Use class-based controllers to handle websocket events. Helps to organize your code using websockets in classes.
-
Install
socket-controllers
:npm install socket-controllers
-
Install
reflect-metadata
shim:npm install reflect-metadata
and make sure to import it in a global place, like app.ts:
import 'reflect-metadata';
-
Install a DI container, for example
typedi
;npm install typedi
-
Create a file
MessageController.ts
import { OnConnect, SocketController, ConnectedSocket, OnDisconnect, MessageBody, OnMessage, } from 'socket-controllers'; import {Service} from 'typedi'; // Only if you are using typedi @SocketController() @Service() // Only if you are using typedi export class MessageController { @OnConnect() connection(@ConnectedSocket() socket: any) { console.log('client connected'); } @OnDisconnect() disconnect(@ConnectedSocket() socket: any) { console.log('client disconnected'); } @OnMessage('save') save(@ConnectedSocket() socket: any, @MessageBody() message: any) { console.log('received message:', message); console.log('setting id to the message and sending it back to the client'); message.id = 1; socket.emit('message_saved', message); } }
-
Create a file
app.ts
import 'es6-shim'; // this shim is optional if you are using old version of node import 'reflect-metadata'; // this shim is required import { SocketControllers } from 'socket-controllers'; import { MessageController } from './MessageController'; import {Container} from 'typedi'; // Only if you are using typedi new SocketControllers({ port: 3001, container: Container, controllers: [MessageController], });
-
Now you can send
save
websocket message using websocket-client.
Controller action marked with @OnConnect()
decorator is called once new client connected.
Controller action marked with @OnDisconnect()
decorator is called once client disconnected.
Controller action marked with @OnDisconnecting()
decorator is called when the client is disconnecting, before the disconnect event.
import { SocketController, OnConnect, OnDisconnect, OnDisconnecting } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnConnect()
save() {
console.log('client connected');
}
@OnDisconnect()
save() {
console.log('client disconnected');
}
@OnDisconnecting()
save() {
console.log('client is disconnecting');
}
}
To get connected socket instance you need to use @ConnectedSocket()
decorator.
import { SocketController, OnMessage, ConnectedSocket } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@ConnectedSocket() socket: any) {
socket.emit('save_success');
}
}
To get received message body use @MessageBody()
decorator:
import { SocketController, OnMessage, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@MessageBody() message: any) {
console.log('received message: ', message);
}
}
If you specify a class type to parameter that is decorated with @MessageBody()
,
socket-controllers will use class-transformer to create instance of the given class type with the data received in the message.
To disable this behaviour you need to specify { transformOption: { transform: false ] }
in SocketControllerOptions when creating a server.
You can define an index to get multiple parameters from the socket event.
import { SocketController, OnMessage, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@MessageBody({index: 0}) param1: any, @MessageBody({index: 1}) param2: any) {
console.log('received message: ', message1);
console.log('received message: ', message2);
}
}
To get received message ack use @MessageAck()
decorator:
import { SocketController, OnMessage, MessageAck, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@MessageBody() message: any, @MessageAck() ack: Function) {
console.log('received message: ', message);
ack('callback message');
}
}
note: ack must be the last parameter in
emit
, otherwise it will benull
To get received query parameter use @SocketQueryParam()
decorator.
import { SocketController, OnMessage, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@SocketQueryParam('token') token: string) {
console.log('authorization token from query parameter: ', token);
}
}
To get connected client id use @SocketId()
decorator.
import { SocketController, OnMessage, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@SocketId() id: string) {}
}
import { SocketController, OnMessage, MessageBody } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
save(@SocketIO() io: any) {
// now you can broadcast messages to specific rooms or namespaces using io instance
}
}
You can use @EmitOnSuccess
decorator:
import { SocketController, OnMessage, EmitOnSuccess } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
@EmitOnSuccess('save_successfully')
save() {
// after this controller executed "save_successfully" message will be emitted back to the client
}
}
If you return something, it will be returned in the emitted message data:
import { SocketController, OnMessage, EmitOnSuccess } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
@EmitOnSuccess('save_successfully')
save() {
// after this controller executed "save_successfully" message will be emitted back to the client with message object
return {
id: 1,
text: 'new message',
};
}
}
You can also control what message will be emitted if there is error/exception during execution:
import { SocketController, OnMessage, EmitOnSuccess, EmitOnFail } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('save')
@EmitOnSuccess('save_successfully')
@EmitOnFail('save_error_range', {errorType: RangeError})
@EmitOnFail('save_error_type', {errorType: TypeError})
@EmitOnFail('save_error')
save() {
if (1 === 1) {
throw new Error('One is equal to one! Fatal error!');
}
return {
id: 1,
text: 'new message',
};
}
}
In this case save_error
message will be sent to the client with One is equal to one! Fatal error!
error message.
