rohitnick/cangaroo

Connect Any App to Any Service

Cangaroo

Cangaroo helps developers integrating their apps with any service. It is a Rails Engine that can be installed on any Rails application and is used as a connection hub from one or multiple applications and external services.

[TODO] add self-explanatory image here.

When/Why you should use Cangaroo?

Cangaroo allows to move the logic related to external services connection and synchronization between multiple applications.

This logic will reside in a shared area without the need to replicate it for any system component. It's especially useful when working with larger system formed by multiple applications that need to send messages to each other and to external services.

This kind of configuration allows:

• don't overload your application resources allowing to scale the number or requests that can be handled;
• in case of errors or downtimes of any system component, Cangaroo stores each message and retry sending it until it's successfully delivered;
• use and contribute to a lot of integrations that has already be done by the community.

Integrations

Cangaroo integrations are pieces of code that allow interacting with external services via API.

An usual flow is:

1. Cangaroo receive some data from an application;
2. Data is send to one or more integrations;
3. Integrations convert data to be compatible with an external service API;
4. Integrations send converted data to the external service.

Cangaroo is born with built-in Wombat extensions compatibility, so you can use any extensions that you can find here.

The whole story

Some time ago Spree decided to shut down Wombat and to release its closed source code to customers only, so we had to decide how to go ahead, and the alternative were:

1. hosting Wombat by ourselves for some of our clients
2. starting a new open source project with an API compliant with Wombat.

We believe in open source so we chose the latter.

The idea is a bit different from Wombat, the goal of this project is to provide a backwards compatible API and give the developers the freedom to change and customize it.

At least for the first release we won't have an admin interface, we believe developers prefer code and using Rails console directly.

We hope this project can help to make the migration from Wombat easier and we believe the Spree/Solidus community will help to make it better.

Dependencies

• rails (>= 4.2.4)
• postgresql

Installation

Cangaroo is a Rails Engine so as usual you can install it by using Bundler by adding it to your application's Gemfile:

  # Gemfile
  gem 'cangaroo'

And then executing:

$ bundle

Install the needed migration with:

$ bin/rake cangaroo:install:migrations

And then executing:

$ bin/rake db:migrate

Now mount the engine into your routes.rb file:

  # routes.rb
  ...
  mount Cangaroo::Engine => "/cangaroo"
  ...

Usage

First create a connection:

  Cangaroo::Connection.create(
    name: 'mystore',
    url: 'http://www.mystore.com',
    key: 'puniethahquoe5aisefoh9ci0Shuaniemei6jahx',
    token: 'ahsh8phuezu3xuhohs6kai5vaB1tae0wiy1shohp'
  )

then create a cangaroo job:

module Cangaroo
  class ShipmentJob < Cangaroo::Job
    connection :mystore
    path '/update_shipment'

    def perform?
      type == 'shipments' &&
      payload['status'] == 'shipped'
    end
  end
end

and add this job to the Rails.configuration.cangaroo.jobs:

  # config/initializers/cangaroo.rb

  Rails.configuration.cangaroo.jobs = [Cangaroo::ShipmentJob]

How it works

Cangaroo provides a Push API where you can send your data. After data has been received, Cangaroo sends data to integrations and webhooks based on your business logic.

This is the detailed flow:

• Cangaroo receives the data
• If some error like "wrong key and token" or "malformed json" is raised Cangaroo returns an HTTP status code based on the error type:
  • 406 for wrong request
  • 401 for Unauthorized
  • 500 for wrong json schema
  • 500 for Cangaroo internal errors
• If there are no errors, for each object in the json body, Cangaroo checks what jobs must be enqueued by calling the #perform? method. Each job returning true to #perform? will be enqueued.

Push API

Cangaroo has just a single endpoint where you can push your data, based on where Cangaroo::Engine is mounted, it will be reachable under the /endpoint path. For example, if the Cangaroo::Engine is mounted under /cangaroo the Push API path will be /cangaroo/endpoint.

When you push to the endpoint the HTTP Request must respect this conventions:

• It must be a POST request
• It must be an application/json request so you have to set the Content-Type header to application/json
• The request must have the X-Hub-Store and X-Hub-Access-Token headers set to a value that exists in the Cangaroo::Connection model (to learn more refer to the Connection documentation below)
• The request body must be a well formatted json.

