github.com/f/graphql.js @v0.6.8

graphql.js: lightweight graphql client

Lightest GraphQL client with intelligent features.

---

Originally inspired by Robert Mosolgo's blog post

Features

• Too small, 4k gzipped.
• No dependencies, plain vanilla JavaScript.
• Plug & Play.
• Isomorphic.
• Runs on most browsers.
• You don't need to install Node.js ecosystem on your computer.
• Query merging to reduce request number.

Overview

GraphQL is based on a very simple HTTP transaction, which sends a request to an endpoint with query and variables.

Many libraries require complex stacks to make that simple request. In any project you don't use React, Relay, you'll need a simpler client which manages your query and makes a simple request.

// Connect...
var graph = graphql("/graphql")

// Prepare...
graph.fragment({
  user: `on User {
    id,
    name
  }`
})

const allUsers = graph(`query { allUsers { ...user } }`)

const createUser = graph(`mutation (@autodeclare) {
  createUser($firstName, $lastName) { ...user }
}`)

await createUser({
  firstName: "John",
  lastName: "Doe"
})

const users = await allUsers()

console.log(users)
// {
//   "allUsers": [{ "id": 1, "name": "John Doe" }]
// }

Installation

You can download graphql.js directly, or you can use Bower or NPM.

CDN

<script src="https://github.com/f/graphql.js/raw/v0.6.8/cdn.jsdelivr.net/npm/graphql.js@0.6.6/graphql.min.js"></script>

Download for Browser

• Development Version - 12kb
• Production Version - 6kb

Using NPM

npm install graphql.js --save

# or

yarn add graphql.js

Using with Rails Asset Pipeline

You can use GraphQL.js with Rails Asset Pipeline using graphqljs-rails.

Using

GraphQL.js is isomorphic. You can use it in both browser and Node.js.

Use in Browser

<script src="https://github.com/f/graphql.js/raw/v0.6.8/path/to/graphql.js"></script>

Use in Node.js

var graphql = require('graphql.js')

Or using import

import graphql from 'graphql.js'

Connection

Create a simple connection to your GraphQL endpoint.

var graph = graphql("http://localhost:3000/graphql", {
  method: "POST", // POST by default.
  headers: {
    // headers
    "Access-Token": "some-access-token"
    // OR "Access-Token": () => "some-access-token"
  },
  fragments: {
    // fragments, you don't need to say `fragment name`.
    auth: "on User { token }",
    error: "on Error { messages }"
  }
})

Executing Queries and Mutations

graph will be a simple function that accepts query and variables as parameters.

graph(`query ($email: String!, $password: String!) {
  auth(email: $email, password: $password) {
    ... auth # if you use any fragment, it will be added to the query.
    ... error
  }
}`, {
  email: "john@doe.com",
  password: "my-super-password"
}).then(function (response) {
  // response is originally response.data of query result
  console.log(response)
}).catch(function (error) {
  // response is originally response.errors of query result
  console.log(error)
})

Prepare Query for Lazy Execution

You can prepare queries for lazy execution. This will allow you to reuse your queries with different variables without any hassle.

var login = graph(`query ($email: String!, $password: String!) {
  auth(email: $email, password: $password) {
    ... on User {
      token
    }
  }
}`)

// Call it later...
login({
  email: "john@doe.com",
  password: "my-super-password"
})

Direct Execution with .run and ES6 Template Tag

If your query doesn't need any variables, it will generate a lazy execution query by default. If you want to run your query immediately, you have three following options:

// 1st option. create and run function.
graph(`...`)()
graph.query(`...`)()
graph.mutate(`...`)()
//...

// 2nd option. create and run function with `run` method.
graph.run(`...`)
graph.query.run(`...`)
graph.mutate.run(`...`)

// 3rd option. create and run function with template tag.
graph`...`
graph.query`...`
graph.mutate`...`

I don't recommend using this. Using it too much may break DRY. Use lazy execution as much as possible.

