Webufs, a Linux-style Web Filesystem Framework.

If you want to use this library directly in your browser, please see the document of the package "@webufs/webufs-browser"

I just started to work on this library. The current versions are all unstable. I will also improve its performance in the future. Please wait and see.

If you have any issues or suggestions, feel free to discuss them on GitHub. Thanks!

Usage

The createDefaultContext method will mount the default in-memory filesystem at / (data will be lost upon closing).

To mount other filesystems, see section Mounting Other FS below.

Basic Directory Operation:

import { createDefaultContext } from '@webufs/webufs'

// Yes, it's async
const ctx = await createDefaultContext()

await ctx.mkdir('a')
await ctx.chdir('/a')
console.log(ctx.getcwd())       /* '/a' */
await ctx.chdir('..')
await ctx.rmdir('./a/')
// ...

// scan directory
fd = await ctx.open('/', { directory: true })
let fileInfos = await fd.getdents()
for (let dirEntry of fileInfos) {
    console.log(`name: ${dirEntry.name}, type: ${dirEntry.type} `)
}

// check if file exists
const status = await ctx.access('abc.png')
if (status) {
    console.error('File abc.png doesn\'t exist. ')
}
// ...

Basic File Operation:

const fd = await ctx.open('a.txt', { create: true })

const src = new Uint8Array([0x30, 0x31, 0x32, 0x33])
await fd.write(src.buffer)
await fd.write(src.buffer)
await fd.write(src.buffer)

const dst = new ArrayBuffer(8)
const dstView = new DataView(dst)
await fd.seek(2, SeekType.SET)          // offset +2 relative to file start 
// You can seek relative to SET, CUR or END.
await fd.read(dst)
// read data from file into dstView...
// NOTE: read(size) will return the number of bytes that are actually read, ranging in the interval [0, size], and
// when less than size, it means that we are near the end of the file.

//  don't forget to close the file (this will also flush pending writing and release reference)
await fd.close()

// delete a file
await ctx.unlink('a.txt')

Read file till the end:

fd = await ctx.open('a.txt', { readonly: true })
let size = 0
const buffer = new ArrayBuffer(0x100)
while (true) {
    const cnt = await fd.read(bufer)
    size += cnt
    if (cnt < 0x100) break
}
await fd.close()

console.log(`File Size = ${size}`)

Discard original content (with size set back to zero) and re-write:

fd = await ctx.open('a.txt', { trunc: true })
const src = new Uint8Array([0x30, 0x31, 0x32, 0x33])
await fd.write(src.buffer)
await fd.close()

Mounting Other FS

We take @webufs/webufs-idb, a filesystem implementation of IndexedDB backend, for example:

import { createDefaultContext } from '@webufs/webufs'
import { IDBFS } from '@webufs/webufs-idb'

const ctx = await createDefaultContext()
ctx.getVFS().registerFSType(IDBFS)
await ctx.mkdir('idb')
await ctx.mount('idbfs', '/idb')
await ctx.chdir('idb')

// See the documentation of '@webufs/webufs-idb' for further detail.

FS umounting not implemented yet.

About Versioning

Change of major version number will introduce incompatibility to both users and FS backend and plugin developers.

Change of minor version number keeps compatibility to users, but may cause API incompatibility to the backend and plugin developers.

Change of patch version number keeps compatibility to both users and developers.

Hope (not implemented yet)

We might also add shell-like support:

await sh.cd('/home/user/').run()
await sh.echo('Hello!').pipe(base64()).to('a.txt').run()
await sh.cmd(find('./src', '-name', '*.ts'))
    .pipe(xargs(cat()))
    .pipe(wc('-l'))
    .run()
/**
 * base64, find, cat are util functions using our VFS API.
 */

That would be cool.

Some Ideas

I found that we can generate a URL from Blob and use it to spawn Web Workers. That might give some ideas...