This package is deprecated in favor of TEO.
GraphQL Joker is the ultimate GraphQL scaffolding tool.
It automates coding process to save your precious time, enhance your work and life experience. In other words, GraphQL Joker write code for you with commands you specified.
With GraphQL Joker, you can create a full-fledged backend server with your complex app logic and running API in less than 3 minutes.
- Motivation
- Design Concept
- Installation
- Create an GraphQL Project
- Generate Resources
- Generate Uploader
- Integrate with Existing Project
- Issues and Helps
- Roadmap
- License
When working on GraphQL projects, we need to define database schema and GraphQL twice. We need to create resolvers for the standardized API. A lot of copying and pasting are going on. It's not elegant to copy code around and find-replace all occurrences. It's also error prone. And sometimes causing unnoticeable errors which wastes time.
Wouldn't be nice if we could have a tool just like Ruby on Rails' scaffold tool to generate code for us?
GraphQL Joker is designed to provide features at least Ruby on Rails scaffold tool has. Aka, generate boilerplate business logic code as much as possible for you.
However, unlike Ruby on Rails, GraphQL Joker is not a full-fledged framework and will never provide a framework for you. It focus on business logic generation. Although GraphQL Joker also has project generation feature, it's trying to hook up the industry standard and battle-tested libraries and components together for you. And it's configurable. To split features into small core chunks and make them combinable and adaptable is a good practice and especially popular in node.js ecosystem, GraphQL Joker embraces this practice. That's what makes GraphQL Joker outstanding and what makes GraphQL Joker really a flexible and configurable scaffolding tool.
GraphQL Joker is a general command line tool, thus you should install it globally.
npm install -g graphql-joker
To create an GraphQL project, use joker app
command.
joker app my-new-app
This will generate your app in 'my-new-app' folder. If you don't specify app name, the app will be created at your current working directory.
Options:
--port
On which port this app is listening on.--git-init
Automatically run 'git init' after project generated.--skip-install
Do not install dependencies.--eslint-config
Changing the default eslint config.--main
Changing the entry filename.
To change default eslint config being used:
joker app my-new-app --eslint-config=your-config
To automatically run git init
:
joker app my-new-app --git-init
API resource generation is the core feature of GraphQL Joker. It's syntax is rather simple and extensible. It follows this basic style:
joker resource ModelName[/optionalPluralVariableName] \
primitiveField[[:Type[typeModifiers]]:defaultValue]... \
referenceField[[:ReferenceType[typeModifiers]]:foreignKey]...
This arguments specification is obscure to see. Let's see some examples.
Let's say you have a model named user, and user has a name, an age and also a list of posts. And you have a model named post, it has title, content and author. Just type like this:
joker resource User name:String age:Int posts:[Post]:author
joker resource Post title:String content:String author:User
Here we specified our first model 'User', with following fields:
name
which is a Stringage
which is an Intposts
which is a list of Posts through foreign key named author
We defined our second model named 'Post', with following fields:
title
which is a Stringcontent
which is also a Stringauthor
which references to User
This creates six files in total, three for User and three for Post. The three files are mongoose model, GraphQL schema and GraphQL resolver.
The autogenerated models/User.js looks like this:
const mongoose = require('mongoose');
const { Schema } = mongoose;
const userSchema = new Schema({
name: String,
age: Number
}, {
timestamps: true,
collection: 'users'
});
module.exports = mongoose.model('User', userSchema);
The autogenerated schemas/User.gql looks like this:
type User {
_id: ID!
name: String
age: Int
posts: [Post]
createdAt: Date
updatedAt: Date
}
input UserInput {
name: String
age: Int
}
type Query {
user(_id: ID!): User
users: [User]
}
type Mutation {
createUser(input: UserInput): User
updateUser(_id: ID!, input: UserInput): User
deleteUser(_id: ID!): User
}
The autogenerated resolvers/User.js looks like this:
module.exports = {
User: {
async posts(root, _, { Post }) {
return await Post.find({ author: root._id });
}
},
Query: {
async user(root, { _id }, { User }) {
return await User.findById(_id);
},
async users(root, { _ }, { User }) {
return await User.find();
}
},
Mutation: {
async createUser(root, { input }, { User }) {
return await User.create(input);
},
async updateUser(root, { _id, input }, { User }) {
return await (await User.findById(_id)).set(input).save();
},
async deleteUser(root, { _id }, { User }) {
return await (await User.findById(_id)).remove();
}
}
};
Besides your schema definition, 5 API are created for you. Those are:
users
query all usersuser
query a user by idcreateUser
create a new userupdateUser
modify an existing userdeleteUser
delete an existing user
Now you can CRUD your resources through API.
