var Router = require('dumb-router');

var router = Router(); //factory; do not use `new`

router.register('/cats', meow);
router.register('/dogs', woof);

function meow() {
  alert('a cat is speaking');
  return 'meow';
}

function woof() {
  alert('a dog is speaking');
  return 'woof';
}

alert(router.execute('/dogs')[0]); // => woof

Route parameters

router.register('/say/:words', say);

function say(path, params) {
  console.log(params.words);
}

router.execute('/say/hello-world');

Nesting routes

router.register('/users/:userId', setUser, function(router) {
  router.register('/edit', showUserForm);
});

router.execute('/users/5/edit');

Scoping routes

router.register('/food', null, function(router) {
  router.register('/cookies', cookies);
  router.register('/cakes', cakes);
});

Return values

When the router executes, it returns an array, with an entry for each route handler.

Return values for outer handlers come before those of nested routes.
Scoping routes do not add to the array of return values.
undefined is added for handlers that do not return anything.

Path building

I lied about dumb-router only doing path -> handler mapping and execution. Being able to validate routes is useful, and we can do it by exposing a bit of the route matching logic.

router.register('/food' null, function(router) {
  router.register('/gourmet', gourmetFood);
  router.register('/fast', fastFood);
});

router.path('food') //throws error, no handler for "/food"
router.path('food', 'gourmet');       // => '/food/gourmet'
router.path('food', 'fast');          // => '/food/fast'
router.path('food', 'home-cooking');  //throws error, no route defined

About

Some context around the origin of this class:

We used this Router class as part of our base Flux Store class, where each Store had an instance of a Router. We only used the Router to provide a convenient DSL for mapping Store private functions (state mutators) to paths. Then, all such Stores would handle "navigate" Actions from the Dispatcher by delegating to the Router's execute method.

This decentralized routing pattern, where each store defined its own response to navigation actions, arose because we wanted to avoid exposing means of mutating store state.

The "single monolithic router" pattern, common in other frameworks, requires that Stores expose some means of mutating their internal State.

Decentralized routing has some drawbacks: duplicated routes, and the need to be loose when matching routes. For example, if two stores have routers, and one store responds to "/users/" and another store responds to "/users/new", and both Stores need to invoke their handlers when navigating to "/users/new", the router has to consider "/users/new" a match for "/users".