Backbone.js + Express.js SPA boilerplate

This project is supposed to be a starter kit helping the community build Single Page Application architectures based on Backbone.js and Express.js frameworks.

A Pragmatic approach is key. The simplest implementation possible, just to show the way, not a finalized application. It covers aspects of front-end design, as well as open API.

You are the one, who can help project to grow. Join now and contribute, your help is much appreciated.

NOTE: It is still in progress. If you would like to contribute, please join the discussion/implementation of currently opened issues.

module.exports = function (app) {
	app.get('/api/emails', function (req, res) {
		res.json({status: 'GET /api/users'});
	});

	app.post('/api/emails', function (req, res) {
		res.json({status: 'POST /api/users'});
	});

	app.put('/api/emails/:id', function (req, res) {
		res.json({status: 'PUT /api/users/' + req.params.id});
	});

	app.del('/api/emails/:id', function (req, res) {
		res.json({status: 'DELETE /api/users/' + req.params.id});
	});
};

To enable API end-point, you should modify app.js file, like

// api endpoints
require('./source/api/emails')(app);
require('./source/api/contacts')(app);
require('./source/api/tasks')(app);

## Authorization and CORS

While you design new open API, authorization is probably one of the most important topics to consider.

There are many ways you can implement authorization. Choices you made will affect how much of API is usable and scalable. Before we jump in, I highly recommend blog post by Vinay Sahni regarding Pragmatic RESTFull API.

Traditional authorization algorithms are typically using some kind of token-based authorization. Means, each user of API is getting registered and special token (pseudo-random looking string) is issued and associated with that user. Token contains user identification + some meta info, encrypted with some simple algorithm as base64. For each request, users sends this token (in HTTP headers of query), server lookups for this token and if it finds match between encrypted userId and token, request treated as authenticated.

Tokens always have to be transported by secured channel as SSL.

The only problem with that approach is that each request requires at least one expensive database call. This could be avoided by applying some cryptography on authorization procedure. One of the approaches is usage of self-signed HMAC tokens. It's easy to implement and it scales very nice.

HMAC based API authorization

The routine could be slitted to such steps:

Client signs ups to API by sending username and password.
Client get registered, password hashed and placed to users collection in DB.
Authorization token is issued.
Client receives token back.
For all further requests client sends token as password in authorization header of HTTP request.
Server validates the token, if it's valid request treated as authenticated.

Note, token validation procedure doesn't require any DB request everything all depends on computation.

HMAC token

Let's take a look how to issue new authorization token. We typically want to reduce token lifetime, in case it's stolen. So, we want to include some meta information about time it's issued as well it's owner. This information is concatenated and HMAC algorithm applied, using servers private key.

After we add same info in open text and encrypt result by base64.

var timespamp = moment();
var message = username + ';' + timespamp.valueOf();
var hmac = crypto.createHmac('sha1', AUTH_SIGN_KEY).update(message).digest('hex');
var token = username + ';' + timespamp.valueOf() + ';' + hmac;
var tokenBase64 = new Buffer(token).toString('base64');

Here, AUTH_SIGN_KEY is private server key and tokenBase64 is the final result, sent back to client.

Authorization

Client stores the token to cookie or localstorage and using it for each API request as part of Basic authentication header.

Request.js example,

request.get({url: url, auth: {user: username, password: token}}, function (err, resp) {
	error = err;
	response = resp;
	done();
});

jQuery example,

$.ajax
({
	type: "GET",
	url: "index1.php",
	dataType: 'json',
	async: false,
	username: username,
	password: token,
	success: function (){
		done()
	}
});

Backbone.sync example,

Backbone.ajax = function() {
	var defaults = arguments[0];
	_.extend(defaults, {
		username: username,
		password: token
	});

	return Backbone.$.ajax.apply(Backbone.$, arguments);
};

Token validation

Now, server receives token back and request need to be authenticated.

Server decodes token base64 and parses out all token information (simple split by ';').
Server computes HMAC signature of received username and timestamp using the same private server key.
It compares it with HMAC received in token.
If HMAC's are different, request is not authorized, 401 reply.
Otherwise, server checks token TTL, if expires , 401 reply.
Otherwise request is authenticated.

If token is compromised or wrong, HMAC guarantees that signatures will never match, except attacker is aware of server private key.

API Authorization implementation

API exposes few methods,

/api/auth/signup
/api/auth/login
/api/auth/validate

Signup, used as initial client registration. Login is called each time, new token have to issued. Validate is used to check token validity (it's used as internal method mostly).

