A Gatsby plugin for fetching source data from the Prismic headless CMS using Prismic's beta GraphQL API. This plugin provides full support for Prismic's preview feature out of the box.
For more context, be sure to checkout Prismic's getting started guide: Using Prismic With Gatsby. This README, however, serves as the most-up-to-date source of information on gatsby-source-prismic-graphql
's latest developments and breaking changes.
Please be sure your Prismic repository has the GraphQL API enabled. It is enabled by default on all new Prismic repositories. If you have an older repository or are unable to access https://[your_repo].prismic.io/graphql
, please reach out to Prismic support to request the GraphQL API.
- Differences From gatsby-source-prismic
- Getting Started
- Usage
- How This Plugin Works
- Development
- Issues & Troubleshooting
gatsby-source-prismic-graphql
(this plugin) fetches data using Prismic's beta GraphQL API and provides full support for Prismic's Preview feature out of the box. It also provides an easy-to-configure interface for page generation.
gatsby-source-prismic
is a different plugin that fetches data using Prismic's REST and Javascript APIs. Previews must be coded up separately.
Install the plugin
npm install --save gatsby-source-prismic-graphql
or
yarn add gatsby-source-prismic-graphql
Add plugin to gatsby-config.js
and configure
{
resolve: 'gatsby-source-prismic-graphql',
options: {
repositoryName: 'gatsby-source-prismic-test-site', // required
defaultLang: 'en-us', // optional, but recommended
accessToken: '...', // optional
prismicRef: '...', // optional, default: master; useful for A/B experiments
path: '/preview', // optional, default: /preview
previews: true, // optional, default: false
pages: [{ // optional
type: 'Article', // TypeName from prismic
match: '/article/:uid', // pages will be generated under this pattern (optional)
path: '/article', // placeholder page for unpublished documents
component: require.resolve('./src/templates/article.js'),
sortBy: 'date_ASC', // optional, default: meta_lastPublicationDate_ASC; useful for pagination
}],
extraPageFields: 'article_type', // optional, extends pages query to pass extra fields
sharpKeys: [
/image|photo|picture/, // (default)
'profilepic',
],
}
}
Edit your gatsby-browser.js
const { registerLinkResolver } = require('gatsby-source-prismic-graphql');
const { linkResolver } = require('./src/utils/linkResolver');
registerLinkResolver(linkResolver);
You can generate pages automatically by providing a mapping configuration under the pages
option in gatsby-config.js
.
Let's assume we have the following page configuration set:
{
pages: [{
type: 'Article',
match: '/blogpost/:uid',
path: '/blogpost',
component: require.resolve('./src/templates/article.js'),
}],
}
If you have two blog posts with UIDs of foo
and bar
, the following URL slugs will be generated:
/blogpost/foo
/blogpost/bar
If you create a new unpublished blogpost, baz
it will be accessible for preview under, assuming you've established a preview session with Prismic:
/blogpost?uid=baz
More on Prismic Previews below.
If the default page generation doesn't cover your use-case, you can provide an optional filter
option to your individual page configurations.
For example, if you had a single Prismic Article type and wanted pages with music
in their UIDs to be generated at a different URL :
{
pages: [{
type: 'Article',
match: '/musicblog/:uid',
filter: data => data.node._meta.uid.includes('music'),
path: '/blogposts',
component: require.resolve('./src/templates/article.js'),
}, {
type: 'Article',
match: '/blog/:uid',
filter: data => !data.node._meta.uid.includes('music'),
path: '/blogposts',
component: require.resolve('./src/templates/article.js'),
}],
}
Given 3 articles with UIDs of why-i-like-music
, why-i-like-sports
and why-i-like-food
, the following URL slugs will be generated:
/musicblog/why-i-like-music
/blog/why-i-like-sports
/blog/why-i-like-food
Sometimes the meta provided by default doesn't contain enough context to be able to filter pages effectively. By passing extraPageFields
to the plugin options, we can extend what we can filter on.
{
extraPageFields: 'music_genre',
pages: [{
type: 'Article',
match: '/techno/:uid',
filter: data => data.node.music_genre === 'techno',
path: '/blogposts',
component: require.resolve('./src/templates/article.js'),
}, {
type: 'Article',
match: '/acoustic/:uid',
filter: data => data.node.music_genre === 'acoustic',
path: '/blogposts',
component: require.resolve('./src/templates/article.js'),
}]
}
Given 2 articles with the music_genre
field set, we'll get the following slugs:
/techno/darude /acoustic/mik-parsons
Prismic allows you to create your content in multiple languages. This library supports that too. When setting up your configuration options in gatsby-config.js
, there are three optional properties you should be aware of: options.defaultLang
, options.langs
, and options.pages[i].langs
. In the following example, all are in use:
{
resolve: 'gatsby-source-prismic-graphql',
options: {
repositoryName: 'gatsby-source-prismic-test-site',
defaultLang: 'en-us',
langs: ['en-us', 'es-es', 'is'],
path: '/preview',
previews: true,
pages: [{
type: 'Article',
match: '/:lang?/:uid',
path: '/article',
component: require.resolve('./src/templates/article.js'),
sortBy: 'date_ASC',
langs: ['en-us', 'es-es', 'is'],
}, {
type: "Noticias",
match: '/noticias/:uid',
path: '/noticias',
component: require.resolve('./src/templates/noticias.js'),
sortBy: 'date_ASC',
langs: ['es-es'],
}],
}
}
In the example above, pages are generated for two document types from Prismic--Articles and Noticias. The latter consists of news stories in Spanish. There are three languages total in use in this blog: US English, Traditional Spanish and Icelandic.
