Warning
You are viewing the technical documentation for Sharetribe. If you are looking for our no-code documentation, see our new help center.

Last updated

Routing

This article explains how routing works in the Sharetribe Web Template

Table of Contents

The Sharetribe Web Template uses React Router for creating routes to different pages. React Router is a collection of navigational components that allow single-page apps to create routing as a part of the normal rendering flow of the React app. Instead of defining on the server what gets rendered when a user goes to URL "somemarketplace.com/about", we just catch all the path combinations and let the app define what page gets rendered.

React Router setup

Route configuration

The template's routing setup is simple. There is just one file to check before you link to existing routes or start creating new routes to static pages: routeConfiguration.js.

This page imports all the page-level components dynamically using Loadable Components. In addition, there's a configuration that specifies all the pages that are currently used within the template:

└── src
    └── routing
        ├── routeConfiguration.js
        └── Routes.js
const AboutPage = loadable(() =>
  import(
    /* webpackChunkName: "AboutPage" */ './containers/AboutPage/AboutPage'
  )
);
const AuthenticationPage = loadable(() =>
  import(
    /* webpackChunkName: "AuthenticationPage" */ './containers/AuthenticationPage/AuthenticationPage'
  )
);
// etc..

// Our routes are exact by default.
// See behaviour from Routes.js where Route is created.
const routeConfiguration = () => {
  return [
    {
      path: '/about',
      name: 'AboutPage',
      component: AboutPage,
    },
    {
      path: '/login',
      name: 'LoginPage',
      component: AuthenticationPage,
      extraProps: { tab: 'login' },
    },
    {
      path: '/signup',
      name: 'SignupPage',
      component: AuthenticationPage,
      extraProps: { tab: 'signup' },
    },
    //...
  ];
};

export default routeConfiguration;

In the code above, path /login renders the AuthenticationPage component with prop tab set to 'login'. In addition, this route configuration has the name: 'LoginPage'.

There are a couple of extra configurations you can set. For example /listings path leads to a page that lists all the listings provided by the current user:

{
  path: '/listings',
  name: 'ManageListingsPage',
  auth: true,
  authPage: 'LoginPage', // the default is 'SingupPage'
  component: ManageListingsPage,
  loadData: pageDataLoadingAPI.ManageListingsPage.loadData,
},

Here we have set this route to be available only for the authenticated users (auth: true) because we need to know whose listings we should fetch. If a user is unauthenticated, they are redirected to LoginPage (authPage: 'LoginPage') before they can see the content of the "ManageListingsPage" route.

There is also a loadData function defined. It is a special function that gets called if a page needs to fetch more data (e.g. from the Marketplace API) after redirecting to that route. The loadData function is explained in more detail in the Loading data section below.

In addition to these configurations, there is also a setInitialValues function that can be defined and passed to a route:

{
  path: '/l/:slug/:id/checkout',
  name: 'CheckoutPage',
  auth: true,
  component: CheckoutPage,
  setInitialValues: pageDataLoadingAPI.CheckoutPage.setInitialValues,
},

This function gets called when some page wants to pass forward some extra data before redirecting a user to that page. For example, we could ask booking dates on ListingPage and initialize CheckoutPage state with that data before a customer is redirected to CheckoutPage.

Both loadData and setInitialValues functions are part of Redux data flow. They are defined in page-specific SomePage.duck.js files and exported through src/containers/pageDataLoadingAPI.js.

How the Sharetribe Web Template renders a route with routeConfiguration.js

The route configuration is used in src/app.js. For example, ClientApp defines BrowserRouter and gives it a child component (Routes) that gets the configuration as routes property.

Here's a simplified app.js code that renders the client-side app:

import { BrowserRouter } from 'react-router-dom';
import Routes from './Routes';
import routeConfiguration from './routeConfiguration';
//...
export const ClientApp = props => {
  return (
    <BrowserRouter>
      <Routes routes={routeConfiguration()} />
    </BrowserRouter>
  );
};

Routes.js renders the navigational Route components. Switch component renders the first Route that matches the location.

import { Switch, Route } from 'react-router-dom';
//...

const Routes = (props, context) => {
  //...
  return (
    <Switch>
      {routes.map(toRouteComponent)}
      <Route component={NotFoundPage} />
    </Switch>
  );

Inside Routes.js, we also have a component called RouteComponentRenderer, which has four important jobs:

  • Calling loadData function, if those have been defined in src/routeConfiguration.js. This is an asynchronous call, a page needs to define what gets rendered before data is complete.
  • Reset scroll position after location change.
  • Dispatch location changed actions to Redux store. This makes it possible for analytics Redux middleware to listen to location changes. For more information, see the Enable analytics guide.
  • Rendering of the page-level component that the Route is connected through the configuration. Those page-level components are Loadable Components. When a page is rendered for the first time, the code-chunk for that page needs to be fetched first.

