Apollo GraphQL server mixin for Moleculer API Gateway
npm i moleculer-apollo-server moleculer-web graphql
This example demonstrates how to setup a Moleculer API Gateway with GraphQL mixin in order to handle incoming GraphQL requests via the default /graphql
endpoint.
"use strict";
const ApiGateway = require("moleculer-web");
const { ApolloService } = require("moleculer-apollo-server");
module.exports = {
name: "api",
mixins: [
// Gateway
ApiGateway,
// GraphQL Apollo Server
ApolloService({
// Global GraphQL typeDefs
typeDefs: ``,
// Global resolvers
resolvers: {},
// API Gateway route options
routeOptions: {
path: "/graphql",
cors: true,
mappingPolicy: "restrict"
},
// https://www.apollographql.com/docs/apollo-server/v2/api/apollo-server.html
serverOptions: {
tracing: true,
engine: {
apiKey: process.env.APOLLO_ENGINE_KEY
}
}
})
]
};
Start your Moleculer project, open http://localhost:3000/graphql in your browser to run queries using graphql-playground, or send GraphQL requests directly to the same URL.
Define queries & mutations in service action definitions
module.exports = {
name: "greeter",
actions: {
hello: {
graphql: {
query: "hello: String"
},
handler(ctx) {
return "Hello Moleculer!"
}
},
welcome: {
params: {
name: "string"
},
graphql: {
mutation: "welcome(name: String!): String"
},
handler(ctx) {
return `Hello ${ctx.params.name}`;
}
}
}
};
Generated schema
type Mutation {
welcome(name: String!): String
}
type Query {
hello: String
}
posts.service.js
module.exports = {
name: "posts",
settings: {
graphql: {
type: `
"""
This type describes a post entity.
"""
type Post {
id: Int!
title: String!
author: User!
votes: Int!
voters: [User]
createdAt: Timestamp
}
`,
resolvers: {
Post: {
author: {
// Call the `users.resolve` action with `id` params
action: "users.resolve",
rootParams: {
"author": "id"
}
},
voters: {
// Call the `users.resolve` action with `id` params
action: "users.resolve",
rootParams: {
"voters": "id"
}
}
}
}
}
},
actions: {
find: {
//cache: true,
params: {
limit: { type: "number", optional: true }
},
graphql: {
query: `posts(limit: Int): [Post]`
},
handler(ctx) {
let result = _.cloneDeep(posts);
if (ctx.params.limit)
result = posts.slice(0, ctx.params.limit);
else
result = posts;
return _.cloneDeep(result);
}
},
findByUser: {
params: {
userID: "number"
},
handler(ctx) {
return _.cloneDeep(posts.filter(post => post.author == ctx.params.userID));
}
},
}
};
users.service.js
module.exports = {
name: "users",
settings: {
graphql: {
type: `
"""
This type describes a user entity.
"""
type User {
id: Int!
name: String!
birthday: Date
posts(limit: Int): [Post]
postCount: Int
}
`,
resolvers: {
User: {
posts: {
// Call the `posts.findByUser` action with `userID` param.
action: "posts.findByUser",
rootParams: {
"id": "userID"
}
},
postCount: {
// Call the "posts.count" action
action: "posts.count",
// Get `id` value from `root` and put it into `ctx.params.query.author`
rootParams: {
"id": "query.author"
}
}
}
}
}
},
actions: {
find: {
//cache: true,
params: {
limit: { type: "number", optional: true }
},
graphql: {
query: "users(limit: Int): [User]"
},
handler(ctx) {
let result = _.cloneDeep(users);
if (ctx.params.limit)
result = users.slice(0, ctx.params.limit);
else
result = users;
return _.cloneDeep(result);
}
},
resolve: {
params: {
id: [
{ type: "number" },
{ type: "array", items: "number" }
]
},
handler(ctx) {
if (Array.isArray(ctx.params.id)) {
return _.cloneDeep(ctx.params.id.map(id => this.findByID(id)));
} else {
return _.cloneDeep(this.findByID(ctx.params.id));
}
}
}
}
};
moleculer-apollo-server supports file uploads through the GraphQL multipart request specification.
