@bybrave/jszip2

Maintained fork of JSZip — create, read and edit .zip files with JavaScript, in Node.js and the browser.

The original has ~187M downloads/month and no release since 2022 (400+ open issues and PRs). This fork keeps the exact same API and modernizes everything underneath it.

npm install @bybrave/jszip2

// CommonJS — drop-in
const JSZip = require("@bybrave/jszip2");

// ESM — finally (top ask of the original backlog)
import JSZip from "@bybrave/jszip2";

const zip = new JSZip();
zip.file("hello.txt", "Hello World\n");
zip.folder("images").file("smile.gif", imgData, { base64: true });
const blob = await zip.generateAsync({ type: "blob" });

const archive = await JSZip.loadAsync(data);
const text = await archive.file("hello.txt").async("string");

ZIP64 (new in 4.3)

Archives and files over 4 GB, and over 65 535 entries (#580, #739 upstream). No configuration needed — ZIP64 kicks in automatically when a size, an offset or the entries count overflows the classic format:

zip.file("huge.bin", hugeStream);
const out = zip.generateNodeStream(); // > 4 GB? ZIP64 structures appear by themselves

The single case needing a flag: streamFiles: true with files over 4 GB (sizes are unknown when each entry starts, so the format can't be decided on the fly) — pass zip64: true, a descriptive error will remind you otherwise. Reading ZIP64 always worked, but sizes beyond 4 GB used to be truncated by 32-bit arithmetic — now exact up to 2^53 − 1 bytes.

Web Streams (new in 4.2)

WHATWG ReadableStream in and out (#345, #830 upstream) — works in browsers, Web Workers and Node.js ≥ 18:

// in: zip a fetch response without buffering it first
const response = await fetch(url);
zip.file("data.json", response.body);

// in: load a zip straight from fetch
const archive = await JSZip.loadAsync((await fetch(zipUrl)).body);

// out: stream the generated zip — backpressure included
return new Response(zip.generateWebStream({ compression: "DEFLATE" }), {
  headers: { "Content-Type": "application/zip" }
});

// out: stream one file's content
await archive.file("video.mp4").webStream().pipeTo(destination);

generateWebStream / webStream mirror generateNodeStream / nodeStream; feature detection via JSZip.support.webstream.

Sync API (new in 4.1)

The most-requested feature of the original (#281, open since 2016): zip and unzip without promises — in CLI scripts, Web Workers, getters, anywhere async doesn't fit.

const zip = new JSZip();
zip.file("hello.txt", "Hello World\n");
const buffer = zip.generateSync({ type: "nodebuffer", compression: "DEFLATE" });

const archive = JSZip.loadSync(buffer);
const text = archive.file("hello.txt").sync("string");

Every async entry point has a sync mirror:

| Async | Sync | |---|---| | zip.generateAsync(options) | zip.generateSync(options) | | JSZip.loadAsync(data, options) / zip.loadAsync(...) | JSZip.loadSync(data, options) / zip.loadSync(...) | | file.async(type) | file.sync(type) |

Same options, same output types (including blob and base64), byte-identical results. The one rule: the data must actually be available synchronously. A file added as a Blob, a Promise or a Node.js stream can't be read without awaiting it, so the sync methods throw a descriptive error naming the file — everything else, including zips loaded with loadAsync, works. The async API is unchanged and still preferred for big archives on the main thread: sync calls block until the archive is done.

What's different from jszip 3.10.1

| Change | Original issue | |---|---| | ESM entry (import JSZip from "@bybrave/jszip2") + exports map with types | #867, #717 | | setimmediate polyfill removed (native setImmediate / MessageChannel) — fixes crashes in sandboxed environments, Tampermonkey, Edge | #909, #934, #596 | | pako upgraded 1.x → 2.x (faster, actively maintained compression) | #720 | | lie Promise polyfill removed — native Promises everywhere | — | | readable-stream dependency removed — Node's own stream in Node, clean stub in browser bundles (no more Cannot find module 'stream' with webpack 5 / modern bundlers) | #704 | | Blob/File input now works in Node.js (read via blob.arrayBuffer() instead of the browser-only FileReader) | — | | Bundlers consume the real source, not a pre-built UMD blob (the original silently substituted dist/jszip.min.js via the browser field) | — | | Build: esbuild instead of grunt + browserify; CI on Node 18/20/22 | — | | 4.1: synchronous API — generateSync, loadSync, file.sync(type) | #281 | | 4.2: Web Streams — ReadableStream in (file, loadAsync) and out (generateWebStream, file.webStream) | #345, #830 | | 4.3: ZIP64 — writing archives/files over 4 GB and over 65 535 entries (auto), exact 64-bit reads | #580, #739 |

Dependency count: 4 → 1 (pako).

Compatibility

• API is unchanged — this is a drop-in replacement for jszip. The full API documentation of the original applies: stuk.github.io/jszip (a copy lives in documentation/).
• Node.js ≥ 18; modern browsers (no IE — it needed the removed polyfills).
• JSZip.external.Promise is still overridable if you need a custom Promise implementation.
• Browser <script> builds: dist/jszip.js and dist/jszip.min.js (global JSZip), shipped in the npm package.

Support

If this package saves you time, you can support maintenance:

Bitcoin (BTC): bc1q37557q5jpeaxqydzwvf3jgj7zhnfpn2td3q40q

Credits & license

Dual-licensed MIT or GPLv3, same as the original — see LICENSE.markdown. Based on JSZip by Stuart Knightley, David Duponchel, Franz Buchinger and António Afonso.