There is source/middleware/auth.js that exposes createToken and validateToken functions. Create token is applied to signup and login api methods, validateToken is applied on every API that requires authorization.

Checkout test/api/auth.specs.js that specifies how authorization works in details, source/api/auth.js for end-point implementation.

CORS

CORS (Cross Origin Resource Sharing) have to be enabled to allow the API to be used from client applications, running on domain different than API deployment domain. For instance, your API can be deployed at https://api.example.com, but application running at https://app.client.com.

In order to allow that, special middleware function created, /source/middleware/cors.js:

function cors() {
	return function (req, res, next) {
		res.header('Access-Control-Allow-Origin', '*');
		res.header('Access-Control-Allow-Methods', 'GET,PUT,POST,DELETE');
		res.header('Access-Control-Allow-Headers', 'X-Requested-With, X-Access-Token, X-Revision, Content-Type');

		next();
	};
}

module.exports = cors;

It have be added during application initialization, like:

	app.use(middleware.cors());

## Backbone.js

Backbone.js is the one of most popular front-end development framework (library). It provides abstractions for models, views, collections and able to handle client-side routing.

Front-end architecture is build on modular structure and relying on AMD to allow build scalable applications.

### RequireJS and CommonJS

RequireJS picked up as asynchronous javascript module loading. RequireJS uses it's own style for defining modules, specifying the dependency as array of strings.

define([
	'/some/dep',
	'another/dep',
	'yet/another/dep',
	'text!./templates/template.html,
	jQuery,
	Backbone'], function(SomeDep, AnotherDep, YetAnotherDep, template, $, Backbone) {
		// module implementation...
	});

With some time spent on Node.js programming, CommonJS style becomes more convenient to use. Fortunately RequireJS has CommonJS style implementation.

define(function (require) {
	// dependencies
	var SomeDep = require('/some/dep');
	var AnotherDep = require('another/dep');

	// export
	return {};
});

### Routing

All routing logic is placed in /core/router.js. There are 3 routes defined in boilerplate.

Each route handler is responsible for starting up new application. Application run function takes ViewManager instance.

### View manager

SPA application typical threat is memory leaks. Memory leaks might appear for a few reasons, one of the most famous reason for Backbone applications are, so called, zombie views.

/core/viewManager.js is responsible for disposing views during switch from one router to another.

Besides of that, it handles transitions during application switch.

### Applications

Application is concept of grouping models, collections, views of unit in one place. The rule is, "one route - one application". Router matches the route, loading the application entry point and passes viewManager (or any other parameters, like id's or query strings) into application.

All applications are apps folder.

app.js is entry point of application and it's responsible for several things:

Fetching initial application data
Instantiating Main View of application

define(function(require) {
	var MainView = require('./views/MainView');

	return {
		run: function (viewManager) {
			var view = new MainView();
			viewManager.show(view);
		}
	};
});

### Main view and subviews

Main view responsible for UI of application. It's quite typically that main view is only instantiating subviews and passing the models/collections further down.

MainView.js keeps track of subviews in this.subviews arrays. Each subview will be closed by ViewManager dispose function.

var MainView = Backbone.View.extend({
	initialize: function () {
		this.subviews = [];
	},

	render: function () {
		var headerView = new HeaderView();
		this.subviews.push(headerView);
		this.$el.append(headerView.render().el);

		var footerView = new FooterView();
		this.subviews.push(footerView);
		this.$el.append(footerView.render().el);

		return this;
	}
});

### Templates

Handlebars is picked up as templating engine, powered by require-handlebars-plugin. Templates are stored on application level in template folder. Handlebars plugin is configured to keep templates in .html files.

View is loading template through !hbs plugin and uses that in render() function.

var HeaderView = Backbone.View.extend({
	template: require('hbs!./../templates/HeaderView'),

	render: function () {
		this.$el.html(this.template({title: 'Backbone SPA boilerplate'}));
		return this;
	}
});

## Transitions

Transitions is a very nice feature for single pages applications. It adds the visual effects of switching from one application to another.

Boilerplate is relying on wonderful animate.css library. core/transition.js is responsible for applying transition style. It's being called from /core/viewManager.js.

Once you decide to have transitions in your app, simply modify master.ejs and add data-transition attribute to application div.

<div class="container">
	<div id="app" class="container" data-transition="fadeOutLeft"></div>
</div>

Checkout the list of available transitions on animate.css page. You can apply anything you want, please note "Out" transition type is suited the best.

## Testing

Testing is key of quality product. Both sides (front and back) have to covered with tests to tackle the complexity.

