24.12 Compressing a Directory Tree: Fine-Tuning

Here's a quick little command that will compress (24.7) files in the current directory and below. It uses find (17.2) to find the files, recursively, and pick the files it should compress:

-size xargs	% find . ! -perm -0100 -size +1 -type f -print | xargs gzip -v

This command finds all files that:

• Are not executable (! -perm -0100), so we don't compress shell scripts and other program files.

• Are bigger than one block, since it won't save any disk space to compress a file that takes one disk block or less. But, depending on your filesystem, the -size +1 may not really match files that are one block long. You may need to use -size +2, -size +1024c, or something else.

• Are regular files (-type f) and not directories, named pipes, etc.

The -v switch to gzip tells you the names of the files and how much they're being compressed. If your system doesn't have xargs, use:

% find . ! -perm -0100 -size +1 -type f -exec gzip -v {} \;

Tune the find expressions to do what you want. Here are some ideas - for more, read your system's find manual page:

! -name \*.gz

Skip any file that's already gzipped (filename ends with .gz ).

-links 1

Only compress files that have no other (hard) links.

-user yourname

Only compress files that belong to you.

-atime +60

Only compress files that haven't been accessed (read, edited, etc.) for more than 60 days.

You might want to put this in a job that's run every month or so by at (40.3) or cron (40.12).