The order is important when defining multiple @EmitOnFail()
decorators, the first matching errorType will be served
Sometimes you may want to not emit success/error message if returned result is null or undefined.
In such cases you can use @SkipEmitOnEmptyResult()
decorator.
import { SocketController, OnMessage, EmitOnSuccess, EmitOnFail, SkipEmitOnEmptyResult } from 'socket-controllers';
@SocketController()
export class MessageController {
@OnMessage('get')
@EmitOnSuccess('get_success')
@SkipEmitOnEmptyResult()
get(): Promise<Message[]> {
return this.messageRepository.findAll();
}
}
In this case if findAll will return undefined, get_success
message will not be emitted.
If findAll will return array of messages, they will be emitted back to the client in the get_success
message.
This example also demonstrates Promises support.
If promise returned by controller action, message will be emitted only after promise will be resolved.
If you need to create and configure socket.io server manually,
you can pass it to the SocketControllers
constructor.
Here is example of creating socket.io server and configuring it with express:
import 'reflect-metadata'; // this shim is required
import { SocketControllers } from 'socket-controllers';
import { Server } from 'socket.io';
import { Container } from 'typedi'; // Only if you are using typedi
const app = require('express')();
const server = require('http').Server(app);
const io = new Server(server);
server.listen(3001);
app.get('/', function (req: any, res: any) {
res.send('hello express');
});
io.use((socket: any, next: Function) => {
console.log('Custom middleware');
next();
});
new SocketControllers({io, container: Container});
You can load all controllers in once from specific directories, by specifying array of directories via options in
createSocketServer
or useSocketServer
:
import 'reflect-metadata'; // this shim is required
import { SocketControllers } from 'socket-controllers';
import { Container } from 'typedi'; // Only if you are using typedi
new SocketControllers({
port: 3000,
container: Container,
controllers: [__dirname + '/controllers/*.js'],
}); // registers all given controllers
To listen to messages only of the specific namespace you can mark a controller with namespace:
@SocketController('/messages')
export class MessageController {
// ...
}
Also you can use dynamic namespace, like express router
patterns:
@SocketController('/messages/:userId')
export class MessageController {
// ...
}
Middlewares are the functions passed to the socketIo.use
method.
Middlewares allows you to define a logic that will be executed each time client connected to the server.
To create your middlewares use @Middleware
decorator:
import { Middleware, MiddlewareInterface } from 'socket-controllers';
@Middleware()
export class CompressionMiddleware implements MiddlewareInterface {
use(socket: any, next: (err?: any) => any) {
console.log('do something, for example get authorization token and check authorization');
next();
}
}
You can limit middlewares to namespaces providing either a string
, RegExp
or Array<string | RegExp>
to the namespace
parameter:
import { Middleware, MiddlewareInterface } from 'socket-controllers';
@Middleware({namespace: '/test'})
export class CompressionMiddleware implements MiddlewareInterface {
use(socket: any, next: (err?: any) => any) {
console.log('do something, for example get authorization token and check authorization');
next();
}
}
Controllers and middlewares should be loaded:
import 'reflect-metadata';
import { SocketControllers } from 'socket-controllers';
import { MessageController } from './MessageController';
import { MyMiddleware } from './MyMiddleware'; // here we import it
import { Container } from 'typedi'; // Only if you are using typedi
const server = new SocketControllers({
port: 3000,
container: Container,
controllers: [MessageController],
middlewares: [MyMiddleware],
});
Also you can load them from directories. Also you can use glob patterns:
import 'reflect-metadata';
import { SocketControllers } from 'socket-controllers';
import { Container } from 'typedi'; // Only if you are using typedi
const server = new SocketControllers({
port: 3000,
container: Container,
controllers: [__dirname + '/controllers/**/*.js'],
middlewares: [__dirname + '/middlewares/**/*.js'],
});
socket-controllers
supports a DI container out of the box. You can inject your services into your controllers and
middlewares. Container must be setup during application bootstrap.
Here is example how to integrate socket-controllers with typedi:
import 'reflect-metadata';
import { SocketControllers } from 'socket-controllers';
import { Container } from 'typedi';
// create and run socket server
const server = new SocketControllers({
port: 3000,
container: Container,
controllers: [__dirname + '/controllers/*.js'],
middlewares: [__dirname + '/middlewares/*.js'],
});
That's it, now you can inject your services into your controllers:
@Service()
@SocketController()
export class MessageController {
constructor(private messageRepository: MessageRepository) {}
// ... controller actions
}
Note: TypeDI won't create instances for unknown classes since 0.9.0, you have to decorate your Class as a
Service()
as well.
