A lightning fast JSON:API serializer for Ruby Objects.
We compare serialization times with Active Model Serializer as part of RSpec performance tests included on this library. We want to ensure that with every change on this library, serialization time is at least 25 times
faster than Active Model Serializers on up to current benchmark of 1000 records. Please read the performance document for any questions related to methodology.
$ rspec
Active Model Serializer serialized 250 records in 138.71 ms
Fast JSON API serialized 250 records in 3.01 ms
- Declaration syntax similar to Active Model Serializer
- Support for
belongs_to
,has_many
andhas_one
- Support for compound documents (included)
- Optimized serialization of compound documents
- Caching
Add this line to your application's Gemfile:
gem 'fast_jsonapi'
Execute:
$ bundle install
You can use the bundled generator if you are using the library inside of a Rails project:
rails g serializer Movie name year
This will create a new serializer in app/serializers/movie_serializer.rb
class Movie
attr_accessor :id, :name, :year, :actor_ids, :owner_id, :movie_type_id
end
class MovieSerializer
include FastJsonapi::ObjectSerializer
set_type :movie # optional
set_id :owner_id # optional
attributes :name, :year
has_many :actors
belongs_to :owner, record_type: :user
belongs_to :movie_type
end
movie = Movie.new
movie.id = 232
movie.name = 'test movie'
movie.actor_ids = [1, 2, 3]
movie.owner_id = 3
movie.movie_type_id = 1
movie
hash = MovieSerializer.new(movie).serializable_hash
json_string = MovieSerializer.new(movie).serialized_json
{
"data": {
"id": "3",
"type": "movie",
"attributes": {
"name": "test movie",
"year": null
},
"relationships": {
"actors": {
"data": [
{
"id": "1",
"type": "actor"
},
{
"id": "2",
"type": "actor"
}
]
},
"owner": {
"data": {
"id": "3",
"type": "user"
}
}
}
}
}
By default fast_jsonapi underscores the key names. It supports the same key transforms that are supported by AMS. Here is the syntax of specifying a key transform
class MovieSerializer
include FastJsonapi::ObjectSerializer
# Available options :camel, :camel_lower, :dash, :underscore(default)
set_key_transform :camel
end
Here are examples of how these options transform the keys
set_key_transform :camel # "some_key" => "SomeKey"
set_key_transform :camel_lower # "some_key" => "someKey"
set_key_transform :dash # "some_key" => "some-key"
set_key_transform :underscore # "some_key" => "some_key"
Attributes are defined in FastJsonapi using the attributes
method. This method is also aliased as attribute
, which is useful when defining a single attribute.
By default, attributes are read directly from the model property of the same name. In this example, name
is expected to be a property of the object being serialized:
class MovieSerializer
include FastJsonapi::ObjectSerializer
attribute :name
end
Custom attributes that must be serialized but do not exist on the model can be declared using Ruby block syntax:
class MovieSerializer
include FastJsonapi::ObjectSerializer
attributes :name, :year
attribute :name_with_year do |object|
"#{object.name} (#{object.year})"
end
end
The block syntax can also be used to override the property on the object:
class MovieSerializer
include FastJsonapi::ObjectSerializer
attribute :name do |object|
"#{object.name} Part 2"
end
end
Attributes can also use a different name by passing the original method or accessor with a proc shortcut:
class MovieSerializer
include FastJsonapi::ObjectSerializer
attributes :name
attribute :released_in_year, &:year
end
Links are defined in FastJsonapi using the link
method. By default, links are read directly from the model property of the same name. In this example, public_url
is expected to be a property of the object being serialized.
You can configure the method to use on the object for example a link with key self
will get set to the value returned by a method called url
on the movie object.
You can also use a block to define a url as shown in custom_url
. You can access params in these blocks as well as shown in personalized_url
class MovieSerializer
include FastJsonapi::ObjectSerializer
link :public_url
link :self, :url
link :custom_url do |object|
"http://movies.com/#{object.name}-(#{object.year})"
end
link :personalized_url do |object, params|
"http://movies.com/#{object.name}-#{params[:user].reference_code}"
end
end
You can specify relationship links by using the links:
option on the serializer. Relationship links in JSON API are useful if you want to load a parent document and then load associated documents later due to size constraints (see related resource links)
class MovieSerializer
include FastJsonapi::ObjectSerializer
has_many :actors, links: {
self: :url,
related: -> (object) {
"https://movies.com/#{object.id}/actors"
}
}
end
This will create a self
reference for the relationship, and a related
link for loading the actors relationship later. NB: This will not automatically disable loading the data in the relationship, you'll need to do that using the lazy_load_data
option:
has_many :actors, lazy_load_data: true, links: {
self: :url,
related: -> (object) {
"https://movies.com/#{object.id}/actors"
}
}
For every resource in the collection, you can include a meta object containing non-standard meta-information about a resource that can not be represented as an attribute or relationship.
meta do |movie|
{
years_since_release: Date.current.year - movie.year
}
end
end
Support for top-level and nested included associations through options[:include]
.
options = {}
options[:meta] = { total: 2 }
options[:links] = {
self: '...',
next: '...',
prev: '...'
}
options[:include] = [:actors, :'actors.agency', :'actors.agency.state']
MovieSerializer.new([movie, movie], options).serialized_json
options[:meta] = { total: 2 }
options[:links] = {
self: '...',
next: '...',
prev: '...'
}
hash = MovieSerializer.new([movie, movie], options).serializable_hash
json_string = MovieSerializer.new([movie, movie], options).serialized_json
You can use is_collection
option to have better control over collection serialization.
If this option is not provided or nil
autedetect logic is used to try understand
if provided resource is a single object or collection.