GraphQL Joker supports a wide range of primitive types:
String
string typeInt
integer typeFloat
float typeBoolean
bool typeDate
date typeEnum
enum type, the type specifier has a different syntaxFile
upload typem the type specifier has a different syntaxMixed
mixed type includes string, int, float, boolean, date, array and objects
When you are defining a field with type mentioned above, GraphQL Joker will treat them as primitive types. When you refer to a type that is not included in the list, GraphQL Joker will treat it as a referecing to another model.
joker resource User disabled:Boolean name:String description:Mixed spouse:User
In the above example, obviously disabled
, name
and description
are
primitive types. spouse
is a reference type which references to User
.
Surround a type with a pair of [], you get an array of that type, for example:
joker resource User spouse:User friends:[User] favoriteSayings:[String]
The field friends
is an array of User
s. And the field favoriteSayings
is
an array of String
s.
There are several ways to implement your own reference types.
The simplest case is one-to-one relation ship.
joker resource User address:Address
joker resource Address user:User:address
In this case, we save the reference into user model, and on address model, we use the foreign key on user model to fetch the user value.
We have two ways to implement this relationship.
joker resource User posts:[Post]:owner
joker resource Post user:User:owner
This is the most common case. We save the reference on the 'many' side, and fetch on the 'many' side model.
joker resource User posts:[Post]
joker resource Post user:User:[posts]
In this case, we are saving the references on the 'one' side, and on 'many' side, we use a pair of [] to indicate it's an array. Be careful of performance when you are doing this way.
In simple cases, we can just do like this.
joker resource User courses:[Course]
joker resource Course users:[User]:[courses]
If there are tons of records, then you may want to use association table.
joker resource Favorite user:User course:Course
joker resource User courses:[Course]:Favorite
joker resource Course users:[User]:Favorite
In this case, we specified a relationship that is have many ... through ...
To create an uploading field, use ...Uploader
as type name. See the following
example:
joker resource User avatar:AvatarUploader
To create an uploader, see Generate Uploader
In the real world practices, fields should be validated. For example, You may want a user's email to match designated format and to be required and unique. You can specify type modifiers.
joker resource User 'email:String/.*@.*\..*/!$'
In the above example, /.*@.*\..*/
means that this field matches this regexp,
!
means required, and $
means unique.
Existing type modifiers includes:
!
required^
index$
unique/regexp/
string only, matches the regexp or not<=n
max for number types, maxlength for string type>=n
min for number types, minlength for string type
You can specify default value to a primitive field with the following syntax.
joker resource Post 'title:String!:Untitled' 'lastUpdate:Date!:`Date.now`'
Here, title's default value is 'Untitled'
, and lastUpdate's default value is
Date.now
. It's a calculated default value, so surround with a pair of
back ticks.
To create nested structure, use the following syntax:
joker resource User posts:[{ title:String content:String comments:[{ \
commenter:User content:String }] }] email:String password:String settings:{ \
sms:Boolean email:Boolean pushNotification:Boolean }
Specify type as {
or [{
, you are going into a nested context. All field
defined after this goes into the nested structure. Use plain }
and }]
tokens
to jump out the nesting context.
To create enum fields, use enum syntax like this:
joker resource User 'gender:Enum(male,female)!'
GraphQL Joker supports reusable nestables and referencing them.
joker nestable Address line1:String line2:String country:String region:String
joker resource User address:addressSchema name:String
Specify the lowercase nestable name append by 'Schema', joker will treat the type as a subschema reference.
If you mistakenly generated something or you spell something wrongly, use the 'destroy' command to delete the autogenerated files. Just append destroy with the original command, it automatically destroys the generated content.
joker destroy resource User name:String
To generate an uploader, use joker uploader
command.
joker uploader FileUploader extends AliOSSUploader bucket=your-bucket-name region=your-region
This generates an base file uploader for you.
GraphQL Joker is designed to be a generic tool. It does not require a project to be a GraphQL Joker project.
Create a file called .jokerrc.json
in project's root directory. And filling it
like this:
{
"schemaDir": "graphql",
"resolverDir": "graphql",
"test": false
}
GraphQL Joker will generate schema files and resolver files into graphql directory, and will not generate unit tests.
GraphQL Joker is not mature yet. If you find anything uncomfortable or confuses you. Any discuss, issue and pull request are welcome.
GraphQL Joker is an ambitious project and it still has a long way to go.
- Version 0.8
- Basic sequelize support
- Version 0.9
- CLI user experience
- use eslint to transform user generated code if available
- configurability
- dependencies reliability
- Version 0.10
- query filter, sorting and pagination feature