tar-split

Extend the upstream golang stdlib archive/tar library, to expose the raw bytes of the TAR, rather than just the marshalled headers and file stream.

The goal being that by preserving the raw bytes of each header, padding bytes, and the raw file payload, one could reassemble the original archive.

Docs

• https://godoc.org/github.com/vbatts/tar-split/tar/asm
• https://godoc.org/github.com/vbatts/tar-split/tar/storage
• https://godoc.org/github.com/vbatts/tar-split/archive/tar

Caveat

Eventually this should detect TARs that this is not possible with.

For example stored sparse files that have "holes" in them, will be read as a contiguous file, though the archive contents may be recorded in sparse format. Therefore when adding the file payload to a reassembled tar, to achieve identical output, the file payload would need be precisely re-sparsified. This is not something I seek to fix imediately, but would rather have an alert that precise reassembly is not possible. (see more http://www.gnu.org/software/tar/manual/html_node/Sparse-Formats.html)

Other caveat, while tar archives support having multiple file entries for the same path, we will not support this feature. If there are more than one entries with the same path, expect an err (like ErrDuplicatePath) or a resulting tar stream that does not validate your original checksum/signature.

Contract

Do not break the API of stdlib archive/tar in our fork (ideally find an upstream mergeable solution)

Std Version

The version of golang stdlib archive/tar is from go1.4.1, and their master branch around a9dddb53f

Example

First we'll get an archive to work with. For repeatability, we'll make an archive from what you've just cloned:

git archive --format=tar -o tar-split.tar HEAD .

Then build the example main.go:

go build ./main.go

Now run the example over the archive:

$ ./main tar-split.tar
2015/02/20 15:00:58 writing "tar-split.tar" to "tar-split.tar.out"
pax_global_header pre: 512 read: 52
.travis.yml pre: 972 read: 374
DESIGN.md pre: 650 read: 1131
LICENSE pre: 917 read: 1075
README.md pre: 973 read: 4289
archive/ pre: 831 read: 0
archive/tar/ pre: 512 read: 0
archive/tar/common.go pre: 512 read: 7790
[...]
tar/storage/entry_test.go pre: 667 read: 1137
tar/storage/getter.go pre: 911 read: 2741
tar/storage/getter_test.go pre: 843 read: 1491
tar/storage/packer.go pre: 557 read: 3141
tar/storage/packer_test.go pre: 955 read: 3096
EOF padding: 1512
Remainder: 512
Size: 215040; Sum: 215040

What are we seeing here?

• pre is the header of a file entry, and potentially the padding from the end of the prior file's payload. Also with particular tar extensions and pax attributes, the header can exceed 512 bytes.
• read is the size of the file payload from the entry
• EOF padding is the expected 1024 null bytes on the end of a tar archive, plus potential padding from the end of the prior file entry's payload
• Remainder is the remaining bytes of an archive. This is typically deadspace as most tar implmentations will return after having reached the end of the 1024 null bytes. Though various implementations will include some amount of bytes here, which will affect the checksum of the resulting tar archive, therefore this must be accounted for as well.

Ideally the input tar and output *.out, will match:

$ sha1sum tar-split.tar*
ca9e19966b892d9ad5960414abac01ef585a1e22  tar-split.tar
ca9e19966b892d9ad5960414abac01ef585a1e22  tar-split.tar.out

Stored Metadata

Since the raw bytes of the headers and padding are stored, you may be wondering what the size implications are. The headers are at least 512 bytes per file (sometimes more), at least 1024 null bytes on the end, and then various padding. This makes for a constant linear growth in the stored metadata, with a naive storage implementation.

Reusing our prior example's tar-split.tar, let's build the checksize.go example:

go build ./checksize.go

$ ./checksize ./tar-split.tar
inspecting "tar-split.tar" (size 210k)
 -- number of files: 50
 -- size of metadata uncompressed: 53k
 -- size of gzip compressed metadata: 3k

So assuming you've managed the extraction of the archive yourself, for reuse of the file payloads from a relative path, then the only additional storage implications are as little as 3kb.

But let's look at a larger archive, with many files.

$ ls -sh ./d.tar
1.4G ./d.tar
$ ./checksize ~/d.tar 
inspecting "/home/vbatts/d.tar" (size 1420749k)
 -- number of files: 38718
 -- size of metadata uncompressed: 43261k
 -- size of gzip compressed metadata: 2251k

Here, an archive with 38,718 files has a compressed footprint of about 2mb.

Rolling the null bytes on the end of the archive, we will assume a bytes-per-file rate for the storage implications.

uncompressed	compressed
~ 1kb per/file	0.06kb per/file

What's Next?

• More implementations of storage Packer and Unpacker

• could be a redis or mongo backend

• More implementations of FileGetter and FilePutter

• could be a redis or mongo backend

• cli tooling to assemble/disassemble a provided tar archive
• would be interesting to have an assembler stream that implements io.Seeker

License

See LICENSE