Cleaning up with Some Shell Commands

Today, I encountered an interesting problem when syncing my music library between two different computers. Whenever I buy a CD (and yes, I still buy CDs on occasion — albeit ones that are used or heavily discounted), I rip it into FLAC. But sometimes, I'll have ripped an MP3 version of the same album on a different computer. Merging the differing libraries fills my main music repository with FLAC and MP3 versions of the same songs, which is not exactly what I want. I'd much rather throw away the lower bitrate files when a lossless one exits.

My solution is for the general form of "given a directory full of files matching pattern X, delete files matching pattern Y", although it will require parameter tweaking to be applied outside of the context of my music library.

find . -name '*.flac' | cut -d"/" -f 1-3 | uniq | while read line; do
  echo rm -f "$line/"*.mp3
done

I'll dissect how it works:

• find . -name '*.flac': Gets the path for all flac files under this root.
• cut -d"/" -f 1-3: Removes the actual filename. My paths are of the form "./Artist/Album/Track.extension", so I cut out everything after the second "/".
• uniq: Ensure that each album directory is only listed once.
• while read line; do [...] done: Grabs each line of input and makes it available inside the loop as the $line variable.

The inside of the loop performs a simple rm – actually, the posted line only echoes the command, which is much better for testing. To adapt this for other purposes, tweak the file matching lines and the cut line. Test first with ls or echo. Don't accidentally delete your data.