Why donate
API Explorer
Upgrade guide
NEW!
The quasar.config file
Convert project to CLI with Vite
Browser Compatibility
Supporting TypeScript
Directory Structure
Commands List
CSS Preprocessors
Routing
Lazy Loading - Code Splitting
Handling Assets
Boot Files
Prefetch Feature
API Proxying
Handling Vite
Handling process.env
State Management with Pinia
Linter
Testing & Auditing
Developing Mobile Apps
Ajax Requests
Opening Dev Server To Public
Quasar CLI with Vite - @quasar/app-vite
SSR Webserver

Notice that your generated /src-ssr contains a file named server.js. This file defines how your SSR webserver is created, managed and served. You can start listening to a port or provide a handler for your serverless infrastructure to use. It’s up to you.

Anatomy

The /src-ssr/server.js file is a simple JavaScript/Typescript file which boots up your SSR webserver and defines what how your webserver starts & handles requests and what it exports (if exporting anything).

WARNING

The /src-ssr/server.js file is used for both DEV and PROD, so please be careful on how you configure it. To differentiate between the two states you can use process∙env∙DEV and process∙env∙PROD.

/**
 * Runs in Node context.
 */

/**
 * Make sure to yarn add / npm install (in your project root)
 * anything you import here (except for express and compression).
 */
import express from 'express'
import compression from 'compression'
import {
  defineSsrCreate,
  defineSsrListen,
  defineSsrClose,
  defineSsrServeStaticContent,
  defineSsrRenderPreloadTag
} from '#q-app/wrappers'

/**
 * Create your webserver and return its instance.
 * If needed, prepare your webserver to receive
 * connect-like middlewares.
 *
 * Can be async: defineSsrCreate(async ({ ... }) => { ... })
 */
export const create = defineSsrCreate((/* { ... } */) => {
  const app = express()

  // attackers can use this header to detect apps running Express
  // and then launch specifically-targeted attacks
  app.disable('x-powered-by')

  // place here any middlewares that
  // absolutely need to run before anything else
  if (process.env.PROD) {
    app.use(compression())
  }

  return app
})

/**
 * You need to make the server listen to the indicated port
 * and return the listening instance or whatever you need to
 * close the server with.
 *
 * The "listenResult" param for the "close()" definition below
 * is what you return here.
 *
 * For production, you can instead export your
 * handler for serverless use or whatever else fits your needs.
 *
 * Can be async: defineSsrListen(async ({ app, devHttpsApp, port }) => { ... })
 */
export const listen = defineSsrListen(({ app, devHttpsApp, port }) => {
  const server = devHttpsApp || app
  return server.listen(port, () => {
    if (process.env.PROD) {
      console.log('Server listening at port ' + port)
    }
  })
})

/**
 * Should close the server and free up any resources.
 * Will be used on development only when the server needs
 * to be rebooted.
 *
 * Should you need the result of the "listen()" call above,
 * you can use the "listenResult" param.
 *
 * Can be async: defineSsrClose(async ({ listenResult }) => { ... })
 */
export const close = defineSsrClose(({ listenResult }) => {
  return listenResult.close()
})

const maxAge = process.env.DEV
  ? 0
  : 1000 * 60 * 60 * 24 * 30

/**
 * Should return a function that will be used to configure the webserver
 * to serve static content at "urlPath" from "pathToServe" folder/file.
 *
 * Notice resolve.urlPath(urlPath) and resolve.public(pathToServe) usages.
 *
 * Can be async: defineSsrServeStaticContent(async ({ app, resolve }) => {
 * Can return an async function: return async ({ urlPath = '/', pathToServe = '.', opts = {} }) => {
 */
export const serveStaticContent = defineSsrServeStaticContent(({ app, resolve }) => {
  return ({ urlPath = '/', pathToServe = '.', opts = {} }) => {
    const serveFn = express.static(resolve.public(pathToServe), { maxAge, ...opts })
    app.use(resolve.urlPath(urlPath), serveFn)
  }
})

const jsRE = /\.js$/
const cssRE = /\.css$/
const woffRE = /\.woff$/
const woff2RE = /\.woff2$/
const gifRE = /\.gif$/
const jpgRE = /\.jpe?g$/
const pngRE = /\.png$/

/**
 * Should return a String with HTML output
 * (if any) for preloading indicated file
 */
export const renderPreloadTag = defineSsrRenderPreloadTag((file/* , { ssrContext } */) => {
  if (jsRE.test(file) === true) {
    return `<link rel="modulepreload" href="${file}" crossorigin>`
  }

  if (cssRE.test(file) === true) {
    return `<link rel="stylesheet" href="${file}" crossorigin>`
  }

  if (woffRE.test(file) === true) {
    return `<link rel="preload" href="${file}" as="font" type="font/woff" crossorigin>`
  }

  if (woff2RE.test(file) === true) {
    return `<link rel="preload" href="${file}" as="font" type="font/woff2" crossorigin>`
  }

  if (gifRE.test(file) === true) {
    return `<link rel="preload" href="${file}" as="image" type="image/gif" crossorigin>`
  }

  if (jpgRE.test(file) === true) {
    return `<link rel="preload" href="${file}" as="image" type="image/jpeg" crossorigin>`
  }

  if (pngRE.test(file) === true) {
    return `<link rel="preload" href="${file}" as="image" type="image/png" crossorigin>`
  }

  return ''
})

TIP

Remember that whatever the listen() function returns (if anything) will be exported from your built dist/ssr/index.js. You can return your ssrHandler for a serverless architecture should you need it.

Parameters

export function <functionName> ({
  app, port, isReady, ssrHandler,
  resolve, publicPath, folders, render, serve
}) => {

Detailing the Object:

{
  app,     // Expressjs app instance (or whatever you return from create())

  port,    // on production: process∙env∙PORT or quasar.config file > ssr > prodPort
           // on development: quasar.config file > devServer > port

  isReady, // Function to call returning a Promise
           // when app is ready to serve clients

  ssrHandler, // Prebuilt app handler if your serverless service
              // doesn't require a specific way to provide it.
              // Form: ssrHandler (req, res, next)
              // Tip: it uses isReady() under the hood already

  // all of the following are the same as
  // for the SSR middlewares (check its docs page);
  // normally you don't need these here
  // (use a real SSR middleware instead)
  resolve: {
    urlPath(path)
    root(arg1, arg2),
    public(arg1, arg2)
  },
  publicPath, // String
  folders: {
    root,     // String
    public    // String
  },
  render(ssrContext),
  serve: {
    static({ urlPath, pathToServe, opts }),
    error({ err, req, res })
  }
}

Usage

WARNING

  • If you import anything from node_modules, then make sure that the package is specified in package.json > “dependencies” and NOT in “devDependencies”.
  • This is usually not the place to add middlewares (but you can do it). Add middlewares by using the SSR Middlewares instead. You can configure SSR Middlewares to run only for dev or only for production too.

Replacing express.js

You can replace the default Express.js Node server with any other connect API compatible one. Just make sure to yarn/npm install its package first.

src-ssr/server.js

import { defineSsrCreate } from '#q-app/wrappers'
import connect from 'connect'
import compression from 'compression'

export const create = defineSsrCreate((/* { ... } */) => {
  const app = connect()

  // place here any middlewares that
  // absolutely need to run before anything else
  if (process.env.PROD) {
    app.use(compression())
  }

  return app
})

Listen on a port

This is the default option that you get when adding SSR support in a Quasar CLI project. It starts listening on the configured port (process∙env∙PORT or quasar.config file > ssr > prodPort).

src-ssr/server.js

export const listen = defineSsrListen(({ app, devHttpsApp, port }) => {
  const server = devHttpsApp || app;
  return server.listen(port, () => {
    if (process.env.PROD) {
      console.log('Server listening at port ' + port);
    }
  });
})

Serverless

If you have a serverless infrastructure, then you generally need to export a handler instead of starting to listen to a port.

Say that your serverless service requires you to name export a variable called handler. Then what you’d need to do is:

src-ssr/server.js

import { defineSsrListen } from '#q-app/wrappers'
export const listen = defineSsrListen(({ app, devHttpsApp, port }) => {
  if (process.env.DEV) {
    // for dev, start listening on the created server
    const server = devHttpsApp || app;
    return server.listen(port, () => {
      // we're ready to serve clients
    })
  }
  else { // in production
    // return an object with a "handler" property
    // that the server script will named-export
    return { handler: app }
  }
})

Please note that the provided ssrHandler is a Function of form: (req, res, next) => void. Should you require to export a handler of form (event, context, callback) => void then you will most likely want to use the serverless-http package (see below).

Example: serverless-http

You will need to manually yarn/npm install the serverless-http package.

src-ssr/server.js

import { defineSsrListen } from '#q-app/wrappers'
import serverless from 'serverless-http'

export const listen = defineSsrListen(({ app, devHttpsApp, port }) => {
  if (process.env.DEV) {
    // for dev, start listening on the created server
    const server = devHttpsApp || app;
    return server.listen(port, () => {
      // we're ready to serve clients
    })
  }
  else { // in production
    return { handler: serverless(ssrHandler) }
  }
})

Example: Firebase function

src-ssr/server.js

import { defineSsrListen } from '#q-app/wrappers'
import * as functions from 'firebase-functions'

export const listen = defineSsrListen(({ app, devHttpsApp, port }) => {
  if (process.env.DEV) {
    // for dev, start listening on the created server
    const server = devHttpsApp || app;
    return server.listen(port, () => {
      // we're ready to serve clients
    })
  }
  else { // in production
    return {
      handler: functions.https.onRequest(ssrHandler)
    }
  }
})