Extracting tarball without specifying compression format.

[RobertF]

I am writing a script that installs updates to an embedded system.

The script is required to extract a update tarball compressed with either .gz .bz2 or .xz.

I have noticed on my gentoo system that I am able to decompress .tar.gz, .tar.bz2 and tar.xz compressed tarballs with only the -xvf option (ie without specifying the the compression filter -z -j or J options)

My question is: Can this behavior be expected for all versions of tar and will it be safe to assume this behavior if using the format "tar -xvf {*.tar.gz,*.tar.bz2,*.tar.xz}" in my script.

[RobertF]

UPDATE:
.tar.gz and .tar.bz2 can be extracted using only the -xvf tar options. However tar.xz requires the -J option included otherwise tarball will not be extracted. Thus I will need to check the extension and use the compression filter option in my script.

[Irithori]

May I suggest to rely on file, rather than a filename suffix?

e.g.:

file -i *.tar.*
filename1.tar.gz:    application/x-gzip; charset=binary
filename2.tar.bz2:  application/x-bzip2; charset=binary
filename3.tar.xz:    application/x-xz; charset=binary

Should be more accurate. Suffix checking can be an additional check, though.

[RobertF]

Good plan. Thanks

Just a note on file though: file will determine that the file is either a gzip, bzip2 or a xz type file but won't tell you if it is a tar file unless you have already extracted the tar from the gzip, bzip2 or a xz type file.

In my script I was hoping to extract the whole tarball with only the tar command. If I could extract each type of compressed tarball with the simple command tar -xvf then I could cut down on a bit of logic. However tar.xz type tarballs require the compression filter option. No problem though. Will just add some checking logic to my script.

However, I strongly agree with you in rather using file types as outputted by the file command rather than relying on the file name wherever possible.