Peeps is a very basic contact management system implemented as an API that follows the JSON API spec.

Other apps will soon be written to demonstrate writing a consumer for this API.

The instructions below were used to create this app.

rails new peeps --skip-javascript

or

rails new peeps -d postgresql --skip-javascript

rake db:create

Add the gem to your Gemfile

gem 'jsonapi-resources'

Then bundle

bundle

Make the following changes to application_controller.rb

class ApplicationController < ActionController::Base
  include JSONAPI::ActsAsResourceController
  
  # Prevent CSRF attacks by raising an exception.
  # For APIs, you may want to use :null_session instead.
  protect_from_forgery with: :null_session
end

OR

class ApplicationController < JSONAPI::ResourceController
  # Prevent CSRF attacks by raising an exception.
  # For APIs, you may want to use :null_session instead.
  protect_from_forgery with: :null_session
end

You can also do this on a per controller basis in your app, if only some controllers will serve the API.

Edit config/environments/development.rb

Eager loading of classes is recommended. The code will work without it, but I think it's the right way to go. See http://blog.plataformatec.com.br/2012/08/eager-loading-for-greater-good/

  # Eager load code on boot so JSONAPI-Resources resources are loaded and processed globally
  config.eager_load = true

config.consider_all_requests_local       = false

This will prevent the server from returning the HTML formatted error messages when an exception happens. Not strictly necessary, but it makes for nicer output when debugging using curl or a client library.

Use the standard rails generator to create a model for Contacts and one for related PhoneNumbers

rails g model Contact name_first:string name_last:string email:string twitter:string

Edit the model

class Contact < ActiveRecord::Base
  has_many :phone_numbers

  ### Validations
  validates :name_first, presence: true
  validates :name_last, presence: true

end

Create the PhoneNumber model

rails g model PhoneNumber contact_id:integer name:string phone_number:string

Edit it

class PhoneNumber < ActiveRecord::Base
  belongs_to :contact
end

rake db:migrate

Use the rails generator to create empty controllers. These will be inherit methods from the ResourceController so they will know how to respond to the standard REST methods.

rails g controller Contacts --skip-assets
rails g controller PhoneNumbers --skip-assets

We need a directory to hold our resources. Let's put in under our app directory

mkdir app/resources

Create a new file for each resource. This must be named in a standard way so it can be found. This should be the single underscored name of the model with _resource.rb appended. For Contacts this will be contact_resource.rb.

Make the two resource files

contact_resource.rb

class ContactResource < JSONAPI::Resource
  attributes :name_first, :name_last, :email, :twitter
  has_many :phone_numbers
end

and phone_number_resource.rb

class PhoneNumberResource < JSONAPI::Resource
  attributes :name, :phone_number
  has_one :contact

  filter :contact
end

Add the routes for the new resources

jsonapi_resources :contacts
jsonapi_resources :phone_numbers

Launch the app

rails server

Create a new contact

curl -i -H "Accept: application/vnd.api+json" -H 'Content-Type:application/vnd.api+json' -X POST -d '{"data": {"type":"contacts", "attributes":{"name-first":"John", "name-last":"Doe", "email":"[email protected]"}}}' http://localhost:3000/contacts

You should get something like this back

HTTP/1.1 201 Created
X-Frame-Options: SAMEORIGIN
X-Xss-Protection: 1; mode=block
X-Content-Type-Options: nosniff
Content-Type: application/vnd.api+json
Etag: W/"809b88231e24ed1f901240f47278700d"
Cache-Control: max-age=0, private, must-revalidate
X-Request-Id: e4a991a3-555b-42ac-af1e-f103a1007edc
X-Runtime: 0.151446
Server: WEBrick/1.3.1 (Ruby/2.2.2/2015-04-13)
Date: Thu, 18 Jun 2015 18:21:21 GMT
Content-Length: 363
Connection: Keep-Alive