Execute all tests

To execute all tests, run

$ npm test

### Backbone.js (front-end) tests

TODO.

### Express.js (back-end) tests

TODO.

### Functional (web driver) tests

TODO.

## SEO

TODO.

## Build for production

Modern web applications contain a lot of JavaScript/CSS files. While application is loading all that recourses have to be in-place, so browser issuing HTTP requests to fetch them. As more application grow, as more requests need to be done.. as slower initial loading is. There are two ways of optimization of initial application loading:

concatenate and minify (decrease HTTP request)
gzip content (decrease payload size)

Application could operate in several modes - development, production. In development mode, we don't care about optimizations at all. Even more, we are interested to get not processed source code, to be able to debug easily. In production mode, we have to apply as much effort as possible to decrease initial load time.

### Concatenate and minify

RequireJS comes together with optimization tool, called r.js. It's able to concatenate and minify both JavaScript and CSS code.

To simplify the process, we'll use GruntJS tasks runner. GruntJS is very handy tool, with great community around and very rich contrib library. There is a special task to handle RequireJS optimizations, called grunt-contrib-requirejs.

Gruntfile.js contains all required configuration. To run grunt,

$ grunt

The result of the grunt run is new folder /public/build that contains 2 files: main.css, main.js - concatenated and minified JavaScript and CSS code.

### Gzip content

Besides concatenation, it's important to compress resources. Express.js includes this functionality out of the box, as compress() middleware function.

### Development and production

The configuration distinction goes in app.js file.

app.configure('development', function(){
	app.use(express.errorHandler());							// apply error handler
	app.use(express.static(path.join(__dirname, 'public')));
	app.use(middleware.serveMaster.development());				// apply development mode master page
});

app.configure('production', function(){
	app.use(express.compress());								// apply gzip content
	app.use(express.static(path.join(__dirname, 'public'), { maxAge: oneMonth }));
	app.use(middleware.serveMaster.production());				// apply production mode master page
});

serveMaster.js middleware component is would serve different version of master page, for different mode. In development mode, it would use uncompressed JavaScript and CSS, in production mode, ones that placed in /public/build folder.

### Cache busting

Caching is in general good since it helps to application to be loaded faster, but it could hurt while you re-deploy application. Browsers do not track the actual content of file, so if the content has changed, but URL still the same, browser will ignore that.

Besides, different browsers have different caching strategies. IE for instance, is 'famous' with is aggressive caching.

Cache busting is widely adopted technique. There are some different implementations for that, but one of the most effective is: name your resources in the way, so if content has changed the name of resource would change as well. Basic implementation is to prefix file names with hash computed on file contents.

Boilerplate uses grunt-hashres task for that (currently I'm using my own fork, hope that changes are promoted to main repo soon). That task transforms the grunt-contrib-requirejs output files main.js, main.css into something like main-23cbb34ffaabd22d887abdd67bfe5b2c.js, main-5a09ac388df506a82647f47e3ffd5187.css.

It also produces /source/client/index.js file serveMaster.js uses to render production master page correctly.

Now, everything that either .js or .css content is changed, build would produce new files and they are guaranteed to be loaded by browser again.

### Optimization results

On a left side you see application running in development mode, on a right side in production mode.

optimization results

Even for such small application as 'TheMailer', the benefits are obvious:

Requests: 55 / 4 ~ 14 times fewer.
Payload: 756Kb / 43.4Kb ~ 17 times smaller.
Load time: 898ms / 153ms ~ 6 times faster.

## Deployment

TODO.

## Yeoman generator

TODO.

Legal Info (MIT License)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Name		Name	Last commit message	Last commit date
Latest commit History 112 Commits
public		public
source		source
test		test
tools/client		tools/client
views		views
.bowerrc		.bowerrc
.gitignore		.gitignore
.jshintrc		.jshintrc
Gruntfile.js		Gruntfile.js
README.md		README.md
app.js		app.js
bower.json		bower.json
package.json		package.json
runMocha.js		runMocha.js

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

Backbone.js + Express.js SPA boilerplate

Contents

Master page

HMAC based API authorization

HMAC token

Authorization

Token validation

API Authorization implementation

CORS

Execute all tests

Legal Info (MIT License)

About

Releases

Packages

Languages

s-stude/backbone-express-spa

Folders and files

Latest commit

History

Repository files navigation

Backbone.js + Express.js SPA boilerplate

Contents

Master page

HMAC based API authorization

HMAC token

Authorization

Token validation

API Authorization implementation

CORS

Execute all tests

Legal Info (MIT License)

About

Resources

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages