Tell Tar to Auto Compress Those Files!

Just discovered a handy shortcut when working with the GNU utility “tar”. Like many other Unix utilities, the switches that you can pass tar change its behavior. To create a “plain” tar file (compressing multiple files down to a single tape archive format – similar to zip) you can execute the following:

bsimpson@Saturn:/tmp$ echo "test" > test
bsimpson@Saturn:/tmp$ tar -cf test.tar test
bsimpson@Saturn:/tmp$ file test*
test:     ASCII text
test.tar: POSIX tar archive (GNU)

The “c” switch creates a new archive, and the “f” switch tells it that the name of the new archive follows. Similarly, we can untar this using the “x” switch, mutually exclusive with “c” for its create counterpart.

We can also apply compression to these new archives:

bsimpson@Saturn:/tmp$ echo "test" > test
bsimpson@Saturn:/tmp$ tar -czf test.tar.gz test
bsimpson@Saturn:/tmp$ file test*
test:        ASCII text
test.tar.gz: gzip compressed data, from Unix, last modified: Tue Jan 11 22:31:34 2011

Or alternately, we can provide the bzip2 compression:

bsimpson@Saturn:/tmp$ echo "test" > test
bsimpson@Saturn:/tmp$ tar -cjf test.tar.bz2 test
bsimpson@Saturn:/tmp$ file test*
test:         ASCII text
test.tar.bz2: bzip2 compressed data, block size = 900k

All well and good, however tar allows you to specify the compression (or lack thereof) based on the new filename. You call it what you want, and tar will figure out what compression to apply. Consider the following:

bsimpson@Saturn:/tmp$ echo "test" > test
bsimpson@Saturn:/tmp$ tar -caf test.tar test
bsimpson@Saturn:/tmp$ tar -caf test.tar.gz test
bsimpson@Saturn:/tmp$ tar -caf test.tar.bz2 test
bsimpson@Saturn:/tmp$ file test*
test:         ASCII text
test.tar:     POSIX tar archive (GNU)
test.tar.bz2: bzip2 compressed data, block size = 900k
test.tar.gz:  gzip compressed data, from Unix, last modified: Tue Jan 11 22:24:43 2011

As a side note, these do not stack. For example “test.tar.gz.bz2” just produces a bzip2 encoded ASCII test file. If you *really* wanted this, you could use the pipe command to chain your compressions.

Hope this saves some time!