.
	def comments
		post.comments.where(:created_by => @scope)
	end
end

In a serializer, @scope is the current authorization scope (usually current_user), which the controller gives to the serializer when you call render :json

As with attributes, you can also change the JSON key that the serializer should use for a particular association.

class PostSerializer < ApplicationSerializer
	attributes :id, :title, :body

	# look up comments, but use +my_comments+ as the key in JSON
	has_many :comments, :key => :my_comments
end

By default, associations will be embedded inside the serialized object. So if you have a post, the outputted JSON will look like:

{
	"post": {
		"id": 1,
		"title": "New post",
		"body": "A body!",
		"comments": [
			{ "id": 1, "body": "what a dumb post" }
		]
	}
}

This is convenient for simple use-cases, but for more complex clients, it is better to supply an Array of IDs for the association. This makes your API more flexible from a performance standpoint and avoids wasteful duplication.

To embed IDs instead of associations, simply use the embed class method:

class PostSerializer < ApplicationSerializer
	embed :ids

	attributes :id, :title, :body
	has_many :comments
end

Now, any associations will be supplied as an Array of IDs:

{
	"post": {
		"id": 1,
		"title": "New post",
		"body": "A body!",
		"comments": [ 1, 2, 3 ]
	}
}

In addition to supplying an Array of IDs, you may want to side-load the data alongside the main object. This makes it easier to process the entire package of data without having to recursively scan the tree looking for embedded information. It also ensures that associations that are shared between several objects (like tags), are only delivered once for the entire payload.

You can specify that the data be included like this:

class PostSerializer < ApplicationSerializer
	embed :ids, :include => true

	attributes :id, :title, :body
	has_many :comments
end

Assuming that the comments also has_many :tags, you will get a JSON like this:

{
	"post": {
		"id": 1,
		"title": "New post",
		"body": "A body!",
		"comments": [ 1 ]
	},
	"comments": [
		{ "id": 1, "body": "what a dumb post", "tags": [ 1, 2 ] },
		{ "id": 1, "body": "i liked it", "tags": [ 1, 3 ] },
	],
	"tags": [
		{ "id": 1, "name": "short" },
		{ "id": 2, "name": "whiny" },
		{ "id": 3, "name": "happy" }
	]
}

NOTE: The embed :ids mechanism is primary useful for clients that process data in bulk and load it into a local store. For these clients, the ability to easily see all of the data per type, rather than having to recursively scan the data looking for information, is extremely useful.

If you are mostly working with the data in simple scenarios and manually making Ajax requests, you probably just want to use the default embedded behavior.