Worker Router

A router for Worker Environments such and Cloudflare Workers or Service Workers.

Work In Progress

This router is inspired by previous work such as tiny-request-router and itty-router, but it improves on them by providing better support for middleware, type inference, nested routing, and broader URL matching for use in service workers.

🆓 Type Inference

The goal of Worker Router is to infer types based on usage so that no explicit typing is required for standard use cases. This allows even JavaScript users to benefit from inline documentation and API discoverability. For example,

const router = new WorkersRouter(basics())
  .get('/about', (req, { userAgent }) => ok())
  .get('/login', unsignedCookies(), (req, { userAgent, cookies }) => ok())

In this example your editor can infer the types and documentation for

userAgent, provided by the basics middleware for the entire router
cookies, provided by the unsignedCookies middleware for this route only

🔋 Functional Middleware

Worker Router middlewares are just function that add properties to a generic context object. As such, they can be mixed and matched using standard tools from functional programming.

For convenience, this module provides a combine utility to combine multiple middlewares into one.

const myReusableMW = combine(
  basics(), 
  signedCookies({ secret: 'password123' }), 
  cookieSession({ user: '' })
);

const router = new WorkersRouter()
  .get('/', myReusableMW, () => ok())
  .post('/', combine(myReusableMW, bodyParser()), () => ok())

Note that type inference is maintained when combining middleware!

🪆 Nested Routing

Worker Router supports delegating entire sub routes to another router:

const itemRouter = new WorkerRouter()
  .get('/', (req, { params }) => ok(`Matched "/item/`))
  .get('/:id', (req, { params }) => ok(`Matched "/item/${params.id}`))

const router = new WorkersRouter()
  .get('/', () => ok('Main Page'))
  .use('/item*', itemRouter)

TODO: Provide parent matches to child router...

⚙️ Ready for Service... Worker

Internally, this router uses URLPattern for routing, which allows it match URLs in the broadest sense. For example, the following router, meant to be used in a Service Worker, can handle internal requests as well as intercept calls to external resources:

// file: "sw.js"
const router = new WorkersRouter()
  .get('/', () => ok('Main Page'))
  .get('/about', () => ok('About Page'))
  .external('https://plausible.io/api/*', req => {
    // intercepted
  })

💥 Error Handling Without Even Trying

Worker Router has first class support for error handling. Its main purpose is to let you write your handlers without having to wrap everything inside a massive try {} catch block. Instead, you can define special recover routes that get invoked when something goes wrong.

const router = new WorkersRouter()
  .get('/', () => ok('Main Page'))
  .get('/about', () => { throw Error('bang') })
  .recover('*', (req, { error, response }) => 
    new Response(`Something went wrong: ${error.message}`, response)
  );

✅ Works with Workers

Worker Router comes with out of the box support for a variety of Worker Environments:

To use it in an environment that provides a global fetch event, use

self.addEventListener('fetch', router)

(This works because the router implements the EventListener interface)

To use it with Cloudflare's module workers, use

export default router

(This works because the router implements a fetch method with compatible interface)

To use it with Deno/Deploy's serve function, use

serve(router.serveCallback)