Lightest GraphQL client with intelligent features.
Originally inspired by Robert Mosolgo's blog post
- 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.
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, $lastName) { ...user }
}`)
await createUser({
firstName: "John",
lastName: "Doe"
})
const users = await allUsers()
console.log(users)
// {
// "allUsers": [{ "id": 1, "name": "John Doe" }]
// }
You can download graphql.js
directly, or you can use Bower or NPM.
<script src="//cdn.jsdelivr.net/npm/graphql.js@0.6.6/graphql.min.js"></script>
npm install graphql.js --save
# or
yarn add graphql.js
You can use GraphQL.js with Rails Asset Pipeline using graphqljs-rails.
GraphQL.js is isomorphic. You can use it in both browser and Node.js.
<script src="/path/to/graphql.js"></script>
var graphql = require('graphql.js')
Or using import
import graphql from 'graphql.js'
Create a simple connection to your GraphQL endpoint.
var graph = graphql("http://localhost:3000/graphql", {
method: "POST", // POST by default.
headers: {
// headers
"Access-Token": "some-access-token"
// OR "Access-Token": () => "some-access-token"
},
fragments: {
// fragments, you don't need to say `fragment name`.
auth: "on User { token }",
error: "on Error { messages }"
}
})
graph
will be a simple function that accepts query
and variables
as parameters.
graph(`query ($email: String!, $password: String!) {
auth(email: $email, password: $password) {
... auth # if you use any fragment, it will be added to the query.
... error
}
}`, {
email: "john@doe.com",
password: "my-super-password"
}).then(function (response) {
// response is originally response.data of query result
console.log(response)
}).catch(function (error) {
// response is originally response.errors of query result
console.log(error)
})
You can prepare queries for lazy execution. This will allow you to reuse your queries with different variables without any hassle.
var login = graph(`query ($email: String!, $password: String!) {
auth(email: $email, password: $password) {
... on User {
token
}
}
}`)
// Call it later...
login({
email: "john@doe.com",
password: "my-super-password"
})
If your query doesn't need any variables, it will generate a lazy execution query by default. If you want to run your query immediately, you have three following options:
// 1st option. create and run function.
graph(`...`)()
graph.query(`...`)()
graph.mutate(`...`)()
//...
// 2nd option. create and run function with `run` method.
graph.run(`...`)
graph.query.run(`...`)
graph.mutate.run(`...`)
// 3rd option. create and run function with template tag.
graph`...`
graph.query`...`
graph.mutate`...`
I don't recommend using this. Using it too much may break DRY. Use lazy execution as much as possible.
You can prefix your queries by simply calling helper methods: .query
, .mutate
or .subscribe
var login = graph.query(`($email: String!, $password: String!) {
auth(email: $email, password: $password) {
... on User {
token
}
}
}`)
var increment = graph.mutate`increment { state }`
var onIncrement = graph.subscribe`onIncrement { state }`
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: $password) {
... on User {
token
}
}`, {declare: true})
This will also create the same query above.
Variable names with matching /_id/i
pattern will be declared as ID
type. Following examples will be declared as IDs:
id: 1
will be declared as$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
}
}
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
.
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 }")
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 make your GraphQL more DRY and improves reusability. With .fragment
method, you'll
manage your fragments easily.
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 } }`)
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 } }`)
You can reuse fragments in your fragments.
graph.fragment({
user: "on User {name, surname}",
login: {
auth: "on User {token, ...user}"
}
})
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 } }`)
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
}
You can use fragments lazily using ES6 template tag queries.
var userProfileToShow = graph.fragment('user.profile')
graph`query { ... ${userProfileToShow} }`
You can create queries using .ql
ES6 template tag.
