Prerender

Client-rendered SPAs are invisible to most crawlers — Googlebot has been less of a problem in recent years, but Facebook, WhatsApp, Slack, Twitter, and most non-Google indexers still expect server-rendered HTML. @mongez/vite can emit a prerender.php and add a rewrite rule to the .htaccess so bot user-agents get a pre-rendered version while humans hit your SPA directly.

Default behaviour

mongezVite();   // preRender: false (opt-in)

No prerender.php is emitted. No crawler routing is added to the .htaccess.

Opting in

mongezVite({
  htaccess: true,     // required — the rewrite rule lives here
  preRender: {
    url: "https://render.mentoor.io",
    crawlers: "Google-Site-Verification|Googlebot|facebook|crawl|WhatsApp|bot|Slack|Twitter",
    delay: 5000,
    cache: false,
  },
});

Options

Key	Type	Default	What it does
url	string	— (required)	The external prerender service to fetch from. Receives the visited URL as a query param and returns rendered HTML.
crawlers	string	Google-Site-Verification|Googlebot|facebook|crawl|WhatsApp|bot|Slack|Twitter|bot	Pipe-separated regex alternation of user-agent fragments. Matched case-insensitively against User-Agent header.
delay	number	5000	Milliseconds to tell the prerender service to wait before rendering — lets your SPA finish loading data before HTML is captured.
cache	boolean	false	When truthy, the generated PHP caches each rendered URL under <outDir>/cache/<sha1(url)>.html.

The generated prerender.php

<?php

function get_content($URL) {
  $hashedUrl = sha1($URL);
  $cachedPagesDirectory = __DIR__ . '/cache';
  $cachedFile = $cachedPagesDirectory . '/' . $hashedUrl . '.html';

  if (file_exists($cachedFile)) {
    $file = file_get_contents($cachedFile);
    if ($file) return $file;
  }

  $ch = curl_init();
  curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
  curl_setopt($ch, CURLOPT_URL, $URL);
  $data = curl_exec($ch);
  curl_close($ch);

  if (!file_exists($cachedPagesDirectory)) {
    mkdir($cachedPagesDirectory, 0777, true);
  }
  file_put_contents($cachedFile, $data);
  return $data;
}

$prerenderUrl = '<your prerender URL>';
$url = $_SERVER['SCRIPT_URI'] ?? ...;
$userAgent = $_SERVER['HTTP_USER_AGENT'];

$params = [
  'url' => $url,
  'delay' => <delay>,
  'cache' => <cache>,
  '__agent' => $userAgent,
];

$url = "$prerenderUrl?" . http_build_query($params);
$content = get_content($url);
echo $content;

It always writes to the cache directory regardless of the cache option — cache: true just enables the read path. If you don’t want any caching, delete the cache directory after each deploy or wrap the script.

The htaccess rewrite

When preRender is set, the .htaccess includes:

# Prerender
RewriteCond %{HTTP_USER_AGENT} .*(<crawlers>).* [NC]
RewriteCond %{REQUEST_URI} !^(/public)
RewriteRule (.*) prerender.php [L,QSA]

So any request whose User-Agent matches the crawler list (and doesn’t start with /public) is routed to prerender.php instead of falling through to the SPA’s index.html.

Worked example

mongezVite({
  htaccess: true,
  preRender: {
    url: "https://render.mentoor.io",
    cache: true,
  },
});

After vite build:

dist/
├── .htaccess          ← contains the prerender rewrite
├── prerender.php      ← proxies to https://render.mentoor.io
├── cache/             ← created lazily on first crawler hit
└── ...

A request from Googlebot:

1. Hits dist/.
2. The .htaccess matches User-Agent: Googlebot... and rewrites to prerender.php.
3. prerender.php checks cache/<sha1>.html. On cache miss it fetches https://render.mentoor.io?url=...&delay=5000&cache=1&__agent=Googlebot/....
4. The response is cached and returned.

A request from a real user agent (Mozilla/...):

1. Hits dist/.
2. The .htaccess does not match the crawler list and falls through to the SPA rewrite (index.html).
3. The SPA renders normally.

Sharp edges

• preRender.url is required when preRender is enabled. defaultOptions.ts has preRender: false, and there is no implicit default for url. If you set preRender: {} without url, generateHtaccess() throws a descriptive build-time error (preRender.url is required when preRender is enabled). Always set url explicitly.
• Requires htaccess: true. The whole generateHtaccess step short-circuits when htaccess: false, so neither .htaccess nor prerender.php is emitted. Pair preRender with htaccess: true explicitly.
• No Apache means no prerender. This pipeline is Apache-specific (.htaccess + PHP). Nginx / Caddy hosts need a different strategy.
• The cache directory is created lazily by the PHP. Make sure the web server has write permission to <outDir>/cache/.
• __agent is forwarded to the service so its rendering can detect the original crawler. Whether your service uses it depends on its implementation.