A Drupal module that provides integration endpoints for the
@amazeelabs/gatsby-source-silverback
Gatsby
plugin. Allows writing custom GraphQL schema definitions based on GraphQL V4 and
automatically send incremental updates to Gatsby.
In version 2, a couple of breaking changes have been introduced due to the new
dependency on the graphql_directives
module.
@resolveEntityReference
and@resolveEntityReferenceRevisions
do not support thesingle
parameter any more. They can be chained with@seek
instead.- Gutenberg directives moved to the
silverback_gutenberg
module and changed in naming and parameters. Refer to the generated directives for more information. - New default value handling requires either nullable custom type invocations,
or
@default
directives on them. Refer to thegraphql_directives
module for more information.
First, simply install and enable the module.
composer require amazeelabs/silverback_gatsby
drush en -y silverback_gatsby
Create a GraphQL folder at the root of your project. This will contain all the schema definitions. To improve IDE support, you can export all the schema definitions in a single file:
drush graphql:directives > graphql/directives.graphqls
It is advised to ignore this file in version control and rather re-create it when needed.
Now you can start to create the project schema defintion and fill in resolvers by using the directives provided.
schema {
query: Query
}
type Query {
page(id: String!): Page @loadEntity(type: "node", id: "$id")
}
type Page @entity(type: "node", bundle: "page") {
path: String! @resolveEntityPath
title: String! @resolveEntityLabel
body: String @resolveProperty(path: "body.value")
}
Now create a new GraphQL server configuration, use the Directable
schema
plugin and make sure to enable the "Silverback Gatsby" extension. The Explorer
or Voyager screens should show root level fields for loading and querying our
type (loadPage
, queryPages
) that you should be able to test now.
Available using @isPath
and @isTemplate
field directives. See
@amazeelabs/gatsby-source-silverback
plugin README for details.
There are directives which create GraphQL resolvers automatically.
This field directive is a shortcut for property_path
data producer.
For example, this schema
type Page @entity(type: "node", bundle: "page") {
body: String @resolveProperty(path: "field_body.0.processed")
}
Will create the following resolver for Page.body
field
$builder->produce('property_path', [
'path' => $builder->fromValue('field_body.0.processed'),
'value' => $builder->fromParent(),
'type' => $builder->fromValue('entity:node:page'),
])
Resolves the relative path to an entity. A shortcut for entity_url
+url_path
data producers.
Example:
type Page @entity(type: "node", bundle: "page") {
path: String! @resolveEntityPath
}
Resolves the references entities. A shortcut for entity_reference
data
producer.
Example:
type Page @entity(type: "node", bundle: "page") {
relatedArticles: [Article]! @resolveEntityReference(field: "field_related_articles", single: false)
parentPage: Page @resolveEntityReference(field: "field_related_articles", single: true)
}
Resolves the entity reference revisions fields, e.g. Paragraphs. A shortcut for
entity_reference_revisions
data producer.
Example:
type Page @entity(type: "node", bundle: "page") {
paragraphs: [PageParagraphs!]! @resolveEntityReferenceRevisions(field: "field_paragraphs", single: false)
singleParagraph: ParagraphText @resolveEntityReferenceRevisions(field: "field_single_paragraph", single: true)
}
To expose Drupal menus to Gatsby, one can use the @menu
directive.
type MainMenu @menu(menu_id: "main") {
items: [MenuItem!]! @resolveMenuItems
}
type MenuItem {
id: String! @resolveMenuItemId
parent: String! @resolveMenuItemId
label: String! @resolveMenuItemLabel
url: String! @resolveMenuItemUrl
}
GraphQL does not allow recursive fragments, so something like this would not be possible:
query Menu {
drupalMainMenu {
items {
...MenuItem
}
}
}
fragment MenuItem on MenuItem {
label
url
children {
# Fragment recursion, not allowed in GraphQL!
...MenuItem
}
}
That's why the menu tree is automatically flattened, and id
and parent
properties are added to each item, so the tree can easily be reconstructed in
the consuming application.
query MainMenu {
drupalMainMenu(langcode: { eq: "en" }) {
items {
id
parent
label
url
}
}
}
The @menu
directive also takes an optional max_level
argument. It can be
used to restrict the number of levels a type will include, which in turn can
optimize caching and Gatsby build times. In many cases, the main page layout
only displays the first level of menu items. When a new page is created and
attached to the third level, Gatsby will still re-render all pages, because the
menu that is used in the header changed. By separating this into two levels, we
can make sure the outer layout really only changes when menu levels are changed
that are displayed.
type MainMenu @menu(menu_id: "main") {...}
# Will only include the first level and trigger updates when a first level item
# changes.
type LayoutMainMenu @menu(menu_id: "main", max_level: 1) {...}
In some cases, the same GraphQL field might have to return different menus, based on the current context. A prominent use case would be a multi-site setup where different menus should be displayed based on the current account Gatsby is using to fetch data with.
In this case, multiple menu id's can be passed to the @menu
directive, and the
resolver will pick the first one that is accessible to the user account.
type MainMenu
@menu(menu_ids: ["site_a_main", "site_b_main"], item_type: "MenuItem")
It checks access for the view label
operation on the Menu
entity, which is
allowed for everybody by default. The consuming project has to implement other
mechanisms to restrict access therefore control which menus are used for which
site.
The last thing to do is to tell Gatsby whenever something noteworthy changes. By
using the @entity
directive in our schema, we already told Drupal to keep
track of all changes related to the entity types we care about. All there is
missing is a Gatsby webhook url to trigger a refresh. We provide this via an
environment variable that is named after our configured GraphQL server.
GATSBY_BUILD_HOOK_[uppercased-server-id]=https://...
So if the server was called My Server
and the automatically generated machine
name is my_server
, then the environment variable would look like this:
GATSBY_BUILD_HOOK_MY_SERVER=https://...
The value is a semicolon-separated list of urls that will be called in case of
an update. This can be http://localhost:8000/__refresh
, for local testing a
Gatsby environment with ENABLE_GATSBY_REFRESH_ENDPOINT=true
, or the build and
preview webhooks provided by Gatsby Cloud.
The Gatsby site has to contain the
@amazeelabs/gatsby-source-silverback
plugin for
this to work.
By default, @amazeelabs/gatsby-source-silverback
behaves like an anonymous user. To change that, simply create a user account
with the required permissions and pass the credentials to the auth_user
and
auth_pass
configuration options of the plugin.
A very common use case would be to create a "preview" user that bypasses content access control and use it for the "Preview" environment on Gatsby cloud, so unpublished content can be previewed. Another sensible case would be to create a "build" user that has access to published content and block anonymous access to Drupal entirely.
There are multiple ways to trigger a Gatsby build:
- on entity save
- via the Drupal UI or Drush.
On the Build tab of the schema configuration, check the Trigger a build on entity save checkbox.
On the same Build tab, click the Gatsby Build button.
This command can be configured in the system cron.
drush silverback-gatsby:build [server_id]