// Add some fragments...
graph.fragment({
username: {
user: `on User {
username
}`,
admin: `on AdminUser {
username,
administrationLevel
}`
}
})
// Get any fragment with its path...
var admin = graph.fragment('username.admin')
// Build your query with using fragment paths or dynamic template variables.
var query = graph.ql`query {
...username.user
...${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
}
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: $postId) {
comment
owner {
name
}
}
}`)
Normally, we make requests as following:
var postId = 123
// This will send a request.
fetchPost({ id: postId }).then(function (response) {
console.log(response.post)
})
// This also will send a request.
fetchComments({ postId: postId }).then(function (response) {
console.log(response.commentsOfPost)
})
This will make two requests:
Use .merge(mergeName, variables)
command to put them into a merge buffer:
var postId = 123
// This won't send a request.
fetchPost.merge('buildPage', { id: postId }).then(function (response) {
console.log(response.post)
})
// This also won't send a request.
fetchComments.merge('buildPage', { postId: postId }).then(function (response) {
console.log(response.commentsOfPost)
})
These will create a buffer with buildPage name, and append the queries to that buffer. You need to use commit(mergeName)
to merge the buffer and send to the server, the response will be consolidated:
// This will send a merged request:
graph.commit('buildPage').then(function (response) {
// All base fields will be in response return.
console.log(response.post)
console.log(response.commentsOfPost)
})
And this will create only one request:
This will create the following merged query generated by graphql.js:
query ($merge024533__id: ID!, $merge141499__postId: ID!) {
merge024533_post: {
post(id: $merge024533__id) {
id
title
text
}
}
merge141499_commentsOfPost: {
comments(postId: $merge141499__postId) {
comment
owner {
name
}
}
}
}
And variables will be generated, too:
{
"merge024533__id": 123,
"merge141499__postId": 123
}
The
merge{number}
aliases won't be passed into your responses, since they will be used for initial seperation.
⚠️ Important Restriction: You cannot use multiple root fields using query merging.⚠️ Important Restriction: Autodeclaration is on by default, do not usealwaysAutodeclare: true
.
You can pass debug: true
to options parameter to get a console output looks like following:
[graphql]: POST http://localhost:3000/graphql
QUERY: query ($email: String!, $password: String!) {
auth(email: $email, password: $password) {
.. login_auth
}
}
fragment info on User { hasPin }
fragment login_auth on User { token, ...info }
VARIABLES: {
"email": "john@doe.com",
"password": "123123"
}
sending as form url-data
Create a GraphQLProvider.js
.
import graphql from 'graphql.js';
/* eslint-disable no-underscore-dangle */
export default {
install(Vue, url, options) {
Vue.mixin({
created() {
this._graph = graphql(url, options);
},
});
Object.defineProperty(Vue.prototype, '$graph', {
get() {
return this._graph;
},
});
},
};
And then you can use this with your Vue app:
import Vue from 'vue';
import GraphQLProvider from './GraphQLProvider';
Vue.use(GraphQLProvider, 'http://localhost:3000/graphql', {
headers: {
// headers...
},
});
// ... in your Vue VM
data() {
return {
hello: '',
};
},
methods: {
makeSomeQuery() {
this.$graph.query(`{hello}`).then(response => {
this.hello = response.hello;
});
},
}
As default, GraphQL.js makes a POST request. But you can change the behavior by setting asJSON
.
var graph = graphql("http://localhost:3000/graphql", {
asJSON: true
});
graphql-tag
converts GraphQL query strings to AST. You can use graphql-tag
with GraphQL.js
graph.query(gql`query { name }`)
Using
graphql-tag
will not allow you to use auto declaration and nested fragments syntaxes since these are not valid query syntax for GraphQL but only for this library.
You can change url anywhere with setUrl
method.
var graph = graphql("http://localhost:3000/graphql", {
asJSON: true
});
// Change url
graph.setUrl('http://www.example.com')
// Run query
graph.query(`{ name }`)
A CRUD ToDo app example code to show how to use GraphQL.js. An implementation can be found at f/graphql.js-demo
var graph = graphql("/graphql", {
alwaysAutodeclare: true,
fragments: {
todo: `on Todo {
id
text
isCompleted
}`
}
})
function getTodos() {
return graph.query.run(`allTodos {
...todo
}`)
}
function addTodo(text) {
return graph.mutate(`todoAdd(text: $text) {
...todo
}`, {
text: text
})
}
function setTodo(id, isCompleted) {
return graph.mutate(`todoComplete(
id: $id,
status: $isCompleted
) {
...todo
}`, {
id: id,
isCompleted: isCompleted
})
}
function removeTodo(id) {
return graph.mutate(`todoRemove(
id: $id
) {
...todo
}`, {
id: id
})
}
MIT License
Copyright (c) 2018 Fatih Kadir Akın
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.