Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

How can I recursively bzip2 all files that do not end with the .bz2 extension in Linux?

share|improve this question

4 Answers 4

up vote 6 down vote accepted

find is your friend. I reckon the following ought to do it:

find <target_dir> -not -name \*.bz2 -exec bzip2 \{\} \;

i.e. if the dir where the files you want to bzip are is /var/log/blah it would be:

find /var/log/blah -not -name \*.bz2 -exec bzip2 \{\} \;
share|improve this answer
    
It may be faster to use + or xargs instead of \;. It's not necessary to escape the curly braces. POSIX supports ! for "not". –  Dennis Williamson Aug 14 '10 at 13:15
    
Why vote down? Will it not work on the platform specified? –  Jason Tan Aug 14 '10 at 13:55
    
No clue why the downvote, this is the best (and correct) answer, I use it all the time. I also use xargs though, can't stand the -exec to find. Also consider using -print0 on the find and -0 to xargs to use NULL separation instead of spaces in case the filenames have spaces. –  troyengel Aug 14 '10 at 14:18
1  
Unless they're all very small files, xargs is probably premature optimization anyways (the actual compression will take most of the time, not the fork and exec of starting up bzip2 repeatedly)... But: find /var/log/blah -not -name \*.bz2 -print0 | xargs -0 bzip2 –  freiheit Aug 14 '10 at 16:56

Here is a one liner. If necessary, install ifne first. On Ubuntu:

sudo apt-get install moreutils

And then:

find YOUR_DIR -type f -not -name \*.bz2 -print0 | ifne xargs -0 bzip2

This does the same as other answers, but you won't get warning messages about bzip2ing directories, and if there is nothing to bzip2 anymore, you won't get warning messages about bzip2ing compressed data to a terminal. It also uses xargs etc.

YOUR_DIR can have wild cards or multiple dirs, e.g. YOUR*DIR or the like.

share|improve this answer

With zsh (setopt extended_glob must be on):

bzip2 **/^*.bz2(.)

** recurses in subdirectories; ^*.bz2 matches everything except *.bz2; (.) restricts to regular files.

With bash 4, if you're ok with ignoring bzip2's complains about being invoked on directories:

bzip2 **/!(*.bz2)
share|improve this answer

Off the top of my head (sorry, don't have a shell handy to test quoting etc.):

for _t in `find . -print |grep -v -E "\.bz$"`; do bzip2 -9 $_t && echo OK $_t || echo FAIL $_t; done

This uses find to find all files, grep to strip out the ones with a .bz2 extension and then feed them one at a time into bzip2. I expect some of the quoting is wrong, though - I'd test the bit in the backquotes separately first.

Good luck! You might want to use xz instead, though - it usually compresses better - or even tar everything up and bzip2 or xz instead.

share|improve this answer
    
... or apparently you can do it all in find. I didn't know it could do '-not' hence the grep etc. here. –  Rup Aug 14 '10 at 12:52
    
I think the -not in find is only a GNU find option - so I don't think you could use it with most proprietary find commands - but since the question specifies Linux, GNU find is probably OK - your approach is probably more portable. –  Jason Tan Aug 14 '10 at 12:56
    
POSIX supports ! for "not". Using a for loop will fail for files that have spaces in their names. You should use find | while read. –  Dennis Williamson Aug 14 '10 at 13:15

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.