tar-stream

tar-stream is a streaming tar parser and generator and nothing else. It operates purely using streams which means you can easily extract/parse tarballs without ever hitting the file system.

Note that you still need to gunzip your data if you have a .tar.gz. We recommend using gunzip-maybe in conjunction with this.

npm install tar-stream

Usage

tar-stream exposes two streams, pack which creates tarballs and extract which extracts tarballs. To modify an existing tarball use both.

It implementes USTAR with additional support for pax extended headers. It should be compatible with all popular tar distributions out there (gnutar, bsdtar etc)

Related

If you want to pack/unpack directories on the file system check out tar-fs which provides file system bindings to this module.

Packing

To create a pack stream use tar.pack() and call pack.entry(header, [callback]) to add tar entries.

1const tar = require('tar-stream')
2const pack = tar.pack() // pack is a stream
3
4// add a file called my-test.txt with the content "Hello World!"
5pack.entry({ name: 'my-test.txt' }, 'Hello World!')
6
7// add a file called my-stream-test.txt from a stream
8const entry = pack.entry({ name: 'my-stream-test.txt', size: 11 }, function(err) {
9  // the stream was added
10  // no more entries
11  pack.finalize()
12})
13
14entry.write('hello')
15entry.write(' ')
16entry.write('world')
17entry.end()
18
19// pipe the pack stream somewhere
20pack.pipe(process.stdout)

Extracting

To extract a stream use tar.extract() and listen for extract.on('entry', (header, stream, next) )

1const extract = tar.extract()
2
3extract.on('entry', function (header, stream, next) {
4  // header is the tar header
5  // stream is the content body (might be an empty stream)
6  // call next when you are done with this entry
7
8  stream.on('end', function () {
9    next() // ready for next entry
10  })
11
12  stream.resume() // just auto drain the stream
13})
14
15extract.on('finish', function () {
16  // all entries read
17})
18
19pack.pipe(extract)

The tar archive is streamed sequentially, meaning you must drain each entry's stream as you get them or else the main extract stream will receive backpressure and stop reading.

Extracting as an async iterator

The extraction stream in addition to being a writable stream is also an async iterator

1const extract = tar.extract()
2
3someStream.pipe(extract)
4
5for await (const entry of extract) {
6  entry.header // the tar header
7  entry.resume() // the entry is the stream also
8}

Headers

The header object using in entry should contain the following properties. Most of these values can be found by stat'ing a file.

1{
2  name: 'path/to/this/entry.txt',
3  size: 1314,        // entry size. defaults to 0
4  mode: 0o644,       // entry mode. defaults to to 0o755 for dirs and 0o644 otherwise
5  mtime: new Date(), // last modified date for entry. defaults to now.
6  type: 'file',      // type of entry. defaults to file. can be:
7                     // file | link | symlink | directory | block-device
8                     // character-device | fifo | contiguous-file
9  linkname: 'path',  // linked file name
10  uid: 0,            // uid of entry owner. defaults to 0
11  gid: 0,            // gid of entry owner. defaults to 0
12  uname: 'maf',      // uname of entry owner. defaults to null
13  gname: 'staff',    // gname of entry owner. defaults to null
14  devmajor: 0,       // device major version. defaults to 0
15  devminor: 0        // device minor version. defaults to 0
16}

Modifying existing tarballs

Using tar-stream it is easy to rewrite paths / change modes etc in an existing tarball.

1const extract = tar.extract()
2const pack = tar.pack()
3const path = require('path')
4
5extract.on('entry', function (header, stream, callback) {
6  // let's prefix all names with 'tmp'
7  header.name = path.join('tmp', header.name)
8  // write the new entry to the pack stream
9  stream.pipe(pack.entry(header, callback))
10})
11
12extract.on('finish', function () {
13  // all entries done - lets finalize it
14  pack.finalize()
15})
16
17// pipe the old tarball to the extractor
18oldTarballStream.pipe(extract)
19
20// pipe the new tarball the another stream
21pack.pipe(newTarballStream)

Saving tarball to fs

1const fs = require('fs')
2const tar = require('tar-stream')
3
4const pack = tar.pack() // pack is a stream
5const path = 'YourTarBall.tar'
6const yourTarball = fs.createWriteStream(path)
7
8// add a file called YourFile.txt with the content "Hello World!"
9pack.entry({ name: 'YourFile.txt' }, 'Hello World!', function (err) {
10  if (err) throw err
11  pack.finalize()
12})
13
14// pipe the pack stream to your file
15pack.pipe(yourTarball)
16
17yourTarball.on('close', function () {
18  console.log(path + ' has been written')
19  fs.stat(path, function(err, stats) {
20    if (err) throw err
21    console.log(stats)
22    console.log('Got file info successfully!')
23  })
24})

Performance

See tar-fs for a performance comparison with node-tar

License

MIT