Prefix Helpers

You can prefix your queries by simply calling helper methods: .query, .mutate or .subscribe

var login = graph.query(`($email: String!, $password: String!) {
  auth(email: $email, password: $password) {
    ... on User {
      token
    }
  }
}`)

var increment = graph.mutate`increment { state }`
var onIncrement = graph.subscribe`onIncrement { state }`

Automatic Declaring with @autodeclare or {declare: true}

Declaring primitive-typed (String, Int, Float, Boolean) variables in query were a little bothering to me. That's why I added an @autodeclare keyword or {declare: true} setting to the processor. It detects types from the variables and declares them in query automatically.

var login = graph.query(`(@autodeclare) {
  auth(email: $email, password: $password) {
    ... on User {
      token
    }
  }
}`)

login({
  email: "john@doe.com", // It's String! obviously.
  password: "my-super-password" // It is, too.
})

This will create following query:

query ($email: String!, $password: String!) {
  auth(email: $email, password: $password) {
    ... on User {
      token
    }
  }
}

You can also pass {declare: true} option to the .query, .mutate and .subscribe helper:

var login = graph.query(`auth(email: $email, password: $password) {
  ... on User {
    token
  }
}`, {declare: true})

This will also create the same query above.

Detecting IDs

Variable names with matching /_id/i pattern will be declared as ID type. Following examples will be declared as IDs:

• id: 1 will be declared as $id: ID!
• post_id: "123af" will be declared as $post_id: ID!
• postID: 3 will be declared as $postID: ID!
• postId: 4 will be declared as $postId: ID!

You can disable auto ID declaration by adding an ! to the end of the variable name:

• id!: 1 will be declared as $id: Int!
• post_id!: "123af" will be declared as $post_id: String!

And, explicitly given types are prioritized.

• postID!CustomId: 3 will be declared as $postID: CustomId!
• postId!UUID: 4 will be declared as $postId: UUID!

var userById = graph.query(`(@autodeclare) {
  user(id: $id) {
    email
  }
}`)

userById({
  id: 15
})

The example above will generate following query:

query ($id: ID!) {
  user(id: $id) {
    email
  }
}

Solving Integer and Float Problem

Let's say you have a rating query that accepts an argument with a Float argument named rating. GraphQL.js will declare 10 value as Integer since it casts using value % 1 === 0 ? 'Int' : 'Float' check.

var rate = graph.query(`(@autodeclare) {
  rating(rating: $rating) {
    rating
  }
}`)

rate({
  rating: 10
})

In this case, you must use ! mark to force your type to be Float as below:

rate({
  "rating!Float": 10
})

This will bypass the casting and declare rating as Float.

Advanced Auto Declaring

Beside you can pass {declare: true} to helpers:

graph.query("auth(email: $email, password: $password) { token }", {declare: true})

Also you can enable auto declaration to run by default using alwaysAutodeclare setting.

var graph = graphql("http://localhost:3000/graphql", {
  alwaysAutodeclare: true
})

After you enable alwaysAutodeclare option, your methods will try to detect types of variables and declare them.

// When alwaysAutodeclare is true, you don't have to pass {declare: true} option.

graph.query("auth(email: $email, password: $password) { token }")

Auto Declaring Custom Types

You can define custom types when defining variables by using a simple "variable!Type" notation. It will help you to make more complex variables:

var register = graph.mutate(`(@autodeclare) {
  userRegister(input: $input) { ... }
}`)

register({
  // variable name and type.
  "input!UserRegisterInput": { ... }
})

This will generate following query:

mutation ($input: UserRegisterInput!) {
  userRegister(input: $input) { ... }
}

Fragments

Fragments make your GraphQL more DRY and improves reusability. With .fragment method, you'll manage your fragments easily.