{"data":{"id":"1","type":"contacts","links":{"self":"http://localhost:3000/contacts/1"},"attributes":{"name-first":"John","name-last":"Doe","email":"[email protected]","twitter":null},"relationships":{"phone-numbers":{"links":{"self":"http://localhost:3000/contacts/1/relationships/phone-numbers","related":"http://localhost:3000/contacts/1/phone-numbers"}}}}}

You can now create a phone number for this contact

curl -i -H "Accept: application/vnd.api+json" -H 'Content-Type:application/vnd.api+json' -X POST -d '{ "data": { "type": "phone-numbers", "relationships": { "contact": { "data": { "type": "contacts", "id": "1" } } }, "attributes": { "name": "home", "phone-number": "(603) 555-1212" } } }' http://localhost:3000/phone-numbers

And you should get back something like this:

HTTP/1.1 201 Created
X-Frame-Options: SAMEORIGIN
X-Xss-Protection: 1; mode=block
X-Content-Type-Options: nosniff
Content-Type: application/vnd.api+json
Etag: W/"b8d0ce0fd869a38dfb812c5ac1afa94e"
Cache-Control: max-age=0, private, must-revalidate
X-Request-Id: 63920c97-247a-43e7-9fe3-87ede9e84bb5
X-Runtime: 0.018539
Server: WEBrick/1.3.1 (Ruby/2.2.2/2015-04-13)
Date: Thu, 18 Jun 2015 18:22:13 GMT
Content-Length: 363
Connection: Keep-Alive

{"data":{"id":"1","type":"phone-numbers","links":{"self":"http://localhost:3000/phone-numbers/1"},"attributes":{"name":"home","phone-number":"(603) 555-1212"},"relationships":{"contact":{"links":{"self":"http://localhost:3000/phone-numbers/1/relationships/contact","related":"http://localhost:3000/phone-numbers/1/contact"},"data":{"type":"contacts","id":"1"}}}}}

You can now query all one of your contacts

curl -i -H "Accept: application/vnd.api+json" "http://localhost:3000/contacts"

And you get this back:

TTP/1.1 200 OK
X-Frame-Options: SAMEORIGIN
X-Xss-Protection: 1; mode=block
X-Content-Type-Options: nosniff
Content-Type: application/vnd.api+json
Etag: W/"512c3c875409b401c0446945bb40916f"
Cache-Control: max-age=0, private, must-revalidate
X-Request-Id: b324bff8-8196-4c43-80fd-b2fd1f41c565
X-Runtime: 0.004106
Server: WEBrick/1.3.1 (Ruby/2.2.2/2015-04-13)
Date: Thu, 18 Jun 2015 18:23:19 GMT
Content-Length: 365
Connection: Keep-Alive

{"data":[{"id":"1","type":"contacts","links":{"self":"http://localhost:3000/contacts/1"},"attributes":{"name-first":"John","name-last":"Doe","email":"[email protected]","twitter":null},"relationships":{"phone-numbers":{"links":{"self":"http://localhost:3000/contacts/1/relationships/phone-numbers","related":"http://localhost:3000/contacts/1/phone-numbers"}}}}]}

Note that the phone_number id is included in the links, but not the details of the phone number. You can get these by setting an include:

curl -i -H "Accept: application/vnd.api+json" "http://localhost:3000/contacts?include=phone-numbers"

and some fields:

curl -i -H "Accept: application/vnd.api+json" "http://localhost:3000/contacts?include=phone-numbers&fields%5Bcontacts%5D=name-first,name-last&fields%5Bphone-numbers%5D=name"

Test a validation Error

curl -i -H "Accept: application/vnd.api+json" -H 'Content-Type:application/vnd.api+json' -X POST -d '{ "data": { "type": "contacts", "attributes": { "name-first": "John Doe", "email": "[email protected]" } } }' http://localhost:3000/contacts

The earlier responses seem pretty snappy, but they are not really returning a lot of data. In a real world system there will be a lot more data. Lets mock some with the faker gem.

