magma

magma is a framework for creating scalable AMQP back ends.

Install Magma

npm install magma

Create a Router

In magma, everything starts with a Router. Routers take in incoming commands and dispatch them to Workers using RabbitMQ. Currently, magma only has one type of Router - the HTTP router. The HTTP router uses express to take incoming HTTP requests and transmit them to a Worker.

var magma  = require('magma');
var router = magma.router();

router.connect(function () {
  router.resource('article');
});

var server = router.listen(3000, function () {
  var host = server.address().address;
  var port = server.address().port;

  console.log('Listening at http://%s:%s', host, port);
});

Resources

The Router above declares one resource: article. In magma, a resource acts as a type of object. Each resource can be created, read, updated, and deleted using HTTP POST, GET, PUT, and DELETE requests respectively. When a resource is declared on the Router, endpoints matching the name of that resource are automatically created. For example:

GET http://.../network

Access Express

The HTTP Router is based around the express framework. You can access the express instance using the Router.app property. This allows you to add helpful features, such as authentication middleware. On a similar note, the Router.listen method is basically a direct proxy to the express listen method.

Create a Worker

The next piece of a magma application is the Worker. Workers are in charge of performing a specific set of tasks related to a resource. While a Worker can handle more than one resource at a time, it's typically best practice to make a new worker for each declared resource. Workers can be spawned and despawned as the workload for the resource increases, and RabbitMQ will automatically handle load balancing between them.

It's important to rememeber that Workers should be—on their own—stateless. Any session or long term storage should be handled using a caching system such as Redis, or a database such as MongoDB. You should write your worker with the assumption that it can be spun up or down at any time.

var magma  = require('magma');
var worker = magma.worker();

worker.connect(function () {
  worker.resource('article', function () {
    this.create = function (req, res) {
      var articleId = req.data.id;                // Access data using `req.data`
      var authHeader = req.headers.authorization; // Access headers using `req.headers`
    };

    this.read = function (req, res) {
      res.success('good job!'); // Send successful responses using `res.success`
    };

    this.update = function (req, res) {
      res.fail('something went wrong!'); // Send failure responses using `res.fail`.
    };

    this.delete = function (req, res) {
      // And so on...
    };
  });
});

When binding a Worker to a Resource, you must implement the create, read, update, and delete methods for that resource. These methods are automatically invoked corresponding to the POST, GET, PUT, and DELETE HTTP requests from the Router.

You can access request data using the req.data object. Query string data as well as POSTed body data will be combined into this key value array.

Remote Procedure Calling

Traditionally it's best to stick with REST-ful API endpoints, which the default router/worker system provides. However, occasionally you need to divert from that path, which is where Magma's RPC feature comes in.

The Router

Start by declaring a router with a custom resource...

var magma  = require('magma');
var router = magma.router();

router.connect(function () {
  router.resource('network', [
    'customAction:get'
  ]);
});

var server = router.listen(3000, funciton () {
  // ...
});

By including a second parameter to the resource method, we have declared a new endpoint on the network resource called customAction. Because we suffixed the endpoint with :get, you will be able to access the custom endpoint like so:

GET http://.../network/customAction

In addition to the :get suffix, you can use :put, :post, and :delete for PUT, POST, and DELETE request respectively. You must include a suffix, there is no default.

The Worker

Now that you have a router defined with your customAction endpoint declared, creating a worker is pretty much business as usual except for one thing: in addition to defining your worker's create, read, update, and delete methods, you will also define a customAction method. The customAction will behave in the same way as the other CRUD methods.

var magma  = require('magma');
var worker = magma.worker();

worker.connect(function () {
  worker.resource('network', function () {
    this.customAction = function (req, res) {
      res.success('customAction');
    };

    this.create = function (req, res) {
      res.success('create');
    };

    this.read = function (req, res) {
      res.success('read');
    };

    this.update = function (req, res) {
      res.success('update');
    };

    this.delete = function (req, res) {
      res.success('delete');
    };
  });
});