To enable uploads, the Upload scalar must be added to the Gateway:
"use strict";
const ApiGateway = require("moleculer-web");
const { ApolloService, GraphQLUpload } = require("moleculer-apollo-server");
module.exports = {
name: "api",
mixins: [
// Gateway
ApiGateway,
// GraphQL Apollo Server
ApolloService({
// Global GraphQL typeDefs
typeDefs: ["scalar Upload"],
// Global resolvers
resolvers: {
Upload: GraphQLUpload
},
// API Gateway route options
routeOptions: {
path: "/graphql",
cors: true,
mappingPolicy: "restrict"
},
// https://www.apollographql.com/docs/apollo-server/v2/api/apollo-server.html
serverOptions: {
tracing: true,
engine: {
apiKey: process.env.APOLLO_ENGINE_KEY
}
}
})
]
};
Then a mutation can be created which accepts an Upload argument. The fileUploadArg
property must be set to the mutation's argument name so that moleculer-apollo-server knows where to expect a file upload. When the mutation's action handler is called, ctx.params
will be a Readable Stream which can be used to read the contents of the uploaded file (or pipe the contents into a Writable Stream). File metadata will be made available in ctx.meta.$fileInfo
.
files.service.js
module.exports = {
name: "files",
settings: {
graphql: {
type: `
"""
This type describes a File entity.
"""
type File {
filename: String!
encoding: String!
mimetype: String!
}
`
}
},
actions: {
uploadFile: {
graphql: {
mutation: "uploadFile(file: Upload!): File!",
fileUploadArg: "file",
},
async handler(ctx) {
const fileChunks = [];
for await (const chunk of ctx.params) {
fileChunks.push(chunk);
}
const fileContents = Buffer.concat(fileChunks);
// Do something with file contents
// Additional arguments:
this.logger.info("Additional arguments:", ctx.meta.$args);
return ctx.meta.$fileInfo;
}
}
}
};
To accept multiple uploaded files in a single request, the mutation can be changed to accept an array of Upload
s and return an array of results. The action handler will then be called once for each uploaded file, and the results will be combined into an array automatically with results in the same order as the provided files.
...
graphql: {
mutation: "upload(file: [Upload!]!): [File!]!",
fileUploadArg: "file"
}
...
moleculer-apollo-server supports DataLoader via configuration in the resolver definition. The called action must be compatible with DataLoader semantics -- that is, it must accept params with an array property and return an array of the same size, with the results in the same order as they were provided.
To activate DataLoader for a resolver, simply add dataLoader: true
to the resolver's property object in the resolvers
property of the service's graphql
property:
settings: {
graphql: {
resolvers: {
Post: {
author: {
action: "users.resolve",
dataLoader: true,
rootParams: {
author: "id",
},
},
voters: {
action: "users.resolve",
dataLoader: true,
rootParams: {
voters: "id",
},
},
...
Since DataLoader only expects a single value to be loaded at a time, only one rootParams
key/value pairing will be utilized, but params
and GraphQL child arguments work properly.
You can also specify options for construction of the DataLoader in the called action definition's graphql
property. This is useful for setting things like `maxBatchSize'.
resolve: {
params: {
id: [{ type: "number" }, { type: "array", items: "number" }],
graphql: { dataLoaderOptions: { maxBatchSize: 100 } },
},
handler(ctx) {
this.logger.debug("resolve action called.", { params: ctx.params });
if (Array.isArray(ctx.params.id)) {
return _.cloneDeep(ctx.params.id.map(id => this.findByID(id)));
} else {
return _.cloneDeep(this.findByID(ctx.params.id));
}
},
},
It is unlikely that setting any of the options which accept a function will work properly unless you are running moleculer in a single-node environment. This is because the functions will not serialize and be run by the moleculer-web Api Gateway.
- Simple
npm run dev
- File Upload
npm run dev upload
- See here for information about how to create a file upload request
- Full
npm run dev full
- Full With Dataloader
- set
DATALOADER
environment variable to"true"
npm run dev full
- set
$ npm test
In development with watching
$ npm run ci
Please send pull requests improving the usage and fixing bugs, improving documentation and providing better examples, or providing some testing, because these things are important.
The project is available under the MIT license.
Copyright (c) 2020 MoleculerJS