Add the faker gem to your Gemfile

gem 'faker', group: [:development, :test]

And add some seed data using the seeds file

# This file should contain all the record creation needed to seed the database with its default values.
# The data can then be loaded with the rails db:seed command (or created alongside the database with db:setup).
#
# Examples:
#
#   movies = Movie.create([{ name: 'Star Wars' }, { name: 'Lord of the Rings' }])
#   Character.create(name: 'Luke', movie: movies.first)

contacts = []
20000.times do
  contacts << Contact.create({
                               name_first: Faker::Name.first_name,
                               name_last: Faker::Name.last_name,
                               email: Faker::Internet.safe_email,
                               twitter: "@#{Faker::Internet.user_name}"
                             })
end

contacts.each do |contact|
  contact.phone_numbers.create({
                                 name: 'cell',
                                 phone_number: Faker::PhoneNumber.cell_phone
                               })

  contact.phone_numbers.create({
                                 name: 'home',
                                 phone_number: Faker::PhoneNumber.phone_number
                               })
end

Now lets add the seed data (note this may run for a while):

bundle install
rails db:seed

Now if we query our contacts we will get a large (20K contacts) dataset back, and it may run for many seconds (about 8 on my system)

curl -i -H "Accept: application/vnd.api+json" "http://localhost:3000/contacts"

There are some things we can do to work around this. First we should add a config file to our initializers. Add a file named jsonapi_resources.rb to the config/initializers directory and add this:

JSONAPI.configure do |config|
  # Config setting will go here
end

We can enable caching so the next request will not require the system to process all 20K records again.

We first need to turn on caching for the rails portion of the application with the following:

rails dev:cache

To enable caching of JSONAPI responses we need to specify which cache to use (and in version v0.10.x and later that we want all resources cached by default). So add the following to the initializer you created earlier:

JSONAPI.configure do |config|
  config.resource_cache = Rails.cache
  # The following option works in versions v0.10 and later
  #config.default_caching = true
 end 

If using an earlier version than v0.10.x we need to enable caching for each resource type we want the system to cache. Add the following line to the contacts ressource:

class ContactResource < JSONAPI::Resource
  caching
  #...
end

If we restart the application and make the same request it will still take the same amount of time (actually a tiny bit more as the resources are added to the cache). However if we perform the same request the time should drop significantly, going from ~8s to ~1.6s on my system for the same 20K contacts.

We might be able to live with performance of the cached results, but we should plan for the worst case. So we need another solution to keep our responses snappy.

Instead of returning the full result set when the user asks for it, we can break it into smaller pages of data. That way the server never needs to serialize every resource in the system at once.

We can add pagination with a config option in the initializer. Add the following to config/initializers/jsonapi_resources.rb:

JSONAPI.configure do |config|
  config.resource_cache = Rails.cache
  # config.default_caching = true

  # Options are :none, :offset, :paged, or a custom paginator name
  config.default_paginator = :paged # default is :none

  config.default_page_size = 50 # default is 10
  config.maximum_page_size = 100 # default is 20 
end

Restart the system and try the request again:

curl -i -H "Accept: application/vnd.api+json" "http://localhost:3000/contacts"

Now we only get the first 50 contacts back, and the request is much faster (about 80ms). And you will now see a links key with links to get the remaining resources in your set. This should look like this:

{
    "data":[...],
    "links": {
    "first":"http://localhost:3000/contacts?page%5Bnumber%5D=1&page%5Bsize%5D=50",
    "next":"http://localhost:3000/contacts?page%5Bnumber%5D=2&page%5Bsize%5D=50",
    "last":"http://localhost:3000/contacts?page%5Bnumber%5D=401&page%5Bsize%5D=50",
    }
}

This will allow your client to iterate over the next links to fetch the full results set without putting extreme pressure on your server.

The default_page_size setting is used if the request does not specify a size, and the maximum_page_size is used to limit the size the client may request.

Note: The default page sizes are very conservative. There is significant overhead in making many small requests, and tuning the page sizes should be considered essential.