swagger-to-graphqlSwagger to GraphQL API adapter
Swagger-to-GraphQL
Swagger-to-GraphQL converts your existing Swagger schema to an executable GraphQL schema where resolvers perform HTTP calls to certain real endpoints. It allows you to move your API to GraphQL with nearly zero effort and maintain both REST and GraphQL APIs. Our CLI tool also allows you get the GraphQL schema in Schema Definition Language.
Try it online! You can paste in the url to your own Swagger schema. There are also public OpenAPI schemas available in the APIs.guru OpenAPI directory.
Features
- Swagger (OpenAPI 2) and OpenAPI 3 support
- Bring you own HTTP client
- Typescript types included
- Runs in the browser
- Formdata request body
- Custom request headers
Usage
Basic server
This library will fetch your swagger schema, convert it to a GraphQL schema and convert GraphQL parameters to REST parameters. From there you are control of making the actual REST call. This means you can reuse your existing HTTP client, use existing authentication schemes and override any part of the REST call. You can override the REST host, proxy incoming request headers along to your REST backend, add caching etc.
import express, { Request } from 'express';
import graphqlHTTP from 'express-graphql';
import { createSchema, CallBackendArguments } from 'swagger-to-graphql';
const app = express();
// Define your own http client here
async function callBackend({
context,
requestOptions,
}: CallBackendArguments<Request>) {
return 'Not implemented';
}
createSchema({
swaggerSchema: `./petstore.yaml`,
callBackend,
})
.then(schema => {
app.use(
'/graphql',
graphqlHTTP(() => {
return {
schema,
graphiql: true,
};
}),
);
app.listen(3009, 'localhost', () => {
console.info('http://localhost:3009/graphql');
});
})
.catch(e => {
console.log(e);
});
Constructor (graphQLSchema) arguments:
export interface Options<TContext> {
swaggerSchema: string | JSONSchema;
callBackend: (args: CallBackendArguments<TContext>) => Promise<any>;
}
-
swaggerUrl
(string) is a path or URL to your swagger schema file. required -
callBackend
(async function) is called with all parameters needed to make a REST call as well as the GraphQL context.
CLI usage
You can use the library just to convert schemas without actually running server
npx swagger-to-graphql --swagger-schema=/path/to/swagger_schema.json > ./types.graphql
Apollo Federation
Apollo federation support can be added by using graphql-transform-federation. You can extend your swagger-to-graphql schema with other federated schemas or the other way around. See the demo with a transformed schema for a working example.
Defining your HTTP client
This repository has:
- node-fetch example. Read more about node-fetch.
- request-promise example. Read more about request.
To get started install node-fetch
and copy the node-fetch example into your server.
npm install node-fetch --save
Implementing your own HTTP client
There a unit test for our HTTP client example, it might be useful when implementing your own client as well.
The function callBackend
is called with 2 parameters:
-
context
is your GraphQL context. Forexpress-graphql
this is the incomingrequest
object by default. Read more. Use this if you want to proxy headers likeauthorization
. For exampleconst authorizationHeader = context.get('authorization')
. -
requestOptions
includes everything you need to make a REST call.
export interface CallBackendArguments<TContext> {
context: TContext;
requestOptions: RequestOptions;
}
RequestOptions
export interface RequestOptions {
baseUrl?: string;
path: string;
method: string;
headers?: {
[key: string]: string;
};
query?: {
[key: string]: string | string[];
};
body?: any;
bodyType: 'json' | 'formData';
}
-
baseUrl
like defined in your swagger schema:http://my-backend/v2
-
path
the next part of the url:/widgets
-
method
HTTP verb:get
-
headers
HTTP headers which are filled using GraphQL parameters:{ api_key: 'xxxx-xxxx' }
. Note these are not the http headers sent to the GraphQL server itself. Those will be on thecontext
parameter -
query
Query parameters for this calls:{ id: 123 }
. Note this can be an array. You can find some examples on how to deal with arrays in query parameters in the qs documentation. -
body
the request payload to send with this REST call. -
bodyType
how to encode your request payload. When thebodyType
isformData
the request should be URL encoded form data. Ensure your HTTP client sends the rightContent-Type
headers.