Simple Fragments

While constructing your endpoint, you can predefine all of your fragments.

var graph = graphql("/graphql", {
  fragments: {
    userInfo: `on User { id, name, surname, avatar }`
  }
})

And you can use your fragments in your queries. The query will pick your fragments and will add them to the bottom of your query.

graph.query(`{ allUsers { ...userInfo } }`)

Nested Fragments

You can nest your fragments to keep them organized/namespaced.

var graph = graphql("/graphql", {
  fragments: {
    user: {
      info: `on User { id, name, surname, avatar }`
    }
  }
})

Accessing them is also intuitive:

graph.query(`{ allUsers { ...user.info } }`)

Using Fragments in Fragments

You can reuse fragments in your fragments.

graph.fragment({
  user: "on User {name, surname}",
  login: {
    auth: "on User {token, ...user}"
  }
})

Lazy Fragments

You can also add fragments lazily. So you can use your fragments more modular.

// Adds a profile fragment
graph.fragment({
  profile: `on User {
    id
    name(full: true)
    avatar
  }`
})

var allUsers = graph.query(`{
  allUsers {
    ... profile
  }
}`)

allUsers().then(...)

Also you can add nested fragments lazily, too:

graph.fragment({
  login: {
    error: `on LoginError {
      reason
    }`
  }
})

graph.fragment({
  something: {
    error: `on SomeError {
      messages
    }`
  }
})

graph.query(`{ login {... login.error } }`)
graph.query(`{ something {... something.error } }`)

Getting Fragments by Path

You can call fragment string by using .fragment method. You have to pass path string to get the fragment.

graph.fragment('login.error')

This will give you the matching fragment code:

fragment login_error on LoginError {
  reason
}

Using Fragments in Tag Query

You can use fragments lazily using ES6 template tag queries.

var userProfileToShow = graph.fragment('user.profile')

graph`query { ... ${userProfileToShow} }`

Query Building

You can create queries using .ql ES6 template tag.

// Add some fragments...
graph.fragment({
  username: {
    user: `on User {
      username
    }`,
    admin: `on AdminUser {
      username,
      administrationLevel
    }`
  }
})

// Get any fragment with its path...
var admin = graph.fragment('username.admin')

// Build your query with using fragment paths or dynamic template variables.
var query = graph.ql`query {
  ...username.user
  ...${admin}
}`

// Use query anywhere...
$.post("/graphql", {query: query}, function (response) { ... })

graph.ql will generate this query string:

query {
  ... username_user
  ... username_admin
}

fragment username_user on User {
  username
}

fragment username_admin on AdminUser {
  username,
  administrationLevel
}

Query Merging: Merge Multiple Queries into One Request

This GIF shows a before/after case to make an example how query merging changes the performance.

graphql.js supports query merging that allows you to collect all the requests into one request.

Assume we've these queries on server, define them just like before we do:

var fetchPost = graph.query(`{
  post(id: $id) {
    id
    title
    text
  }
}`)

var fetchComments = graph.query(`{
  commentsOfPost: comments(postId: $postId) {
    comment
    owner {
      name
    }
  }
}`)

Normally, we make requests as following:

var postId = 123

// This will send a request.
fetchPost({ id: postId }).then(function (response) {
  console.log(response.post)
})

// This also will send a request.
fetchComments({ postId: postId }).then(function (response) {
  console.log(response.commentsOfPost)
})

This will make two requests:

Use .merge(mergeName, variables) command to put them into a merge buffer:

var postId = 123

// This won't send a request.
fetchPost.merge('buildPage', { id: postId }).then(function (response) {
  console.log(response.post)
})

// This also won't send a request.
fetchComments.merge('buildPage', { postId: postId }).then(function (response) {
  console.log(response.commentsOfPost)
})

These will create a buffer with buildPage name, and append the queries to that buffer. You need to use commit(mergeName) to merge the buffer and send to the server, the response will be consolidated:

// This will send a merged request:
graph.commit('buildPage').then(function (response) {
  // All base fields will be in response return.
  console.log(response.post)
  console.log(response.commentsOfPost)
})

And this will create only one request:

This will create the following merged query generated by graphql.js:

```graphql query ($merge024533__id: ID!, $merge141499__postId: ID!) { merge024533_post: { post(id: $merge024533__id) {