You can enable scoped controllers by providing a scopedContainerGetter
function in SocketServerOptions. This function should return a new container that will be used to instantiate the controller and its dependencies.
You will get a new instance for each event in the controller.
The scopedContainerGetter
function receives the SocketEventContext
.
The scopedContainerDisposer
function receives the container instance you created with scopedContainerGetter
after the socket action is finished. Use this function to dispose the container if needed.
import 'reflect-metadata';
import { SocketControllers, SocketEventContext } from 'socket-controllers';
import { Container, ContainerInstance, Token } from "typedi";
const myDiToken = new Token();
// create and run socket server
const server = new SocketControllers({
port: 3000,
container: Container,
scopedContainerGetter: (args: SocketEventContext) => {
const container = Container.of(YOUR_REQUEST_CONTEXT);
container.set(myDiToken, 'MY_VALUE');
return container;
},
scopedContainerDisposer: (container: ContainerInstance) => {
container.dispose();
},
controllers: [__dirname + '/controllers/*.js'],
middlewares: [__dirname + '/middlewares/*.js'],
});
Interceptors allow you to wrap your event handlers in higher order functions. With interceptors you can add logging or modify the incoming or outgoing data for event handlers.
import {
SocketController,
OnMessage,
EmitOnSuccess,
EmitOnFail,
SkipEmitOnEmptyResult,
UseInterceptor,
MessageBody
} from 'socket-controllers';
const interceptor: InterceptorInterface = {
use: (ctx: SocketEventContext, next: () => any) => {
ctx.messageArgs[0] = 'modified message from controller - ' + ctx.messageArgs[0];
const resp = next();
return 'modified response from controller - ' + resp; // modified response from controller - modified response from method - reponse
},
};
@Service()
class Interceptor implements InterceptorInterface {
async use(ctx: SocketEventContext, next: () => any) {
ctx.messageArgs[0] = 'modified message from method - ' + ctx.messageArgs[0];
const resp = await next();
return 'modified response from method - ' + resp; // modified response from method - reponse
}
}
@SocketController()
@UseInterceptor(interceptor)
export class MessageController {
@OnMessage('get')
@EmitOnSuccess('get_success')
@SkipEmitOnEmptyResult()
@UseInterceptor(Interceptor)
get(@MessageBody() message: string): Promise<Message[]> {
console.log(message); // modified message from controller - modified message from method - original message
return 'response';
}
}
Interceptors are executed in order of definition, starting with the controller interceptors.
Signature | Description |
---|---|
`@SocketController(namespace?: string\ | Regex)` |
@OnMessage(messageName: string) |
Registers controller's action to be executed when socket receives message with given name. |
@OnConnect() |
Registers controller's action to be executed when client connects to the socket. |
@OnDisconnect() |
Registers controller's action to be executed when client disconnects from the socket. |
@OnDisconnecting() |
Registers controller's action to be executed when client is disconnecting from the socket. |
@ConnectedSocket() |
Injects connected client's socket object to the controller action. |
@SocketIO() |
Injects socket.io object that initialized a connection. |
@MessageBody() |
Injects received message body. |
@SocketQueryParam(paramName: string) |
Injects query parameter from the received socket request. |
@SocketId() |
Injects socket id from the received request. |
@SocketRequest() |
Injects request object received by socket. |
@SocketRooms() |
Injects rooms of the connected socket client. |
@NspParams() |
Injects dynamic namespace params. |
@NspParam(paramName: string) |
Injects param from the dynamic namespace. |
@Middleware() |
Registers a new middleware to be registered in the socket.io. |
@EmitOnSuccess(messageName: string) |
If this decorator is set then after controller action will emit message with the given name after action execution. It will emit message only if controller succeed without errors. If result is a Promise then it will wait until promise is resolved and emit a message. |
@EmitOnFail(messageName: string) |
If this decorator is set then after controller action will emit message with the given name after action execution. It will emit message only if controller throw an exception. If result is a Promise then it will wait until promise throw an error and emit a message. |
@SkipEmitOnEmptyResult() |
Used in conjunction with @EmitOnSuccess and @EmitOnFail decorators. If result returned by controller action is null or undefined then messages will not be emitted by @EmitOnSuccess or @EmitOnFail decorators. |
Take a look on samples in ./sample for more examples of usage.
- If you are interested to create controller-based express or koa server use routing-controllers module.
- If you need to use dependency injection in use typedi module.