Home

Awesome

Routington

NPM version Build status Test coverage Dependency Status License Downloads Gittip

Routington is a trie-based URL router. Its goal is only to define and match URLs. It does not handle methods, headers, controllers, views, etc., in anyway. It is faster than traditional, linear, regular expression-matching routers, although insignficantly, and scales with the number of routes.

The purpose of this router isn't for performance, but to bring more structure to URL routing. The intention is for you to build a framework on top either in node.js or in the browser.

Implementations:

API

node Node = Routington()

var routington = require('routington')
var router = routington()

router is the root Node in the trie. All nodes will have router as furthest ancestor.

Node

Every node on a tree is an instance of Node. You only construct the root. A node has the following properties:

nodes []Node = router.define(route)

var nodes = router.define('/:identity(page|petition)/:id([0-9a-f]{24})')

Each fragment of the route, delimited by a /, can have the following signature:

You should always name your regular expressions otherwise you can't use the captured value. The regular expression is built using new RegExp('^(' + regex + ')$', 'i'), so you need to escape your string, ie \\w. You can always pre-define names or regular expressions before. For example, I can define:

router.define('/page/:id(\\w{3,30})')

// later, :id will have the same regexp
// so you don't have to repeat yourself
router.define('/page/:id/things')

match {} = router.match(url)

router.define('/page/:id(\\w{3,30})')
var match = router.match('/page/taylorswift')

match, unless null, will be an object with the following properties:

Building a Router on top of Routington

Each URL you define creates a node, and you are free to do whatever you'd like with each node as long you don't overwrite any prototype properties (basically just define, match, and parse). Adding any features to routington shouldn't be necessary.

For example, suppose you want to attach callbacks to a node by extending routington:

router.get('/:id/:controller', function (req, res, next) {
  console.log('do something')
})

You can attach the middleware to a node.GET array:

router.get = function (path, handler) {
  var node = router.define(path)[0]
  node.GET = node.GET || []
  node.GET.push(handler)
}

Now, dispatching is easy:

function dispatcher(req, res, next) {
  var match = router.match(url.parse(req.url).pathname)
  if (!match)
    // this is a 404

  var node = match.node
  var callbacks = node[req.method]
  if (!callbacks)
    // this is a 405

  // execute all the callbacks.
  // async.series won't actually work here,
  // but you get the point.
  async.series(callbacks, next)
}

Properties attached to the node will be exposed on the match. For example, suppose you wanted to label a node:

var node = router.define('/:id/:controller')[0]
node.label = 'controller'

When matched, it will be available via match.node.label:

var match = router.match('/someid/somecontroller')
assert(match.node.label === 'label')

Since reaching into match.node is a little inconvenient and you probably don't want your end users to touch it, you should expose in your dispatcher:

var match = router.match(url.parse(req.url).pathname)

// ...

req.param = match.param
req.label = match.node.label

Browser Support

IE9+