Add sparse file support #35
Conversation
* add test cases for multiple extended headers * fix bug where 80+ extended headers would crash tar
There was a problem hiding this comment.
Thanks for the PR, @zwhitchcox! I'd like to see a few small changes before merging.
| }); | ||
| ``` | ||
|
|
||
| If you prefer the raw data chunks as they appear in the archive (without reconstructing zeros), you can call `entry.bytes({ raw: true })`: |
There was a problem hiding this comment.
When you say "without reconstructing zeros" do you mean that the resulting byte array will have zeroes in it? That's just spacer data, right? Forgive my ignorance, but when would that ever be useful?
There was a problem hiding this comment.
This is the default behavior of tar.
Basically, if someone sends you a sparse tarball, you don't have to do anything differently if you just pipe to a write stream for the file. So, for someone with no knowledge of sparse files, it "just work".
However, if you're a more advanced user, you can get the sparse offset's and lengths, and use fs.write(fd, buffer, offset, length), which will be more efficient, because you're not writing the extra data.
Add support for sparse files using old gnu format
This is the current default for gnu tar generated tar files, including extended headers.
Also supports raw mode, which will read the sparse file map and only emit chunks actually containing data corresponding to the sparse map. This is useful to save memory while streaming a sparse file.