The json body contains data that will be processed by Cangaroo, the following is an example of an order that will be processed on Cangaroo:

{
  "orders": [
    {
      "id": "O154085346",
      "status": "complete",
      "email": "[email protected]"
    }
  ]
}

The root objects of the json body must contain an array with the objects that Cangaroo needs to process. The only required field for the objects contained in the arrays will be the id key. Push API also supports multiple objects so a request with the following body:

  {
     "orders":[
        {
           "id":"O154085346172",
           "state":"cart"
        },
        {
           "id":"O154085343224",
           "state":"payed"
        }
     ],
     "shipments":[
        {
           "id":"S53454325",
           "state":"shipped"
        },
        {
           "id":"S53565543",
           "state":"waiting"
        }
     ]
  }

will create 2 orders and 2 shipments.

When Cangaroo receives the request it responds with a 200(OK) HTTP status code and the response body will contain numbers of the objects in the payload, for example for the previous request the response will be:

  {
    "orders": 2,
    "shipments": 2
  }

if something goes wrong Cangaroo responds with an HTTP error code with an error message in the body, for example:

  {
    "error": "The property '#/orders/0' did not contain a required property of 'id' in schema"
  }

Connection

Connection are services that can send and receive data from Cangaroo. Each connection must have these fields:

• name - (required, String) A generic name for this connection
• url - (required, String) The url where Cangaroo pushes the data
• key - (required, String) It's used for authentication (used to check the request's 'X-Hub-Store' header)
• token - (required, String) It's used for authentication (used to check the request's 'X-Hub-Access-Token' header)
• basic_auth - (optional, Boolean) Defaults to false. If you would like to use HTTP basic auth in your integration instead of Wombat's key + token. Basic auth is handled Stripe-style, without a username using key as your password.
• parameters - (optional, Hash) Used as parameters when Cangaroo makes a request to this connection

For now we don't have a Web GUI so you have to create the connection on your own by running the code somewhere on your server, for example from the Rails console:

  Cangaroo::Connection.create(
    name: 'mystore',
    url: 'http://www.mystore.com',
    key: 'puniethahquoe5aisefoh9ci0Shuaniemei6jahx',
    token: 'ahsh8phuezu3xuhohs6kai5vaB1tae0wiy1shohp',
    parameters: {
      'channel': 'mysubstore'
    }
  )

Cangaroo Jobs

Jobs are where the payload is pushed to the configured connection.

To allow a job to be executed add it to the Rails.configuration.cangaroo.jobs configuration, for example in an initializer:

  # config/initializers/cangaroo.rb

  Rails.configuration.cangaroo.jobs = [Cangaroo::AddOrderJob, Cangaroo::UpdateShipmentJob]

The Cangaroo::Job class inherits from ActiveJob::Base, so you can use any 3rd-party queuing library supported by ActiveJob. When the job is performed Cangaroo makes a POST request to the connection with the configured path and build the json body with the result of the #transform instance method merged with this attributes:

• request_id - is the job_id coming from ActiveJob::Base
• parameters - are the parameters configured by the parameters class method

You can use the following Cangaroo::Job class methods to configure the job's behaivor:

• connection - is the connection name (see connection for more info)
• path - this path will be appended to your connection.url
• parameters - these parameters will be merged with connection.parameters, they will be added to the json body.

it also has a #perform? instance method that must be implemented. This method must return true or false as Cangaroo will use it to understand if the job must be performed. Inside the #perform? method you'll be able to access the source_connection, type and payload instance attributes.

The #transform instance method can be overridden to customize the json body request, it will have the source_connection, type and payload variables (like the #perform? method) and must return an Hash.

The following is an example of a Cangaroo::Job:

  module Cangaroo
    class ShipmentJob < Cangaroo::Job
      connection :mystore
      path '/update_shipment'
      parameters({ timestamp: Time.now })

      def transform
        payload = super
        payload['shipment']['updated_at'] = Time.now
        payload
      end

      def perform?
        type == 'shipments' &&
        payload['status'] == 'shipped'
      end
    end
  end

Suppose that the mystore connection has a url set to "http://mystore.com" an the payload is something like:

  { "id": "S123", "status": "shipped" }

It will do a POST request to http://mystore.com/update_shipment with this json body:

{
  "request_id": "088e29b0ab0079560dea5d3e5aeb2f7868af661e",
  "parameters": {
    "timestamp": "2015-11-04 14:14:30 +0100"
  },
  "shipment": {
    "id": "S123",
    "status": "shipped"
  }
}

This project rocks and uses MIT-LICENSE.

Tests

Tests are written using rspec. bundle exec rake will run the test suite. If you create an additional migration, you'll need to rm -Rf spec/dummy to ensure the test database schema contains your latest changes.