Server-Side Rendering (SSR)
Server-Side Rendering (SSR) means rendering front-end frameworks(for example React, Vue, Solid, etc) to html
in Node.js(Server Side), and hydrating the rendered html
on the client.
This document describes how to built a SSR application on top of Farm from scratch.
Example Projectsâ
Farm provides a list of SSR examples for popular frameworks:
Project Structureâ
A SSR typical application often have the following source file structure:
.
âââ index.html
âââ farm.config.ts
âââ farm.config.server.ts
âââ server.js
âââ src
âââ index-client.tsx
âââ index-server.tsx
âââ main.tsx
index.html
: Entry html of the application that running on the client(browser)farm.config.ts
: farm config that builds the project to clientfarm.config.server.ts
: Farm config that builds the project to Node.js(server)server.js
: Server script that should be deployed for productionsrc/index-client.tsx
: Client entry scriptssrc/index-server.tsx
: Server entry scriptssrc/main.tsx
: Application code shared for both client and server
index.html
need to reference index-client.tsx
and include a placeholder where the server-rendered markup
should injected:
<div id="root"><div>app-html-to-replace</div></div>
<script src="./src/index-client.tsx"></script>
You should replace <div>app-html-to-replace</div>
to the server-rendered markup
.
We have to build the SSR application twice, one for client
(browser) and one for server
(Node.js). So farm.config.ts
and farm.config.server.ts
are needed, we'll discuss the details in later sections.
Setting up Dev Serverâ
For above example, farm.config.ts
is used to build the project for browser and setting up DevServer for server rendering. The content of farm.config.ts
normally would be:
import path from 'path';
import { defineConfig } from '@farmfe/core';
export default defineConfig({
compilation: {
input: {
index_client: './index.html'
},
output: {
path: './build'
},
},
server: {
hmr: true,
cors: true,
middlewares: [
// register a middleware that render the application on the server,
// inject server rendered markup and return final index.html
(server) => {
server.app().use(async (ctx, next) => {
await next();
// handle index.html or SPA fallback
if (ctx.path === '/' || ctx.status === 404) {
// loading the server entry, and render it by ctx.path
const render = await import(path.join(process.cwd(), 'dist', 'index.js')).then(
(m) => m.default
);
const renderedHtml = render(ctx.path);
// get compiled index.html content from server.getCompiler()
// The html is compiled for client with all client bundles injected
const template = server
.getCompiler()
.resource('index_client.html')
.toString();
// replace the placeholder to rendered markup and return it as html
const html = template.replace(
'<div>app-html-to-replace</div>',
renderedHtml
);
ctx.body = html;
ctx.type = 'text/html';
ctx.status = 200;
}
console.log('ctx.path outer', ctx.path);
});
}
]
},
plugins: ['@farmfe/plugin-react', '@farmfe/plugin-sass']
});
In above example, a middleware
is required for rendering the application to markup and serve it as html. Normal workflow for SSR in the middleware
:
- Load compiled server entry: A
index-server
entry which exports arender
function is required, we need toimport(server_entry_path)
to get therender
function. - Get compiled client index.html: All client bundles and Farm runtime are injected to
index.html
, so the client canhydrate
successfully. - Replace the placeholder to rendered markup: Replace the placeholder and return the
final html
.
In this example, we are building a SPA
SSR application with if (ctx.path === '/' || ctx.status === 404) {
, if you are building a MPA
SSR application, guard ctx.path
to your pages.
Building for Node.jsâ
farm.config.server.ts
is used to build the project for Node.js, producing the compiled server entry which can be used to rendering the application to markup on the server side.
import { defineConfig } from '@farmfe/core';
export default defineConfig({
compilation: {
input: {
index: './src/index-server.tsx'
},
output: {
path: './dist',
targetEnv: 'node'
}
},
plugins: [
[
'@farmfe/plugin-react',
{
refresh: false,
development: false
}
],
'@farmfe/plugin-sass'
]
});
For farm.config.server.ts
, we set input
to server entry and output.targetEnv
to node
.
By default, Farm compiles server entry script to esm
, if you want to compile it to cjs
, try set output.format
.
Develop SSR Projectâ
You have start compilation for both client
and server
, for example, you may have following scripts
in package.json
:
{
"name": "@farmfe-examples/react-ssr",
"scripts": {
"start": "farm start",
"start:server": "farm watch --config farm.config.server.mjs",
}
}
When starting your SSR project, you should run both npm run start
and npm run start:server
in different terminal.
Building for Productionâ
You have build both client
and server
, for example, you may add following command to scripts
:
{
"name": "@farmfe-examples/react-ssr",
"scripts": {
"start": "farm start",
"start:server": "farm watch --config farm.config.server.mjs",
"build": "farm build",
"build:server": "farm build --config farm.config.server.mjs"
}
}
When building for production, you should run both npm run build
and npm run build:server
, the client bundles will be emitted to build
dir, and the server bundles will be emitted to dist
dir.
For production, you need a node server
for rendering
and serving rendered html
, in this example, we provide a server.js
as production server:
import path from 'node:path';
import { fileURLToPath } from 'node:url'
import fsp from 'fs/promises';
import express from 'express';
function resolve(p) {
const __dirname = path.dirname(fileURLToPath(import.meta.url));
return path.resolve(__dirname, p);
}
// create a node production server
async function createServer() {
let app = express();
// serve the client builds as static assets, you can also deploy client builds to CDN or separate dev server as you wish.
app.use(express.static(resolve('build')));
// listen '/' route, you can replace it to the routes you use.
app.use('/', async (req, res) => {
let url = req.originalUrl;
try {
let template;
let render;
// load client html
template = await fsp.readFile(resolve('build/index_client.html'), 'utf8');
// load server render function
render = await import(resolve('dist/index.js')).then(
(m) => m.default
);
// render the application to markup
const markup = render(url);
let html = template.replace(
'<div>app-html-to-replace</div>',
markup
);
// return the rendered html with client bundles, the client bundles hydrate the server rendered markup and make it interactive
res.setHeader('Content-Type', 'text/html');
return res.status(200).end(html);
} catch (error) {
console.log(error.stack);
res.status(500).end(error.stack);
}
});
return app;
}
// create and listen the server
createServer().then((app) => {
app.listen(3000, () => {
console.log('HTTP server is running at http://localhost:3000');
});
});
We use express
as server here, but you can use whatever server frameworks you want. The rendering process are the same:
- Loading client compiled html
- Loading
render
function from compiled server script - Call
const markup = render(url)
function to get the server-side rendered markup of your application - Replace the
placeholder
inclient index.html
to therendered markup
and return the replaced html as final result
Static-Site Generation(SSG)â
The same flow of SSG is the same as SSR, the difference is SSG that emits to replaced html
to the final resources. Example scripts for SSG:
// load client html
const template = await fsp.readFile(resolve('build/index_client.html'), 'utf8');
// load server render function
const render = await import(resolve('dist/index.js')).then(
(m) => m.default
);
const pages = renderDirEntry('src/pages');
for (const page of pages) {
// render the application to markup
const markup = render(url);
const html = template.replace(
'<div>app-html-to-replace</div>',
markup
);
// emit the static generated page, for example writing it to disk
emitPage(page, html);
}