Skip to content

segpacto/gql-gateway

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

30 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Apollo GraphQL Gateway

Description

This module provides a GraphQL Gateway that allows the interaction with Swagger based REST APIs, by autogenerating and merging their GraphQL schemas. 🚀
Through this gateway, it is possible to easily establish aggregations between the downstream REST services using GraphQL generated types, queries and mutations.

Related topics

How this GraphQL-Gateway actually works?

  1. Read and parse the Swagger specifications from all given endpoints.
  2. For each Swagger specification auto-generate the GraphQL Types, Queries and Mutations; as well as auto-generate the APIs based resolvers.
  3. Merge our local GraphQl definitions containing the aggregations and extensions along with the previous generated schemas.
  4. Serve an Apollo GraphQl server with all agreggations.

Getting started

Installation

npm install --save gql-gateway

Getting started

Basic - Serve a basic GraphQL Gateway from public services

const gateway = require('gql-gateway')

const endpointsList = [
  { name: 'petstore_service', url: 'https://petstore.swagger.io/v2/swagger.json' },
  { name: 'fruits_service', url: 'https://api.predic8.de/shop/swagger' }
]

gateway({ endpointsList })
  .then(server => server.listen(5000))
  .then(console.log('Service is now running at port: 5000'))
  .catch(err => console.log(err))

Advanced - Adding aggregations

const gateway = require('gql-gateway')

const localSchema = `
  extend type Order {
    pet: Pet 
  } 
`

const resolvers = {
  /* 
  Query : {
     ....
  } 
  */   
  Order: {
    pet: {
      fragment: '... on Order {petId}',
      async resolve (order, args, context, info) {
        const schema = await context.resolveSchema('pet_service')

        return info.mergeInfo.delegateToSchema({
          schema,
          operation: 'query',
          fieldName: 'getPetById',
          args: { petId: order.petId },
          context,
          info
        })
      }
    }
  }
}

const config = {
  port: 5000,
  playgroundBasePath: 'gql-gateway'
}

const endpointsList = [
  { name: 'pet_service', url: 'https://petstore.swagger.io/v2/swagger.json' }
]

const apolloServerConfig = { playground: { endpoint: config.playgroundBasePath } }

gateway({ resolvers, localSchema, endpointsList, apolloServerConfig })
  .then(server => server.listen(config.port))
  .then(console.log(`Service is now running at port: ${config.port}`))
  .catch(err => console.log(err))

What just happened?

  • On localSchema we declare the aggregations that we would like to have by extending the original schemas (to get the original schemas, queries and mutations it is recommended to publish the service and then take a look at them before start adding aggregations).
  • On resolvers we declare the way how to resolve the model Order, for this we use graphql delegations, where we specify on which of the autogenerated queries or mutations we relay to obtain the pet property in Order, in this case getPetById.

Note that on the fragment part we declare petId as required field to obtain the pet property, so petId is going to be injected from the Order to the resolver even if it haven't been requested originally.

Configuration options explained

Name Default Description
localSchema empty Schema that contains the aggregations that we want to establish between the REST API services
resolvers empty Resolvers that implement delegation. See samples above
endpointsList required Contains a list of json swagger endpoints where to retrieve definitions to build the graphql schemas. Minimum one element
apolloServerConfig empty Apollo Server configuration (https://www.apollographql.com/docs/apollo-server/api/apollo-server/#apolloserver)
contextConfig empty Object that contains middlewares and also used to inject data into the Context (https://www.apollographql.com/docs/apollo-server/api/apollo-server/#apolloserver)
logger console Default logger is the console

Format of localSchema parameters:

Name Default Description
name required Is used to identify the service
url required Url of the service swagger in json format
headers empty Headers passed to request the json swagger service, in case any kind of particular auth is needed
onLoaded empty Function that process the swaggerJson once is loaded so changes on the flight can be introduced. If passed Must return the swaggerJson back

Format of function onLoaded parameters:

Name Default Description
swaggerJson Swagger JSON schema Contains the loaded Swagger Json schema
service object Contains the localSchema that was loaded

onLoaded function Ex :

const onLoaded = (swaggerJson, service) => {
  swaggerJson.schemes = ['http', 'https']
  return swaggerJson
}
const endpointsList = [
  { name: 'pet_service', url: 'https://petstore.swagger.io/v2/swagger.json', onLoaded }
]

Using the apolloServerConfig parameter:

const apolloServerConfig = { 
  playground: { 
    endpoint: config.playgroundBasePath 
  } 
}

Technical Explanation

Below, we describe how to interact between services swagger based using agreggations(relations).
In this example we take the User and Product services as example.

The User service:

...
paths:
    "/users":
        get:
            description: "Return an Array of existing users"
            responses:
                '200':
                    description: "successful operation"
                    schema:
                        type: array
                        items:  
                            "$ref": "#/definitions/User"
...
definitions :
    User:
        type: object
        properties:
            userId:
                type: string
            firstname:
                type: string
            lastname:
                type: string
...

The Product service:

...
paths:
  paths:
    '/products/{userId}':
      get:
        tags:
          - Product
        parameters:
          - name: userId
            in: path
            description: ID of the user to fetch last products
            required: true
            type: string
        summary: Return a summary of the last products
        description: Return a sumary of the user products
        responses:
          '200':
            description: successful operation
            schema:
                type: array
                items:  
                    "$ref": "#/definitions/Product"
...
definitions :
    Product:
        type: object
        properties:
            productId:
                type: string
            userId:
                type: string
            name:
                type: string
            type:
                type: string
...

Once the graphql gateway read from those services their swagger specification, our server generates the following:

type Queries {
    get_products_userId(userId: String!): Products!
    get_users(): [User]!
}

Custom aggregations / relations

The next step is to extend the GraphQL definitions to introduce our custom global aggregations:

  • Extend GraphQl Types definitions:
    extend type User {
        products: Product
    }

    # You can always declare the relation in one direction
    extend type Product {
        user: User
    }

This will automatically indicate to the GraphQl server that the Type User will have another field named products, actually the Product service relation.

  • Finally, extend GraphQl resolvers:
User: {
    products: {
      fragment: '... on User {userId}',
      async resolve (user, args, context, info) {
        return info.mergeInfo.delegateToSchema({
          schema: userSchema,
          operation: 'query',
          fieldName: 'get_products_userId',
          args: {
            userId: user.userId  // here we hook the relation identifier
          },
          context,
          info
        })
      }
    }
  },
  • And now we can magically query:
query {
  get_users {
    firstname,
    lastname,
    products {
      name,
      type
    }
  }
}