GraphQL schema components.
This project is designed to make module or component based development of graphql schemas easy.
Read more about the idea here:
This is very similar to the excellent graphql-modules
project — but a little closer to our own internal paradigm already in use for over a year and a half and adds some features such as exclude
root types from imports
and memoize resolvers.
In fact, this module utilizes the graphql-toolkit
developed by the graphql-modules
team to merge types and resolvers.
The future
Experimental / alpha for now.
Repository structure
lib
- the graphql-component code.test/examples/example-listing/property-component
- a component implementation for Property
.test/examples/example-listing/reviews-component
- a component implementation for Reviews
.test/examples/example-listing/listing-component
- a component implementation composing Property
and Reviews
.test/examples/example-listing/server
- the "application".
Running the example
Can be run with node examples/server/index.js
or npm start
which will start with debug flags.
Debugging
Enable debug logging with DEBUG=graphql-component:*
Activating mocks
To intercept resolvers with mocks execute this app with GRAPHQL_DEBUG=1
enabled.
API
GraphQLComponent(options)
- the component class, which may also be extended. Its options include:
types
- A string or array of strings representing typeDefs and rootTypes.resolvers
- An object containing resolver functions.imports
- An optional array of imported components for the schema to be merged with.context
- An optional object { namespace, factory } for contributing to context.directives
- An optional object containing custom schema directives.useMocks
- Enable mocks.preserveMockResolvers
- Preserve type resolvers in mock mode.mocks
- An optional object containing mock types.
A new GraphQLComponent instance has the following API:
schema
- getter that returns an executable schema.context
- context builder.schemaDirectives
- schema directives for this component.types
- this component's types.resolvers
- this component's resolvers.imports
- this component's imports.execute
- accepts a graphql query.
General usage
Creating a component using the GraphQLComponent class:
const GraphQLComponent = require('graphql-component');
const { schema, context } = new GraphQLComponent({ types, resolvers });
Encapsulating state
Typically the best way to make a re-useable component with instance data will be to extend GraphQLComponent
.
const GraphQLComponent = require('graphql-component');
const resolvers = require('./resolvers');
const types = require('./types');
const mocks = require('./mocks');
class PropertyComponent extends GraphQLComponent {
constructor({ useMocks, preserveTypeResolvers }) {
super({ types, resolvers, mocks, useMocks, preserveTypeResolvers });
}
}
module.exports = PropertyComponent;
This will allow for configuration (in this example, useMocks
and preserveTypeResolvers
) as well as instance data per component (such as data base clients, etc).
Aggregation
Example to merge multiple components:
const { schema, context } = new GraphQLComponent({
imports: [
new Property(),
new Reviews()
]
});
const server = new ApolloServer({
schema,
context
});
Excluding root fields from imports
You can exclude root fields from imported components:
const { schema, context } = new GraphQLComponent({
imports: [
{
component: new Property(),
exclude: ['Mutation.*']
},
{
component: new Reviews(),
exclude: ['Mutation.*']
}
]
});
This will keep from leaking unintended surface area. But you can still delegate calls to the component's schema to enable it from the API you do expose.
Directly executing components
Components can be directly executed via the execute
function. The execute
function is basically a passthrough to graphql.execute
and is mostly useful for components calling imported components and the like.
For example, this allows one component to invoke another component and still get the benefits of that component's schema type resolvers and validation.
execute(input, options)
accepts an input
string and an optional options
object with the following fields:
root
- root object.context
- context object value.variables
- key:value mapping of variables for the input.
The execute
function also adds some helper fragments. For any type you query in a component, a helper fragment will exist to query all fields.
Example extending Property
to include a reviews
field that delegates to another component:
class PropertyComponentReviews extends GraphQLComponent {
constructor({ useMocks, preserveTypeResolvers }) {
const propertyComponent = new PropertyComponent();
const reviewsComponent = new ReviewsComponent();
super ({
types: [
`type Property { reviews: [Review] }`
],
resolvers: {
Property: {
reviews(_, args, context) {
return reviewsComponent.execute(`query { reviewsByPropertyId(id: ${_.id}) { ...AllReview }}`, { context });
}
}
},
imports: [
propertyComponent,
{
component: reviewsComponent,
exclude: ['*']
}
]
});
}
}
For the Review
type in the reviewsComponent
, a helper fragment will exist as AllReview
that provides all fields.
Adding to context
Example context argument:
const context = {
namespace: 'myNamespace',
factory: function ({ req }) {
return 'my value';
}
};
After this, resolver context
will contain { ..., myNamespace: 'my value' }
.
Context middleware
It may be necessary to transform the context before invoking component context.
const { schema, context } = new GraphQLComponent({types, resolvers, context});
context.use('transformRawRequest', ({ request }) => {
return { req: request.raw.req };
});
Using context
now in apollo-server-hapi
for example, will transform the context to one similar to default apollo-server
.
Mocks
graphql-component accepts mocks in much the same way that Apollo does but with one difference.
Instead of accepting a mocks object, it accepts (importedMocks) => mocksObject
argument. This allows components to utilize the mocks from other imported components easily.