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

Last updated

Hosted marketplace texts with Asset Delivery API

This article describes how hosted marketplace texts work in the Sharetribe Web Template.

Table of Contents

Marketplace texts can be managed both in the built-in marketplace text files and in Sharetribe Console. This article describes how the template uses the hosted marketplace texts and merges them with the built-in marketplace texts.

Hosted marketplace texts

Hosted marketplace texts refer to the texts that the marketplace operator can edit in Sharetribe Console. The client app then needs to fetch it using the Asset Delivery API.

It is good to note that even if the operator adds some hosted marketplace texts using Sharetribe Console, the template still needs to have a built-in marketplace text file for the marketplace text keys that do not have a value in the hosted asset. That way, the UI can still render something meaningful for the parts of the page that the operator has not modified.

The template specifies the path to the hosted marketplace texts as part of the app-wide configuration in config/configDefault.js. These hosted marketplace texts live in a file called content/translations.js, since language-specific marketplace text files make it fairly easy to translate the template to languages other than the default English.

// CDN assets for the app. Configurable through Sharetribe Console.
// Currently, only translation.json is available.
const appCdnAssets = {
  translations: 'content/translations.json',
};

In addition, the template has a global Redux file (src/ducks/hostedAssets.duck.js), which exports a Redux Thunk function called fetchAppAssets. This is the function that actually makes the calls to the Asset Delivery API.

There are two ways to fetch marketplace text assets using Asset Delivery API: by version or by alias.

Fetching marketplace texts by version

All assets are identifiable by their version, and versions are immutable. Therefore if you fetch assets by version, they can be cached for an extended period of time. Read more about caching assets. When fetching marketplace texts by version, the result is cached for an extended period of time, which helps to avoid unnecessary data loading. Since Asset Delivery API sets Cache-Control header for these responses, the browser knows to cache these responses on its own.

Hosted assets are versioned as a whole asset tree - a bit similar to how Git works. Individual asset files might have not changed when the whole version has changed and this might cause HTTP redirects. Since the template uses Sharetribe SDK, the response always contains the data for the requested asset, even if the asset has not changed in the specified version.

sdk.assetByVersion({
  path: 'content/translations.json',
  version: '<some-hash-string>',
});

Fetching marketplace texts by alias

In addition to fetching assets by version, you can fetch them by a specific alias instead of a version. Currently, the marketplace text asset can be fetched with the alias latest, which returns the most recently updated version of the marketplace text file. The response also contains the version information for the most recent asset, so that subsequent fetches can be done based on asset version.

When fetching by alias, the cache time is a few seconds for the Dev and Test environments and up to 5 minutes for the Live environment. In other words, it can take up to 5 minutes for marketplace text updates to be visible in a Live environment. These cache times are subject to change.

sdk.assetByAlias({
  path: 'content/translations.json',
  alias: 'latest',
});

How production build works with hosted marketplace texts

This setup is in use if you run yarn start in your host environment or yarn run dev-server on your local machine.

  1. Export fetchAppAssets from src/index.js to make it available for the server.

  2. SSR: initialize the store

    └── server
        └── dataLoader.js
    let translations = {};
    const store = configureStore({}, sdk);
    return store.dispatch(fetchAppAssets(config.appCdnAssets));
  3. SSR: fetchAppAssets thunk fetches the latest version of content/translations.json asset by using latest alias with sdk.assetByAlias

    • This ensures that the server-side rendering has the most recent version of asset to render the page
  4. SSR: make a loadData call if the route specifies it and

      .then(fetchedAssets => {
        translations = fetchedAssets?.translations?.data || {};
        return Promise.all(dataLoadingCalls);
    })
  5. SSR: asset version is saved to the store and passed to the browser through the preloadedState.

    • Server-side rendering must match with client-side rendering when browser hydrates the server-side rendered content.
    • To ensure that client-side rendering has the same version of marketplace texts, the asset version is passed (through preloaded state) to front end.
    • Note: The template does not pass the marketplace text asset itself from the server to browser. The reason is that if browser fetches the versioned asset file directly, it can leverage browser's cache. So, every page load, after the initial one, will use the translation.json file from the browser's local cache.
      .then(() => {
        return { preloadedState: store.getState(), translations };
      })
  6. SSR: call renderApp function, which renders the <ServerApp> in src/app.js

    • Hosted marketplace text file from Asset Delivery API is passed as props to ServerApp.
    └── server
        └── renderer.js
    // Render the app with given route, preloaded state, hosted microcopy.
    const { head, body } = renderApp(
      requestUrl,
      context,
      preloadedState,
      translations,
      collectWebChunks
    );
    └── src
        └── app.js
    <ServerApp
      url={url}
      context={serverContext}
      helmetContext={helmetContext}
      store={store}
      hostedTranslations={hostedTranslations}
    />
  • Hosted marketplace texts are then merged with the default marketplace texts in ServerApp component.
    export const ServerApp = props => {
      const { url, context, helmetContext, store, hostedTranslations = {}, hostedConfig = {} } = props;
      const appConfig = mergeConfig(hostedConfig, defaultConfig);
      HelmetProvider.canUseDOM = false;
      return (
        <Configurations appConfig={appConfig}>
          <IntlProvider
            locale={appConfig.localization.locale}
            messages={{ ...localeMessages, ...hostedTranslations }}
            textComponent="span"
          >
      {/* etc. */}
  1. Browser: initialize the store with a preloaded state
    • The asset version that the SSR used, is included in that preloaded state
  2. Browser: fetchAppAssets thunk fetch assets using asset version: calls sdk.assetByVersion.
  3. Browser: make loadData call
  4. Browser: hydrate the and pass marketplace texts as props.
  • Hosted marketplace texts are merged with default marketplace texts in ClientApp component.

    <ClientApp store={store} hostedTranslations={translations} hostedConfig={hostedConfig} />,
    

How development build works with hosted marketplace texts

This is setup is in use if you run yarn run dev on local machine.

  1. Browser: initialize the store
  2. Browser: fetchAppAssets thunk fetches assets using latest alias by calling sdk.assetByAlias
    • Because SSR is not available to fetch the latest version of the asset files, this call needs to be made from browser.
  3. Browser: make loadData call
  4. Browser: render the <ClientApp> and pass marketplace texts as props.

Read more

If you want to read more, here are some pointers: