⚡ Vite + Express

@vitejs integration module for @expressjs

If you like vite-express make sure to leave a ⭐ star!
You can also

• 📦 Installation & usage
• 🚚 Shipping to production
• 📖 Multipage Apps
• 🤖 Transforming HTML
• 🤔 How does it work?
• 📝 Documentation

🕑 Click here to see the changelog

With Vite you can easily bootstrap your project and just start working without figuring everything out. That's great for front-end apps, but when you want to include server-side into the mix, things get quite complicated. Thanks to vite-express you can just as easily start writing full-stack app in seconds.

import express from "express";
import ViteExpress from "vite-express";

const app = express();

app.get("/message", (_, res) => res.send("Hello from express!"));

ViteExpress.listen(app, 3000, () => console.log("Server is listening..."));

You can also bind into the express app, to be able to do things such as specifying custom host address or creating your own server instance (e.g., when you want to use the https: protocol).

import express from "express";
import ViteExpress from "vite-express";

const app = express();

const server = app.listen(3000, "0.0.0.0", () =>
  console.log("Server is listening...")
);

ViteExpress.bind(app, server);

⚡ vite-express takes care of

• injecting necessary middleware to serve static files from your express server
• managing unhandled routes to make client-side routing possible

The only thing that is left to you is to code! 🎉

The easiest way to setup a Vite Express app is to use 🏗️ create-vite-express package

1. Run the CLI from your terminal

   yarn create vite-express

2. Follow the prompts to configure your project using your favourite framework.

3. Open app folder, install dependencies and run the app in development mode

   cd YOUR_APP_NAME
   yarn
   yarn dev

4. Open your browser at http://localhost:3000

5. Change the client code and see the beauty of HMR in action!

Congrats, you've just created your first vite-express app! 🎉 Happy hacking!

Alternatively you can use create-vite package to setup the client and then add an express server to it if your favourite framework isn't supported by create-vite-express.

1. Start by creating Vite project

   yarn create vite

2. Follow the prompts to configure your project using your favourite framework.

3. Install express and vite-express packages

   yarn add express vite-express

4. Create a server script inside project root directory

   //e.g server.js
   import express from "express";
   import ViteExpress from "vite-express";
   
   const app = express();
   
   app.get("/message", (_, res) => res.send("Hello from express!"));
   
   ViteExpress.listen(app, 3000, () => console.log("Server is listening..."));

5. Run the express script

   node server.js

6. Open your browser at http://localhost:3000

7. Change the client code and see the beauty of HMR in action!

Congrats, you've just created your first vite-express app! 🎉 Happy hacking!

By default vite-express runs in development mode, when server uses Vite's Dev Server in middleware mode (which means that no separate Vite process is running) utilizing the power of HMR and native browser modules. This is not suitable for production as described here, so in production we want to serve static files that Vite spits out during it's build process. That's why you need to invoke vite build command first. Then you need to run your app in production mode.

You have these options to achieve that

• Run the code with NODE_ENV=production variable, either by inlining it with the command

  NODE_ENV=production node server.ts

  Or by using dotenv or other envs tool.

• Use ViteExpress.config() and set mode to production

  import express from "express";
  import ViteExpress from "vite-express";
  
  const app = express();
  ViteExpress.config({ mode: "production" })
  
  app.get("/message", (_, res) => res.send("Hello from express!"));
  
  ViteExpress.listen(app, 3000, () => console.log("Server is listening..."));

vite-express will always try to use Vite to resolve vite.config.*s file. But in production environment it is very possible that it will not be installed. When Vite package cannot be used, vite-express falls back to reading the config file as a plain text file, trying to extract root, base and outDir values. If you don't want to ship vite.config to production you can specify custom inline config using ViteExpress.config({ inlineViteConfig: ... }). Remember that when using inline config there are two sources of truth, so you need to make sure vite.config used by Vite, and inline config used by vite-express are synchronized.

Order of config file resolve process:

1. When inlineViteConfig is set to some value it will be used as a configuration object, completely ignoring vite.config file.

   import express from "express";
   import ViteExpress from "vite-express";
   
   ViteExpress.config({ 
      inlineViteConfig: { 
         base: "/admin", 
         build: { outDir: "out" }
      } 
   });
   
   ViteExpress.listen(express(), 3000);

   You can specify any combination of those values, it will be merged with the default values to get resulting configuration object.

   Default config values are:

   • root = process.cwd()
   • base = /
   • outDir = dist

2. When inlineViteConfig is not defined, vite-express will try to use Vite to resolve the config.

3. If the package is not present, vite.config.*s file will be loaded as a plain file and config values will be extracted using plain text manipulation methods. That's why root, base and outDir need to be defined as json valid values: strings.

4. If config file cannot be used to extract the values, defaults defined above will be used.

With Vite you can have multiple HTML entry points as described here. From v0.11.0 this feature is also supported by vite-express. Your app will automatically load correct index.html file when you navigate to path, just like in Vite.

Suppose you have the following source code structure:

├── package.json
├── vite.config.js
├── index.html
├── main.js
└── nested
    ├── index.html
    ├── nested.js
    └── subroute
        └─ index.html

When you navigate to any client-side route vite-express tries to find an index.html file matching request path as much as it is possible. What is means is that if you go to /nested/subroute/my/secret/path, files that vite-express will try to render are:

1. nested/subroute/my/secret/index.html
2. nested/subroute/my/index.html
3. nested/subroute/index.html
4. nested/index.html
5. index.html

In this case nested/subroute/index.html would be picked because that file exist, but in another case it would fall back to nested/index.html, and in case of another failure to root index.html. As soon as nested/subroute/my/secret/index.html will exist in the file structure, it can be used for this request.

Why nested/subroute/my/secret/path/index.html isn't considered? Because it needs to have a trailing /. That's how Vite is doing it. If you want, you can always use express redirect to add this trailing slash at the end.

app.get('^nested/subroute/my/secret/path$', (_, res) => {
   res.redirect("nested/subroute/my/secret/path/")
});

⚠️ Please remember that you still have to configure Vite so that it will resolve all these files correctly in its build step.

You can specify transformer function that takes two arguments - HTML as a string and Request object - and returns HTML as a string with any string related transformation applied. It can be used to inject your custom metadata on the server-side. It can also be an async function if you need to retrieve the data to inject from a database or some other remote source.

This transformer function is invoked right before sending the HTML to the client in the index-serving middleware that vite-express injects at the end of the middleware stack.

Imagine a situation in which your index.html file looks like this

<html>
   <head>
      <!-- placeholder -->
   </head>
   <body>
      <div id="root"></div>
   </body>
</html>

You can then use custom transformer function to replace the HTML comment with any string you like. It can be a custom meta tag. You can use request object to extract additional information about request such as requested page.

import express from "express";
import ViteExpress from "vite-express";
import someMiddleware from "./some/middleware";

const app = express()

function transformer(html: string, req: express.Request) {
   return html.replace(
      "<!-- placeholder -->", 
      `<meta name="custom" content="${req.baseUrl}"/>`
   )
}

app.use(someMiddleware())

ViteExpress.config({ transformer })
ViteExpress.listen(app, 3000);

The HTML served to the client will then look something like this

<html>
   <head>
     <meta name="custom" content="/"/>
   </head>
   <body>
      <div id="root"></div>
   </body>
</html>

The way vite-express works is quite simple. As soon as you invoke ViteExpress.listen:

• Static files serving middleware is injected at the end of express middlewares stack, but you can change that by using ViteExpress.static middleware to precisely describe at what point do you want to serve static files. This middleware takes care of serving static files from the Vite Dev Server in development mode and in production mode express.static is used instead.

• A GET routes handler get("*") is registered at the end of middleware stack to handle all the routes that were unhandled by you. We do this to ensure that client-side routing is possible.

Because ViteExpress.listen is an async function, in most cases it doesn't matter when you invoke it, but it is generally the best to do it at the end of file to avoid get("*") handler overriding your routes.

Used to pass in configuration object with each key optional.

ViteExpress.config({ /*...*/ });

type ViteConfig = {
   root?: string;
   base?: string;
   build?: { outDir: string };
   server?: { hmr?: boolean | HmrOptions };
}

You can read more about HmrOptions in Vite docs.

Used to inject necessary middlewares into the app and start listening on defined port. Should replace app.listen() in your base express application. Due to its async nature can be invoked at any time but should generally be invoked at the end to avoid interfering with other middlewares and route handlers.

• app - express application returned from invoking express()
• port: number - port that server will be listening on
• callback?: () => void - function that will be invoked after server starts listening

Returns the same http.Server that is returned by express when running app.listen()

const app = express();
const httpServer = ViteExpress.listen(app, 3000, () => console.log("Server is listening!"));

When HTTP server has been closed using the close() method, vite-express also closes the Vite dev server asynchronously. After that, a vite:close event is emitted so you can listen to that event to have a guarantee that all server resources were freed.

httpServer.on("vite:close", handleViteClose);
httpServer.close();

Used to inject necessary middleware into the app, but does not start the listening process. Should be used when you want to create your own http/https server instance manually e.g. when you use socket.io library. Same as listen, can be invoked at any time because it is async, but it is advised to invoke it when you already registered all routes and middlewares, so that it can correctly hook into the express app.

• app - express application returned from invoking express()
• server: http.Server | https.Server - server instance that is returned when invoking http.createServer
• callback?: () => void - function that will be invoked after Vite dev server is started and vite-express injects all middleware

const app = express();
const server = http.createServer(app).listen(3000, () => {
   console.log("Server is listening!")
});
ViteExpress.bind(app, server);

Used as a typical express middleware to indicate to vite-express the exact moment when you want to register static serving logic. You can use this method to prevent some of your request blocking middleware, such as authentication/authorization, from blocking files coming from your server, which would make displaying for example login page impossible because of blocked html, styles and scripts files.

Another use-case is when you want to override options that vite-express passes to express.static middleware in production mode. In development mode it doesn't have any effect.

Example:

import express from "express"
import yourAuthMiddleware from "some/path"

const app = express()

app.use(ViteExpress.static({ maxAge: "1d" }))
app.use(yourAuthMiddleware())

app.get("/", ()=> /*...*/ )

ViteExpress.listen(app, 3000,  () => console.log("Server is listening!"))

You should use it when the default behaviour of serving static files at the end of middleware chain doesn't work for you because you block requests in some way.

Used when you want to build the app to production programically. It is adviced to use vite build command, but can be freely used in some edge scenarios (e.g. in some automation scripts) as it does the same thing.

ViteExpress.build();

Useful in case you want to know the Vite configuration parameters currently in use. For example, for writing a console.log message with some of that information in the listen callback.

Example:

import express from "express";
import ViteExpress from "vite-express";

const app = express();

const port = 3000;
const host = "localhost";

const server = app.listen(port, host);

ViteExpress.bind(app, server, async () => {
  const { root, base } = await ViteExpress.getViteConfig();
  console.log(`Serving app from root ${root}`);
  console.log(`Server is listening at http://${host}:${port}${base}`);
});

MIT