Global namespace pollution when using rollupOptions.output.format umd

[TomPradat]

Describe the bug

I'm building a bundle with umd format as a target. When doing so I expect the result bundle to be wrapped in an iife function. The configuration looks like this :

  build: {
    rollupOptions: {
      output: {
        format: "umd",
        entryFileNames: "bundle.js",
      },
    },
  },

Instead, I have variables declared outside the iife :

This is a problem because the variables' declaration might override previous variables with the same name. This is how I encountered the issue.

Workaround :

So far, I have used a homemade plugin to wrap my final code in a iife :

import { defineConfig } from "vite";

// https://vitejs.dev/config/
export default defineConfig({
  build: {
    rollupOptions: {
      output: {
        format: "umd",
        entryFileNames: "bundle.js",
      },
    },
  },
  plugins: [viteWrapCodeInIIFE({ files: ["bundle.js"] })],
});

function viteWrapCodeInIIFE(options = {}) {
  return {
    name: "vite-wrap-code-in-iife",
    apply: "build",
    enforce: "post",
    generateBundle(outputOptions, bundle) {
      for (const [fileName, chunkOrAsset] of Object.entries(bundle)) {
        if (
          chunkOrAsset.type === "chunk" &&
          options.files &&
          options.files.includes(fileName)
        ) {
          chunkOrAsset.code = `(function () {${chunkOrAsset.code}})();`;
        }
      }
    },
  };
}

I'm willing to work on a PR. I believe the fix would be an extension of the #7948 so that is also applies when rollupOptions.output.format is umd or iife.

Reproduction

See steps

Steps to reproduce

1. Bootstrap a vite project

npm init vite@latest test -- --template vanilla && cd test && npm install

2. Install the "react-query" package that causes the build to have variables defined outside the iife

npm install @tanstack/react-query

3. Create a vite.config.js with the umd format as a target

cat <<EOF > vite.config.js
import { defineConfig } from "vite";

// https://vitejs.dev/config/
export default defineConfig({
  build: {
    rollupOptions: {
      output: {
        format: "umd",
        entryFileNames: "bundle.js",
      },
    },
  },
});
EOF

4. Change the main.js

cat <<EOF > main.js
import { QueryClient } from "@tanstack/react-query";

const queryClient = new QueryClient();
EOF

5. Build the bundle

npm run build

6. See the result in dist/bundle.js

System Info

System:
    OS: macOS 13.4.1
    CPU: (8) arm64 Apple M1 Pro
    Memory: 330.16 MB / 16.00 GB
    Shell: 5.9 - /bin/zsh
  Binaries:
    Node: 18.16.0 - ~/.nvm/versions/node/v18.16.0/bin/node
    Yarn: 3.5.1 - ~/.nvm/versions/node/v18.16.0/bin/yarn
    npm: 9.5.1 - ~/.nvm/versions/node/v18.16.0/bin/npm
    pnpm: 8.15.4 - ~/Library/pnpm/pnpm
    bun: 1.0.14 - ~/.bun/bin/bun
  Browsers:
    Brave Browser: 124.1.65.132
    Chrome: 126.0.6478.127
    Firefox Nightly: 121.0a1
    Safari: 16.5.2
  npmPackages:
    vite: ^5.3.1 => 5.3.3

Used Package Manager

npm

Logs

No response

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines.
• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Make sure this is a Vite issue and not a framework-specific issue. For example, if it's a Vue SFC related bug, it should likely be reported to vuejs/core instead.
• Check that this is a concrete bug. For Q&A open a GitHub Discussion or join our Discord Chat Server.
• The provided reproduction is a minimal reproducible example of the bug.

[sapphi-red]

I haven't checked deeply but I guess we should use opts.format === 'iife' || opts.format === 'umd' instead of config.build.lib for people using IIFE/UMD format without turning lib mode on.

vite/packages/vite/src/node/plugins/esbuild.ts

Line 320 in 22b2994

if (config.build.lib) {

[TomPradat]

I haven't checked deeply but I guess we should use opts.format === 'iife' || opts.format === 'umd' instead of config.build.lib for people using IIFE/UMD format without turning lib mode on.

vite/packages/vite/src/node/plugins/esbuild.ts

Line 320 in 22b2994

if (config.build.lib) {

This is what I was thinking too 👍 . I can work on a PR as soon as this is confirmed as a bug

[sapphi-red]

Yeah, I think it's a bug.

[syranide]

Even with format=IIFE (at least for non-libs) it seems that some level of global namespace pollution occurs:
#16443 (comment)

[eXaminator]

We also ran into this issue today because the _ variable was redefined and underscore didn't work anymore in our old'ish application. Thank you @TomPradat for the Workaround, it works like a charm! Would be great if this could be the default though.

[christopher1986]

Any updates on this bug? I also noticed that helpers are added to the global space instead of inside IIFE.

I believe this should also be corrected when not using build.lib as was fixed by this pull request.

[nealoke]

I'm running writing a small script that people will embed on their website and am I'm running into a this issue with Vite (version 7.1.3) where pollution of the global scope happens even when the output is iife.

Config

export default defineConfig(({ mode }) => {
	return {
		build: {
			emptyOutDir: false,
			minify: true,
			sourcemap: mode === 'production' ? false : true,
			rollupOptions: {
				input: {
					tracker: './src/tracker/main.ts',
				},
				output: {
					entryFileNames: 'pmt.js',
					format: 'iife',
				},
			},
			target: 'es2020',
			outDir: './src/backend/assets/dist',
		},
		plugins: [tsconfigPaths()],
	};
});

Expected output

(function() {
    "use strict";
    var k = Object.defineProperty;
    var C = (l, r, d) => r in l ? k(l, r, {
        enumerable: !0,
        configurable: !0,
        writable: !0,
        value: d
    }) : l[r] = d;
    var a = (l, r, d) => C(l, typeof r != "symbol" ? r + "" : r, d);
    ...

Actual output

var k = Object.defineProperty;
var C = (l, r, d) => r in l ? k(l, r, {
    enumerable: !0,
    configurable: !0,
    writable: !0,
    value: d
}) : l[r] = d;
var a = (l, r, d) => C(l, typeof r != "symbol" ? r + "" : r, d);
(function() {
    "use strict";
    ...

Let me know if this needs to be created as a separate issue, but to me seems very related?