For Articles, we are instructing the plugin to generate pages for articles of all three languages. But, because there is a question mark (?
) after the :lang
portion of the match
property (/:lang?/:uid
), we only include the locale tag in the URL slug for languages that are not the defaultLang
specified above (i.e., 'en-us'). So for the following languages, these are the slugs generated:
- US English:
/epic-destinations
- Spanish:
/es-es/destinos-increibles
- Icelandic:
/is/reykjadalur
If we had not specified a defaultLang
, the slug for US English would have been /en-us/epic-destinations
. And, in fact, including the langs: ['en-us', 'es-es', 'is']
declaration for this particular document type (Articles
) is unnecessary because we already specified that as the default language set right after defaultLang
in the plugin options.
For Noticias, however, we only want to generate pages for Spanish documents of that type (langs
is [es-es]
). We decide that in this context, no locale tag is needed in the URL slug; "noticias" is already enough indication that the contents are in Spanish. So we omit the :lang
match entirely and specify only match: '/noticias/:uid'
.
This is an example of how these three properties can be used together to offer maximum flexibility. To see this in action, check out the languages example app.
To use short language codes (e.g. /fr/articles
) instead of the default (e.g. /fr-fr/articles
), you can set options.shortenUrlLangs
to true
.
Keep in mind that if you use this option & have multiple variants of a language (e.g. en-us and en-au) that would be shortened to the same value, you should add UIDs to your URLs to differentiate them.
It is very easy to fetch data from Prismic in your pages:
import React from 'react';
import { RichText } from 'prismic-reactjs';
export const query = graphql`
{
prismic {
page(uid:"homepage", lang:"en-us") {
title
description
}
}
}
`
export default function Page({ data }) => <>
<h1>{RichText.render(data.prismic.title)}</h1>
<h2>{RichText.render(data.prismic.description)}</h2>
</>
Previews are enabled by default, however they must be configured in your prismic instance/repository. For instructions on configuring previews in Prismic, refer to Prismic's guide: How to set up a preview.
When testing previews, be sure you are starting from a valid Prismic preview URL/path. The most reliable way to test previews is by using the preview button from your draft in Prismic. If you wish to test the Preview locally, catch the URL that opens immediately after clicking the preview link:
https://[your-domain.tld]/preview?token=https%3A%2F%[your-prismic-repo].prismic.io%2Fpreviews%2FXRag6xAAACA...ABwjduaa%3FwebsitePreviewId%3DXRA...djaa&documentId=XRBH...jduAa
Then replace the protocol and domain at the beginning of the URL with your localhost:PORT
instance, or wherever you're wanting to preview from.
This URL will be parsed and replaced by the web app and browser with the proper URL as specified in your page configuration.
You can use StaticQuery
as usual, but if you would like to preview them, you must use the withPreview
function.
import { StaticQuery, graphql } from 'gatsby';
import { withPreview } from 'gatsby-source-prismic-graphql';
const articlesQuery = graphql`
query {
prismic {
...
}
}
`;
export const Articles = () => (
<StaticQuery
query={articlesQuery}
render={withPreview(data => { ... }, articlesQuery)}
/>
);
useStaticQuery
is not yet supported.
Fragments are supported for both page queries and static queries.
Within page components:
import { graphql } from 'gatsby';
const fragmentX = graphql` fragment X on Y { ... } `;
export const query = graphql`
query {
...X
}
`;
const MyPage = (data) => { ... };
MyPage.fragments = [fragmentX];
export default MyPage;
With StaticQuery:
import { StaticQuery, graphql } from 'gatsby';
import { withPreview } from 'gatsby-source-prismic-graphql';
const fragmentX = graphql` fragment X on Y { ... } `;
export const query = graphql`
query {
...X
}
`;
export default () => (
<StaticQuery
query={query}
render={withPreview(data => { ... }, query, [fragmentX])}
/>
);
You can use this plugin to dynamically fetch data for your component using prismic.load
. Refer to the pagination example to see it in action.
import React from 'react';
import { graphql } from 'gatsby';
export const query = graphql`
query Example($limit: Int) {
prismic {
allArticles(first: $limit) {
edges {
node {
title
}
}
}
}
}
`;
export default function Example({ data, prismic }) {
const handleClick = () =>
prismic.load({
variables: { limit: 20 },
query, // (optional)
fragments: [], // (optional)
});
return (
// ... data
<button onClick={handleClick}>load more</button>
);
}
Pagination can be accomplished statically (i.e., during initialy page generation) or dynamically (i.e., with JS in the browser). Examples of both can be found in the pagination example.