Linking

Linking needs special handling in a single page application (SPA). Using HTML <a> tags will cause the browser to redirect the user to the given "href" location. That will cause all resources to be fetched again, which is a slow and unnecessary step for a SPA. Instead, we just need to tell our router to render a different page by adding or modifying the location through the browser's history API.

React Router exports a couple of navigational components (e.g. <Link to="/about">About</Link>) that could be used for linking to different internal paths. Since the Sharetribe Web Template is meant to be a starting point for customization, we want all the paths to be customizable too. That means that we can not use paths directly when redirecting a user to another Route. For example, a marketplace for German customers might want to customize the LoginPage path to be /anmelden instead of /login - and that would mean that all the Links to it would need to be updated.

This is the reason why we have created names for different routes in src/routeConfiguration.js. We have a component called <NamedLink name="LoginPage" /> and its name property creates a link to the correct Route even if the path is changed in routeConfiguration.js. Needless to say that those names should only be used for internal route mapping.

Here is a more complex example of NamedLink:

// Link to LoginPage:
<NamedLink name="LoginPage" />log in</NamedLink>

// Link to ListingPage with path `l/<listing-uuid>/<listing-title-as-url-slug>/`:
<NamedLink name="ListingPage" params={{ id: '<listing-uuid>', slug: '<listing-title-as-url-slug>' }}>some listing</NamedLink>

// Link to SearchPage with query parameter: bounds
<NamedLink name="SearchPage" to={{ search: '?bounds=60.53,22.38,60.33,22.06' }}>Turku city</NamedLink>

NamedLink is widely used in the template, but there are some cases when we have made a redirection to another page if some data is missing (e.g. CheckoutPage redirects to ListingPage, if some data is missing or it is old). This can be done by rendering a component called NamedRedirect, which is a similar wrapper for the Redirect component.

There's also a component for external links. The reason why it exists is that there's a security issue that can be exploited when a site is linking to external resources. ExternalLink component has some safety measures to prevent those. We recommend that all the external links are created using ExternalLink component instead of directly writing <a> anchors.

// Bad pattern: <a href="externalsite.com">External site</a>
// Recommended pattern:
<ExternalLink href="externalsite.com">External site</ExternalLink>

Loading data

If a page component needs to fetch data, it can be done as a part of navigation. A page-level component has a related modular Redux file with a naming pattern: PageName.duck.js. To connect the data loading with navigation, there needs to be an exported function called loadData in that file. That function returns a Promise, which is resolved when all the asynchronous Redux Thunk calls are completed.

For example, here's a bit simplified version of loadData function on ListingPage:

export const loadData = (params, search) => dispatch => {
  const listingId = new UUID(params.id);

  return Promise.all([
    dispatch(showListing(listingId)), // fetch listing data
    dispatch(fetchTimeSlots(listingId)), // fetch timeslots for booking calendar
    dispatch(fetchReviews(listingId)), // fetch reviews related to this listing
  ]);
};

The loadData function needs to be separately mapped in routeConfiguration.js. To do that, the data loading functions are collected into pageDataLoadingAPI.js file.

└── src
    └── containers
        └── pageDataLoadingAPI.js

Loading the code that renders a new page

The template uses route-based code splitting. Different pages are split away from the main code bundle and those page-specific code chunks are loaded separately when the user navigates to a new page for the first time.

This means that there might be a fast flickering of a blank page when navigation happens for the first time to a new page. To remedy that situation, the template forces the page-chunks to be preloaded when the mouse is over NamedLink. In addition, Form and Button components can have a property enforcePagePreloadFor="SearchPage". That way the specified chunk is loaded before the user has actually clicked the button or executed form submit.

Read more about code-splitting.

Analytics

It is possible to track page views to gather information about navigation behaviour. Tracking is tied to routing through Routes.js where RouteRendererComponent dispatches LOCATION_CHANGED actions. These actions are handled by a global reducer (routing.duck.js), but more importantly, analytics.js (a Redux middleware) listens to these changes and sends tracking events to configured services.

└── src
    ├── routing
    |  └── Routes.js
    ├──analytics
    |  └── analytics.js
    └── ducks
        └── routing.duck.js

For more information, see the Enable analytics guide.

A brief introduction to SSR

Routing configuration is one of the key files to render any page on the server without duplicating routing logic. We just need to fetch data if loadData is defined on page component and then use ReactDOMServer.renderToString to render the app to string (requested URL is a parameter for this render function).

So, instead of having something like this on the Express server:

app.get('/about', handleAbout);

We basically catch every path call using * on server/index.js:

app.get('*', (req, res) => {

and then we ask our React app to

  1. load data based on current URL (and return this preloaded state from Redux store)
  2. render the correct page with this preloaded state (renderer also attaches preloadedState to HTML-string to hydrate the app on the client-side)
  3. send rendered HTML string as a response to the client browser
dataLoader
  .loadData(req.url, sdk /* other params */)
  .then(preloadedState => {
    const html = renderer.render(req.url /* and other params */);
    //...
    res.send(html);
  });