This is a Next.js project bootstrapped with create-next-app.
- copy .env.example to .env.local
- The only necessary variable is the
NEXT_PUBLIC_DRUPAL_BASE_URL. Set this to your local drupal installation url.
You can configure your Drupal environment to use this as a "preview".
- In the Drupal environment, go to
/admin/config/services/next/sites/addpage. - Enter a label of your choice.
- "Base URL" will be
http://localhost:3000. - "Preview URL" will be
http://localhost:3000/api/draft. - "Preview Secret" can be any string of your choice. This should match the
DRUPAL_PREVIEW_SECRETenvironment variable. - "Revalidate URL"
http://localhost:3000/api/revalidate. Only necessary to test cache invalidations in preview mode. - "Revalidate secret" will be any string of your choice. This should match
DRUPAL_REVALIDATE_SECRETenvironment variable. . Only necessary to test cache invalidations in preview mode. - To test authenticated "Draft Mode" navigate to
/admin/config/services/consumer. At least 1 "Consumer" should already exist.- Edit the consumer
- The "Client ID" can be any string of your choice. It should match the
DRUPAL_DRAFT_CLIENTenvironment variable. - The "New Secret" can be any string of your choice. It should match the
DRUPAL_DRAFT_SECRETenvironment variable. - Choose an appropriate "User", like any "Site Manager"
- For "Scopes" select "Site Manager" and "Decoupled Site User"
# Install Dependencies
yarn install
# Run dev server
yarn dev
# Or run preview server
yarn previewOpen http://localhost:3000 with your browser to see the result.
You can start editing the page by modifying app/page.tsx. The page auto-updates as you edit the file.
This project uses next/font to automatically optimize and load Inter, a custom Google Font.
Story book is a great way to edit components without the need for any Drupal connection. All stories and setup are saved in the ./.storybook directory.
yarn storybookThis will open a new browser window to http://localhost:6006.
View more Storybook documentation.
This project uses both typescript checks and ESLinting. These are run on CI services, but not on production environments since the dev dependencies are not installed on production.
yarn lintThis project makes use of both JSON API and GraphQL API endpoints from the Drupal environment. When a user is in "Draft
Mode", the APIs will use the DRUPAL_DRAFT_CLIENT & DRUPAL_DRAFT_SECRET environment variables to fetch an OAuth token.
This token allows either API to fetch authenticated only data. But while in "draft mode", the pages will be built at
request time. "Draft mode" should only be used for previewing content when a user is editing. "Draft mode" is only enabled
when a user hits the /api/draft route from the Drupal environment. It establishes a cookie
that is then used for subsequent page requests. Note that while in "draft mode", every page load will request fresh data
from the CMS system. This can have negative performance impacts on both platforms.
The JSON API is used for data points that are more simple and don't require very complex data such as paragraph entities. Things like the config pages and the main menu are fetched from JSON API. These APIs also use GET methods. This way they can be easily cached by Drupal/Varnish/CDN services and result in faster data transfer.
JSON API functions are found in the ./src/lib/drupal directory.
GraphQL endpoint /graphql accepts POST methods only. GraphQL allows us to create very nested queries using unions. We
can easily fetch every single piece of information in a single request to build out the entire page, except views. Views
are fetched separately to allow us to make them more dynamic in the future and also to avoid some unwanted errors that
come from the first render in Drupal.
GraphQL types and fetch methods are generated automatically using yarn graphql. If a content type, field, vocabulary,
paragraph type, etc. are created/edited/deleted in the Drupal environment, the queries in ./src/lib/gql
will need to be updated. Most of the changes can be implemented in the fragments.drupal.gql
file. To make it easy, Drupal provides fragments you can copy as a starting point. Navigate to /admin/config/graphql_compose/fragments
to view those fragments. Once the fragments and/or queries have been modified, simply run yarn graphql to rebuild the
typescript types and fetcher queries.
If a field is added in the Drupal environment that is "required", that field must be populated for each entity. GraphQL is strict and will throw an error if you include that field in a query, but the data is null. To solve this, either populate the data in Drupal or make the field optional.
Next.js caches data fetches pretty heavy. On top of that, in production builds, the data and pages are build and cached.
If you experience any issues during development, delete the .next directory and restart your local server.
In the layout and pages, we set the revalidate variable to false. This caches the page and layout build indefinitely.
Layouts and page caches are treated separately and can be invalidated independently of each other, while also allowing
specific parts of each to be invalidated. A route handler is provided that allows the CMS system to invalidate
appropriate areas of the site. Making a GET request to /api/revalidate?secret=[secret]&slug=/[slug] with the correct
parameters will accomplish this invalidation. Passing a slug in the form /tags/foo:bar will invalidate the cache tags
for foo:bar using the revalidateTag function. The
reason for this is the Next.js Drupal module only provides a single API url for on demand invalidation. So we have to
implement our own logic.
The layout consists of the global elements on all pages. This consists of the global header, footer, and the menu. Any
site wide settings should also be used in the layout. The main menu in the header has cache tags: menus & menu:main.
The config pages have the cache tag config-pages since all config pages are fetched with a single request.
When a layout cache is invalidated, it has no impact on the route caches below. However, it will trigger every route to be rebuilt upon the next request. This shouldn't impact the CMS system since the route caches are still available.
Page routes are cache separately from Layouts. When invalidating a route or any fetch requests on the route, the layout
caches will not be impacted. Using the route handler, if we invalidate the slug /foo/bar using the revalidatePath
function, it will invalidate any fetch request that was used to build that single page and no other pages. Requests
like list paragraphs, or external fetches will be re-executed when the page is requested.
Pages may contain list paragraphs. Those paragraphs have a separate fetch so they can be invalidated when a content
changes in the CMS. Each view contains a cache tag in the form views:[content_type] that correlates to the content
type in the Drupal CMS. When this cache tag is invalidated, any route that contains that list paragraph will be rebuilt,
but only the list paragraph data will be re-fetched from the CMS.
Images are optimized on the hosting platform. It is recommended to use the original image from the source so that the derived images will be at the best resolution quality. Next.js provides extensive documentation about image optimization. Optimized images will then be cached on the hosting provider and stored for 31 days, unless triggered to be cleared out. Vercel has documentation explaining how their cache is handled.
Files, pdf, txt, etc., assets are referenced directly from the CMS. Their cache is managed by the Drupal hosting provider and/or Varnish/CDN/Etc.
To learn more about Next.js, take a look at the following resources:
- Next.js Documentation - learn about Next.js features and API.
- Learn Next.js - an interactive Next.js tutorial.
You can check out the Next.js GitHub repository - your feedback and contributions are welcome!
Production version on Vercel uses the main branch. 1.x branch is used for releases and is merged into main when a
release is to be performed. dev and test branches have their respective vercel and Acquia environments and are used as sandboxes.