Autodetect logic is compatible with most DB toolkits (ActiveRecord, Sequel, etc.) but cannot guarantee that single vs collection will be always detected properly.
options[:is_collection]
was introduced to be able to have precise control this behavior
nil
or not provided: will try to autodetect single vs collection (please, see notes above)true
will always treat input resource as collectionfalse
will always treat input resource as single object
Requires a cache_key
method be defined on model:
class MovieSerializer
include FastJsonapi::ObjectSerializer
set_type :movie # optional
cache_options enabled: true, cache_length: 12.hours
attributes :name, :year
end
In some cases, attribute values might require more information than what is
available on the record, for example, access privileges or other information
related to a current authenticated user. The options[:params]
value covers these
cases by allowing you to pass in a hash of additional parameters necessary for
your use case.
Leveraging the new params is easy, when you define a custom attribute or relationship with a block you opt-in to using params by adding it as a block parameter.
class MovieSerializer
include FastJsonapi::ObjectSerializer
attributes :name, :year
attribute :can_view_early do |movie, params|
# in here, params is a hash containing the `:current_user` key
params[:current_user].is_employee? ? true : false
end
belongs_to :primary_agent do |movie, params|
# in here, params is a hash containing the `:current_user` key
params[:current_user].is_employee? ? true : false
end
end
# ...
current_user = User.find(cookies[:current_user_id])
serializer = MovieSerializer.new(movie, {params: {current_user: current_user}})
serializer.serializable_hash
Custom attributes and relationships that only receive the resource are still possible by defining the block to only receive one argument.
Conditional attributes can be defined by passing a Proc to the if
key on the attribute
method. Return true
if the attribute should be serialized, and false
if not. The record and any params passed to the serializer are available inside the Proc as the first and second parameters, respectively.
class MovieSerializer
include FastJsonapi::ObjectSerializer
attributes :name, :year
attribute :release_year, if: Proc.new { |record|
# Release year will only be serialized if it's greater than 1990
record.release_year > 1990
}
attribute :director, if: Proc.new { |record, params|
# The director will be serialized only if the :admin key of params is true
params && params[:admin] == true
}
end
# ...
current_user = User.find(cookies[:current_user_id])
serializer = MovieSerializer.new(movie, { params: { admin: current_user.admin? }})
serializer.serializable_hash
Conditional relationships can be defined by passing a Proc to the if
key. Return true
if the relationship should be serialized, and false
if not. The record and any params passed to the serializer are available inside the Proc as the first and second parameters, respectively.
class MovieSerializer
include FastJsonapi::ObjectSerializer
# Actors will only be serialized if the record has any associated actors
has_many :actors, if: Proc.new { |record| record.actors.any? }
# Owner will only be serialized if the :admin key of params is true
belongs_to :owner, if: Proc.new { |record, params| params && params[:admin] == true }
end
# ...
current_user = User.find(cookies[:current_user_id])
serializer = MovieSerializer.new(movie, { params: { admin: current_user.admin? }})
serializer.serializable_hash
Attributes and relationships can be selectively returned per record type by using the fields
option.
class MovieSerializer
include FastJsonapi::ObjectSerializer
attributes :name, :year
end
serializer = MovieSerializer.new(movie, { fields: { movie: [:name] } })
serializer.serializable_hash
Option | Purpose | Example |
---|---|---|
set_type | Type name of Object | set_type :movie |
key | Key of Object | belongs_to :owner, key: :user |
set_id | ID of Object | set_id :owner_id or ```set_id { |
cache_options | Hash to enable caching and set cache length | cache_options enabled: true, cache_length: 12.hours, race_condition_ttl: 10.seconds |
id_method_name | Set custom method name to get ID of an object (If block is provided for the relationship, id_method_name is invoked on the return value of the block instead of the resource object) |
has_many :locations, id_method_name: :place_ids |
object_method_name | Set custom method name to get related objects | has_many :locations, object_method_name: :places |
record_type | Set custom Object Type for a relationship | belongs_to :owner, record_type: :user |
serializer | Set custom Serializer for a relationship | has_many :actors, serializer: :custom_actor or has_many :actors, serializer: MyApp::Api::V1::ActorSerializer |
polymorphic | Allows different record types for a polymorphic association | has_many :targets, polymorphic: true |
polymorphic | Sets custom record types for each object class in a polymorphic association | has_many :targets, polymorphic: { Person => :person, Group => :group } |
fast_jsonapi
also has builtin Skylight integration. To enable, add the following to an initializer:
require 'fast_jsonapi/instrumentation/skylight'
Skylight relies on ActiveSupport::Notifications
to track these two core methods. If you would like to use these notifications without using Skylight, simply require the instrumentation integration:
require 'fast_jsonapi/instrumentation'
The two instrumented notifcations are supplied by these two constants:
FastJsonapi::ObjectSerializer::SERIALIZABLE_HASH_NOTIFICATION
FastJsonapi::ObjectSerializer::SERIALIZED_JSON_NOTIFICATION
It is also possible to instrument one method without the other by using one of the following require statements:
require 'fast_jsonapi/instrumentation/serializable_hash'
require 'fast_jsonapi/instrumentation/serialized_json'
Same goes for the Skylight integration:
require 'fast_jsonapi/instrumentation/skylight/normalizers/serializable_hash'
require 'fast_jsonapi/instrumentation/skylight/normalizers/serialized_json'
Please see contribution check for more details on contributing
We use RSpec for testing. We have unit tests, functional tests and performance tests. To run tests use the following command:
rspec
To run tests without the performance tests (for quicker test runs):
rspec spec --tag ~performance:true
To run tests only performance tests:
rspec spec --tag performance:true
Join the Netflix Studio Engineering team and help us build gems like this!