fix require cache memory leak

packages/docusaurus/src/common.d.ts

@@ -15,9 +15,13 @@ export type AppRenderResult = {
   collectedData: PageCollectedData;
 };
 
-export type AppRenderer = (params: {
-  pathname: string;
-}) => Promise<AppRenderResult>;
+export type AppRenderer = {
+  render: (params: {pathname: string}) => Promise<AppRenderResult>;
+
+  // It's important to shut down the app renderer
+  // Otherwise Node.js require cache leaks memory
+  shutdown: () => Promise<void>;
+};
 
 export type PageCollectedData = {
   // TODO Docusaurus v4 refactor: helmet state is non-serializable

packages/docusaurus/src/ssg/ssg.ts

@@ -6,7 +6,6 @@
  */
 
 import fs from 'fs-extra';
-import {createRequire} from 'module';
 import path from 'path';
 import _ from 'lodash';
 import evaluate from 'eval';
@@ -19,6 +18,7 @@ import {
   type SSGTemplateCompiled,
 } from './ssgTemplate';
 import {SSGConcurrency, writeStaticFile} from './ssgUtils';
+import {createSSGRequire} from './ssgNodeRequire';
 import type {SSGParams} from './ssgParams';
 import type {AppRenderer, AppRenderResult, SiteCollectedData} from '../common';
 import type {HtmlMinifier} from '@docusaurus/bundler';
@@ -58,6 +58,8 @@ export async function loadAppRenderer({
 
   const filename = path.basename(serverBundlePath);
 
+  const ssgRequire = createSSGRequire(serverBundlePath);
+
   const globals = {
     // When using "new URL('file.js', import.meta.url)", Webpack will emit
     // __filename, and this plugin will throw. not sure the __filename value
@@ -67,7 +69,7 @@ export async function loadAppRenderer({
 
     // This uses module.createRequire() instead of very old "require-like" lib
     // See also: https://github.com/pierrec/node-eval/issues/33
-    require: createRequire(serverBundlePath),
+    require: ssgRequire.require,
   };
 
   const serverEntry = await PerfLogger.async(
@@ -86,7 +88,15 @@ export async function loadAppRenderer({
       `Server bundle export from "${filename}" must be a function that renders the Docusaurus React app.`,
     );
   }
-  return serverEntry.default;
+
+  async function shutdown() {
+    ssgRequire.cleanup();
+  }
+
+  return {
+    render: serverEntry.default,
+    shutdown,
+  };
 }
 
 export function printSSGWarnings(
@@ -191,6 +201,8 @@ export async function generateStaticFiles({
     {concurrency: SSGConcurrency},
   );
 
+  await renderer.shutdown();
+
   printSSGWarnings(results);
 
   const [allSSGErrors, allSSGSuccesses] = _.partition(
@@ -235,7 +247,7 @@ async function generateStaticFile({
 }): Promise<SSGSuccessResult & {warnings: string[]}> {
   try {
     // This only renders the app HTML
-    const result = await renderer({
+    const result = await renderer.render({
       pathname,
     });
     // This renders the full page HTML, including head tags...

packages/docusaurus/src/ssg/ssgNodeRequire.ts

@@ -0,0 +1,49 @@
+/**
+ * Copyright (c) Facebook, Inc. and its affiliates.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+import {createRequire} from 'module';
+
+export type SSGNodeRequire = {
+  require: NodeJS.Require;
+  cleanup: () => void;
+};
+
+// The eval/vm.Script used for running the server bundle need a require() impl
+// This impl has to be relative to the server bundler path
+// This enables the server bundle to resolve relative paths such as:
+// - require('./assets/js/some-chunk.123456.js')
+//
+// Unfortunately, Node.js vm.Script doesn't isolate memory / require.cache
+// This means that if we build multiple Docusaurus localized sites in a row
+// The Node.js require cache will keep growing and retain in memory the JS
+// assets of the former SSG builds
+// We have to clean up the node require cache manually to avoid leaking memory!
+// See also https://x.com/sebastienlorber/status/1848399310116831702
+export function createSSGRequire(serverBundlePath: string): SSGNodeRequire {
+  const realRequire = createRequire(serverBundlePath);
+
+  const allRequiredIds: string[] = [];
+
+  const ssgRequireFunction: NodeJS.Require = (id) => {
+    const module = realRequire(id);
+    allRequiredIds.push(id);
+    return module;
+  };
+
+  const cleanup = () => {
+    allRequiredIds.forEach((id) => {
+      delete realRequire.cache[realRequire.resolve(id)];
+    });
+  };
+
+  ssgRequireFunction.resolve = realRequire.resolve;
+  ssgRequireFunction.cache = realRequire.cache;
+  ssgRequireFunction.extensions = realRequire.extensions;
+  ssgRequireFunction.main = realRequire.main;
+
+  return {require: ssgRequireFunction, cleanup};
+}