Prismic pagination is cursor-based. See Prismic's Paginate your results article to learn about cursor-based pagination.
By default, pagination will be sorted by last publication date. If you would like to change that, specify a sortBy
value in your page configuration in gatsby-config.js
.
When coupled with prismic.load
, as demonstrated in the index page of the pagination example, other pages can be fetched dynamically using page and cursor calculations.
GraphQL documents from Prismic have a cursor--a base64-encoded string that represents their order, or page number, in the set of all documents queried. We provide two helpers for converting between cursor strings and page numbers:
getCursorFromDocumentIndex(index: number)
getDocumentIndexFromCursor(cursor: string)
For basic linking between the pages, metadata for the previous and next pages are provided to you automatically via pageContext
in the paginationPreviousMeta
and paginationNextMeta
properties. These can be used in conjunction with your linkResolver
to generate links between pages without any additional GraphQL query. For an example of this, take a look at the <Pagination />
component in the pagination example's article.js
.
If you would like to gather other information about previous and next pages (say a title or image), simply modify your page query to retrieve those documents. This also is demonstrated in the same pagination example with the <EnhancedPagination />
component and the page's GraphQL query.
The latest versions of this plugin support gatsby-image by adding a new property to GraphQL types that contains fields that match the sharpKeys
array (this defaults to /image|photo|picture/
) to the Sharp
suffix.
Note: When querying, make sure to also query the source field. For example:
query {
prismic {
Article(id: "123") {
title
articlePhoto
articlePhotoSharp {
childImageSharp {
fluid(maxWidth: 400, maxHeight: 250) {
...GatsbyImageSharpFluid
}
}
}
}
}
}
You can also get access to specific crop sizes from Prismic by passing the crop
argument:
query {
prismic {
Author(id: "123") {
name
profile_picture
profile_pictureSharp(crop: "face") {
childImageSharp {
fluid(maxWidth: 500, maxHeight: 500) {
...GatsbyImageSharpFluid
}
}
}
}
}
}
NOTE Images are not transformed in preview mode, so be sure to fall back to the default image when the sharp image is null
.
import Img from 'gatsby-image';
import get from 'lodash/get';
// ...
const sharpImage = get(data, 'prismic.Author.profile_pictureSharp.childImageSharp.fluid');
return sharpImage ? (
<Img fluid={sharpImage} />
) : (
<img src={get(data, 'prismic.Author.profile_picture.url')} />
);
Later, we may add an Image
component that does this for you and leverages the new Prismic Image API as a fallback for preview modes.
You can use this plugin in combination with Prismic's built-in experiments functionality, and a hosting service like Netlify, to run content A/B tests.
Experiments in Prismic are basically branches of the core content, split into 'refs' similar to git branches. So if you want to get content from a certain experiment variation, you can pass the corresponding ref through to Prismic in your request, and it will return content based on that ref's variation.
A/B experiments are tricky to implement in a static website though; A/B testing needs a way to dynamically serve up the different variations to different website visitors. This is at odds with the idea of a static, non-dynamic website.
Fortunately, static hosting providers like Netlify allow you to run A/B tests at a routing level. This makes it possible for us to build multiple versions of our project using different source data, and then within Netlify split traffic to our different static variations.
Therefore, we can use A/B experiments from Prismic in the following way:
-
Setup an experiment in Prismic.
-
Create a new git branch of your project which will be used to get content. You will need to create a separate git branch for each variation.
-
In that git branch, edit/add the optional 'prismicRef' parameter (documented above). The value of this should be the ref of the variation this git branch is for.
-
Push the newly created branch to your git repo.
-
Now go to your static hosting provider (we'll use Netlify in this example), and setup split testing based on your git branches/Prismic variations.
-
Now your static website will show different experimental variations of the content to different users! At this point the process is manual and non-ideal, but hopefully we'll be able to automate it more in the future.
-
The plugin creates a new page at
/preview
(by default, you can change this), that will be your preview URL you setup in the Prismic admin interface.It will automatically set cookies based on the query parameters and attempt to find the correct page to redirect to with your linkResolver.
-
It uses a different
babel-plugin-remove-graphql-queries
on the client.The modified plugin emits your GraphQL queries as a string so they can be read and re-used on the client side by the plugin.
-
Once redirected to a page with the content, everything will load normally.
In the background, the plugin takes your original Gatsby GraphQL query, extracts the Prismic subquery and uses it to make a GraphQL request to Prismic with a preview reference.
Once data is received, it will update the
data
prop with merged data from Prismic preview and re-render the component.
git clone git@github.com:birkir/gatsby-source-prismic-graphql.git
cd gatsby-source-prismic-graphql
yarn install
yarn setup
yarn start
# select example to work with
cd examples/default
yarn start
Please raise an issue on GitHub if you have any problems.
This is a Gatsby limitation. You can bypass this limitation by adding the following:
export const query = graphql` ... `;
const MyPage = () => { ... };
MyPage.query = query; // <-- set the query manually to allow hot-reload.