Skip to content
Use GraphQL without a server, directly from anywhere to any API.
JavaScript
Branch: master
Clone or download
1
Latest commit 4762378 Jan 18, 2020
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
demo Fix broken demo Jan 9, 2020
src Split subscription and non-subscription requests into different fns Jan 5, 2020
.eslintrc.js Add eslint Jan 5, 2020
.gitignore
.nowignore Sort out deploy Dec 27, 2019
.prettierrc Get simple typedefs and resolvers working in the browser Dec 27, 2019
Readme.md Update docs in preparation for a release Jan 5, 2020
demo.gif Get simple typedefs and resolvers working in the browser Dec 27, 2019
now.json Sort out deploy Dec 27, 2019
package.json 0.0.16 Jan 18, 2020
rollup.config.js Add pluging to analyze bundle size during dev Dec 30, 2019
types.d.ts Better types and more static type guards Jan 18, 2020

Readme.md

serverless-graphql

Run GraphQL on any schema directly from anywhere without the need for a graphql server.

Supports Query, Mutation and Subscription types.

This is super experimental.

Installation

npm install @zeusdeux/serverless-graphql

You can use other package managers as well.

API

makeExecutableSchema

This function accepts an options object with two properties for now and returns a GraphQLSchema.

Return type

GraphQLSchema

e.g.,

import { makeExecutableSchema } from '@zeusdeux/serverless-graphql'

const typeDefs = gql`
  type Query {
    hello: String!
  }
`

const resolvers = {
  Query: {
    hello: () => 'world!'
  }
}

const schema = makeExecutableSchema({ typeDefs, resolvers })

// this schema can then be consumed by other packages
// such as apollo-server, etc to run a dev server for example.

options.typeDefs

These are the type definitions of your GraphQL server.

e.g.,

type Query {
  authors: [Author]!
  author(name: String!): Author!
  books: [Book]!
}

type Author {
  name: String!
  age: Int!
  addedAt: String!
  books: [Book]!
}

type Book {
  name: String!
  publisher: String!
  publishedYear: Int!
  authors: [Author!]
}

input AuthorInput {
  name: String!
  age: Int!
}

input BookInput {
  name: String!
  publisher: String!
  publishedYear: Int!
}

type Mutation {
  addAuthor(author: AuthorInput!): Author!
  addBookForAuthor(authorName: String!, book: BookInput): Book!
}

options.resolvers

This an object contain the resolvers for the types provided in the typeDefs. If any resolver is missing, the default resolver is used.

The resolver function can also be async and use the async/await syntax.

If a resolver throws an error, the query execution is aborted and the error is returned to the user in the errors key of the response.

A working example matching the schema above is provided below.

const AuthorDB = {}
const BookDB = {}

const delay = ms => {
  return new Promise(res => {
    setTimeout(res, ms)
  })
}

const resolvers = {
  Query: {
    authors() {
      return Object.values(AuthorDB)
    },
    author(_, { name }) {
      if (name in AuthorDB) {
        return AuthorDB[name]
      }

      throw new Error(`No author named ${name} found in database`)
    },
    books() {
      return Object.values(BookDB)
    }
  },

  // Author: {}, // not providing this lets us use the default resolvers

  Mutation: {
    addAuthor(_, { author }) {
      const { name, age } = author

      // add author if it doesn't exist in DB
      if (!(name in AuthorDB)) {
        AuthorDB[name] = {
          name,
          age,
          addedAt: new Date().toISOString(),
          books: []
        }
      }

      return AuthorDB[name]
    },

    async addBookForAuthor(_, { authorName, book: { name: bookName, publisher, publishedYear } }) {
      await delay(2500)

      if (authorName in AuthorDB) {
        const author = AuthorDB[authorName]
        const authorHasBook = !!author.books.filter(book => book.name === bookName).length
        const bookIdx = `${bookName}:${publisher}:${publishedYear}`
        const book = BookDB[bookIdx]

        if (authorHasBook) {
          return book
        }

        if (book) {
          book.authors.push(author)
          BookDB[bookIdx] = book
        } else {
          BookDB[bookIdx] = {
            name: bookName,
            publisher,
            publishedYear,
            authors: [author]
          }
        }

        author.books.push(BookDB[bookIdx])

        return BookDB[bookIdx]
      }

      throw new Error(`Author with name ${name} not found`)
    }
  }
}

getQueryRunner

This function accepts the same options as makeExecutableSchema or an object containing a GraphQLSchema i.e., { schema: GraphQLSchema }. It returns an object with two properties, graphql and subscribe.

Return type

type GetQueryRunnerResult<T = { [key: string]: any }> = {
  graphql: (request: string | QueryRunnerOptions) => Promise<ExecutionResult<T>>
  subscribe: (request: string | QueryRunnerOptions) => Promise<AsyncIterable<ExecutionResult<T>>>
}

type QueryRunnerOptions = {
  req: string
  variables?: { [key: string]: any }
  root?: any
  context?: any
  operationToRun?: string
}

type ExecutionResult<T> = {
  errors?: ReadonlyArray<GraphQLError>
  data?: T | null
}

Both functions returned by getQueryRunner can be provided with the operation as a string or as a string assigned to req property of an object. The object also supports variables which let you use graphql variables, context which is passed to all resolvers, root value which gets passed as the root value to the executor and operationToRun which can be used to select which operation to run if many are provided in the request (req) string.

