ZIP-go

zip-go was designed with the goal of saving multiple large files generated by the browsers without holding any of the data in memory by streaming/piping data to a destination. While the zip format isn't technically built for streaming and each file entry needs some pre-header information that tells how large a file is and what the crc checksum is... all of this can be stored in the central directory header also at the end of a zip file...

Install

zip-go is an ESM-only module - you are not able to import it with require. If you are unable to use ESM in your project you can use the async import('zip-go') from CommonJS to load zip-go asynchronously.
npm install zip-go

Requirements

• BigInt support
• ReadableStream
• WritableStream
• Reading compressed file entries requires DecompressionStream
• Reading a zip file entries is done with Blob-like objects. NodeJS users can use fs.openAsBlob or fetch-blob

It can't as of yet write zip files larger than 4 GiB as it has no zip64 support but it can read those.

Creating a ZIP

import Writer from 'zip-go/lib/write.js'

const s3 = 'https://s3-us-west-2.amazonaws.com/bencmbrook/'
const files = ['NYT.txt', 'water.png', 'Earth.jpg'].values()

// Creates a regular ReadableStream that will pull file like objects
const myReadable = new ReadableStream({
  async pull(controller) {
    const { done, value } = files.next()
    if (done) return controller.close()
    const { body } = await fetch(s3 + value)

    return controller.enqueue({
      name: `/${value}`,
      stream: () => body,
    })
  }
})

const stream = myReadable
  .pipeThrough(new ZipWriter())

const blob = await new Response(stream).blob()

if you would like to work it more manually you can do that as well.

import StreamSaver from 'streamsaver';
import Writer from 'zip-go/lib/write.js';

const { readable, writable } = new Writer();
const writer = writable.getWriter();

// Set up StreamSaver
const fileStream = streamSaver.createWriteStream('archive.zip');

// Add a WebIDL File like object that at least have name and a stream method
// that returns a whatwg ReadableStream
writer.write({
  name: '/cat.txt',
  lastModified: +new Date(123),
  stream: () => new Response('mjau').body
})

writer.write(
  new File(['woof'], 'dog.txt', { lastModified: +new Date(123)})
)

readable.pipeTo(destination)

writer.close()

Reading a zip

This read method only read the central directory (end of the file) to figure out all about each entry. Each Entry returns a WebIDL File like object with added properties that are more zip specific

import read from 'zip-go/lib/reader.js'

for await (const entry of read(blob)) {
  console.log(entry)
  console.log(entry.name)
  console.log(entry.size)
  console.log(entry.type)
  console.log(entry.directory)

  const ab = await entry.arrayBuffer()
  const text = await entry.text()
  const readableStream = entry.stream()

  // returns a real web File Object, if the entry is uncompressed
  // it will just slice the zip with it's start/end position
  const file = await entry.file()
}