Chrome Dev Summit 2018 is happening now and streaming live on YouTube. Watch now.

Workbox Routing

What is Workbox Routing?

A service worker can intercept network requests for a page. It may respond to the browser with cached content, content from the network or content generated in the service worker.

workbox-routing is a module which makes is easy to "route" these requests to different functions that provide responses.

How Routing is Performed

When a network request causes a service worker fetch event, workbox-routing will attempt to respond to the request using the supplied routes and handlers.

Workbox Routing Diagram

The main things to note from the above is:

  • The type of request is important.
    • By default, Routes are registered for 'GET' requests. If you wish to intercept other types of requests, you’ll need to specify the method.
  • The order of the Route registration is important.
    • If multiple Routes are registered that could handle a request, the Route that is registered first will be used to respond to the Request.

There are a few ways to register a route, you can use callbacks, regular expressions or Route instances each of which we’ll look at next.

Matching and Handling in Routes

A "route" in workbox is nothing more than two functions: a “matching” function to determine if the route should match a request and a “handling” function, which should handle the request and respond with a Response.

Workbox comes with some helpers that’ll perform the matching and handling for you, but if you ever find yourself wanting different behaviour, writing a custom match and handler function is the best option.

A match function will be given a FetchEvent and a URL object and you can match against a request by returning a truthy value. For a simple example, you could match against a specific URL like so:

const matchCb = ({url, event}) => {
  return (url.pathname === '/special/url');
};

Most use cases can be covered by examining / testing either the url or the event.request to match against a Request.

A "handler" will be given the url and event as well and you can determine how to respond, whether it’s from the network, from the cache or generated in the service worker.

const handlerCb = ({url, event, params}) => {
  return fetch(event.request)
  .then((response) => {
    return response.text();
  })
  .then((responseBody) => {
    return new Response(`${responseBody} <!-- Look Ma. Added Content. -->`);
  });
};

Your handler must return a Promise that resolves to a Response. The params value is the value returned by the "match" function. This may be useful if you parsed the URL or request and want to pass values into the “handler” for a matching request.

You can register these callbacks via like so:

workbox.routing.registerRoute(matchCb, handlerCb);

The only limitation is that "match" must synchronously return a truthy value, you can’t perform any asynchronous work. The reason for this is that the Router must synchronously respond to the fetch event or allow falling through to other fetch events.

The "handler" callback should return a Promise that resolves to a Response. Where that Response comes from is up to you, the network, a cache or generated in the service worker.

Normally the "handler" callback would use one of the strategies provided by workbox-strategies like so:

workbox.routing.registerRoute(
  matchCb,
  workbox.strategies.staleWhileRevalidate()
);

In this page we’ll focus on workbox-routing but you can learn more about these strategies on workbox.strategies.

How to Register a Regular Expression Route

A common practice is to use a regular expression instead of a "match" callback. Workbox makes this easy to implement like so:

workbox.routing.registerRoute(
  new RegExp('/styles/.*\.css'),
  handlerCb
);

For requests from the same origin, this regular expression will match as long as the request’s URL matches the regular expression.

  • https://example.com/styles/main.css
  • https://example.com/styles/nested/file.css
  • https://example.com/nested/styles/directory.css

However, for cross-origin requests, regular expressions must match the beginning of the URL. The reason for this is that it’s unlikely that with a regular expression new RegExp('/styles/.*\.css') you intended to match a third-parties CSS files.

  • https://cdn.third-party-site.com/styles/main.css
  • https://cdn.third-party-site.com/styles/nested/file.css
  • https://cdn.third-party-site.com/nested/styles/directory.css

If you did want this behaviour, you just need to ensure that regular expression matches the beginning of the URL. If we wanted to match the requests for https://cdn.third-party-site.com we could use the regular expression new RegExp('https://cdn.third-party-site.com.*/styles/.*\.css').

  • https://cdn.third-party-site.com/styles/main.css
  • https://cdn.third-party-site.com/styles/nested/file.css
  • https://cdn.third-party-site.com/nested/styles/directory.css

If you wanted to match both local and third parties you can use a wildcard at the start of your regular expression, but this should be done with caution to ensure it doesn’t cause unexpected behaviors in you web app.

How to Register a Navigation Route

If your site is a single page app, you can use a NavigationRoute to return a specific response for all navigation requests.

workbox.routing.registerNavigationRoute('/single-page-app.html');

Whenever a user goes to your site in the browser, the request for the page will be a navigation request and will be served the cached page '/single-page-app.html'. (Note: You should have the page cached via workbox-precaching or through your own installation step).

By default this will respond to all navigation requests, if you want to restrict it to respond to a subset of URL’s you can use the whitelist and blacklist options to restrict which pages will match this route.

workbox.routing.registerNavigationRoute('/single-page-app.html', {
  whitelist: [
    new RegExp('/blog/')
  ],
  blacklist: [
    new RegExp('/blog/restricted/'),
  ]
});

The only thing to note is that the blacklist will win if a URL is in both the whitelist and blacklist.

Set a Default Handler

If you want to supply a "handler" for requests that don’t match a route, you can set a default handler.

workbox.routing.setDefaultHandler(({url, event, params}) => {
  ...
});

Set a Catch Handler

In the case of any of your routes throwing an error, you can capture and degrade gracefully by setting a catch handler.

workbox.routing.setCatchHandler(({url, event, params}) => {
  ...
});

Defining a Route for Non-GET Requests

All routes by default are assumed to be for 'GET' requests.

If you would like to route other requests, like a 'POST' request, you need to define the method when register the request, like so:

workbox.routing.registerRoute(
  matchCb,
  handlerCb,
  'POST'
);
workbox.routing.registerRoute(
  new RegExp('/api/.*\.json'),
  handlerCb,
  'POST'
);

Router Logging

You should be able to determine the flow of a request using the logs from workbox-routing which will highlight which URLs are being processed through Workbox.

Routing Logs

If you need more verbose information you can set the log level to debug to view logs on requests not handled by the Router. See our debugging guide for more info on setting the log level.

Debug and Log Routing Messages

Advanced Usage

If you want to have more control over when the Workbox Router is given requests you can create your own Router instance and call it’s handleRequest() method whenever you want to use the router to respond to a request.

const router = new DefaultRouter();
self.addEventListener('fetch', (event) => {
  const responsePromise = router.handleRequest(event);
  if (responsePromise) {
    // Router found a route to handle the request
    event.respondWith(responsePromise);
  } else {
    // No route found to handle the request
  }
});

When using the Router directly you will also need to use the Route class, or any of the extending classes to register routes.

const router = new DefaultRouter();
router.registerRoute(new Route(matchCb, handlerCb));
router.registerRoute(new RegExpRoute(new RegExp(...), handlerCb));
router.registerRoute(new NavigationRoute(handlerCb));