The functions on the returned object are explained below.

graphql

This function accepts a GraphQL request to execute on the given schema or schema generated from { typeDefs, resolvers } (i.e., the args to getQueryRunner). It should only be used for queries and mutations.

It returns a Promise which resolves to an object containing data and errors properties, one of which is usually populated. data maps to the resolved response. errors contains errors that occured during execution. This maps to the ExecutionResult type shown above.

e.g.,

import { makeExecutableSchema, getQueryRunner, gql } from '@zeusdeux/serverless-graphql'

const typeDefs = gql`
  type Query {
    hello: String!
  }
`

const resolvers = {
  Query: {
    hello: () => 'world!'
  }
}

const { graphql: runQuery } = getQueryRunner({ typeDefs, resolvers })

runQuery('{ hello }').then(({ data }) => console.log(data)) // logs {hello: "world!"}

// or

const schema = makeExecutableSchema({ typeDefs, resolvers })
const { graphql: runQuery } = getQueryRunner({ schema })

runQuery('{ hello }').then(({ data }) => console.log(data)) // logs {hello: "world!"}

An example with the variables option being used

import { getQueryRunner, gql } from '@zeusdeux/serverless-graphql'

const typeDefs = gql`
  type Query {
    hello: String!
  }

  type Mutation {
    getTime(iso: Boolean = false): String
  }
`

const resolvers = {
  Query: {
    hello: () => 'world!'
  },
  Mutation: {
    getTime(_, { iso }) {
      const t = new Date()
      return iso ? t.toISOString() : t.getTime()
    }
  }
}

const { graphql: runQuery } = getQueryRunner({ typeDefs, resolvers })

runQuery({
  req: gql`
    mutation whatTimeIsIt($iso: Boolean) {
      currentServerTime: getTime(iso: $iso)
    }
  `,
  variables: {
    iso: true
  }
}).then(({ data }) => console.log(data)) // logs server time in ISO format

subscribe

This function accepts a GraphQL subscription request to execute on the given schema or schema generated from { typeDefs, resolvers } (i.e., the args to getQueryRunner). It should only be used for subscription requests.

It returns a Promise that resolves to an async iterable that can be consumed using the new for await...of syntax for example.

The values generated by the async iterable have the same shape and type as those returned by the graphql function mentioned above. I.e., Promises that resolve to { data?, errors? } aka ExecutionResult.

e.g.,

import { getQueryRunner, gql } from '@zeusdeux/serverless-graphql'

const delay = ms => {
  return new Promise(res => {
    setTimeout(res, ms)
  })
}

const typeDefs = gql`
  type Subscription {
    getNumber: Int!
  }
`

const resolvers = {
  Subscription: {
    getNumber: {
      subscribe: () => generateNumbers()
    }
  }
}

async function* generateNumbers() {
  let x = 1
  while (x < 5) {
    // synthetic delay
    await delay(1000)

    // whatever is yielded is passed as the root value
    // to the resolve function next to the subscribe.
    // In this example, there isn't one which means
    // GraphQL uses the default resolver which in
    // this case is resolve(root) = root.getNumber
    yield {
      getNumber: x++
    }
  }
}

async function main() {
  const { subscribe } = getQueryRunner({ typeDefs, resolvers })

  for await (let x of await subscribe('subscription { getNumber }')) {
    console.log('Number:', x) // prints Number: 1, Number: 2, and so on
  }
}

main().catch(err => console.log.bind('😱', err))

gql

An alias for String.raw. This lets prettier format the type definitions for you automatically.

Packages published

This repo publishes a umd package with its external dependencies, graphql and graphql/utilities, all bundled up together as a UMD bundle for ease of usage from a browser.

For usage from an application, it does not bundle up any of its external dependencies (graphql and graphql/utilities). It instead marks graphql (which includes graphql/utilities) as a peer dependency under peerDependencies. It is up to the consumer of this package to install the peerDependencies in their application. This is because, at any given time, only one instance of graphql can be installed otherwise it starts throwing when it's functions are consumed.

UNPKG

UNPKG serves umd/serverless-gql.umd.js which is minified and contains the dependencies within.

CJS

Usage of this package using require or in CJS environments consumes dist/serverless-gql.cjs.js.

ESM

In the context of an ESM based codebase, bundlers can choose to consume dist/serverless-gql.esm.js which is a build with import/export statements kept intact for better tree-shaking to enable dead code elimination.

Development

  1. Fork (if not the author) and clone the repository
  2. cd in the directory and run npm i
  3. npm install the peerDependencies of this package if you will be building the code (to publish, test, etc)
  4. Make your changes as required on a well named branch
  5. Open a PR against the origin repository
You can’t perform that action at this time.