Bash: Find folders with less than x files

Question

How would I go about finding all the folders in a directory than contain less than x number of .flac files?

Solution 1

• For every subdirectory, print the subdirectory name if there are at most 42 .flac files in the subdirectory. To execute a command on the directories, replace -print by -exec … \;. POSIX compliant.

  find . -type d -exec sh -c 'set -- "$0"/*.flac; [ $# -le 42 ]' {} \; -print
  

  Note that this command won't work to search for directories containing zero .flac files ("$0/*.flac" expands to at least one word). Instead, use

  find . -type d -exec sh -c 'set -- "$0"/*.flac; ! [ -e "$1" ]' {} \; -print
  

• Same algorithm in zsh. **/* expands to all the files in the current directory and its subdirectories recursively. **/*(/) restricts the expansion to directories. {.,**/*}(/) adds the current directory. Finally, (e:…:) restricts the expansion to the matches for which the shell code returns 0.

  echo {.,**/*}(/e:'set -- $REPLY/*.flac(N); ((# <= 42))':)
  

  This can be broken down in two steps for legibility.

  few_flacs () { set -- $REPLY/*.flac(N); ((# <= 42)); }
  echo {.,**/*}(/+few_flacs)
  

_{Changelog:
​• handle x=0 correctly.}

Solution 2

Replace $MAX with your own limit:

find -name '*.flac' -printf '%h\n' | sort | uniq -c | while read -r n d ; do [ $n -lt $MAX ] && printf '%s\n' "$d" ; done

Note: This will print all the subdirectories with a number of .flac files between 0 and $MAX (both excluded).

Comments

[cYrus]

Thanks for the -r, I really missed that! But I can't see any problem with echo, it correctly works with directories like: st r\a \`n-[g\"e

[cYrus]

The first command prints all the subdirectories. $# is always at least 1 hence [ $# -le 42 ] is true when there are no flac files in the subdirectory.

[Gilles 'SO- stop being evil']

@cYrus: If there are no flac files, there are fewer than 42 flac files. Ok, I should have mentioned that my solutions only work for 42 > 0, so they won't work to search for directories containing no flac files (you need a different, simpler command).

[cYrus]

I'm not talking about searching directories without flac files (x=0). I was saying that if a directory contains no flac files $# is 1 because $1 is literally path-to-dir/*.flac (as there's no expansion) the directory is printed anyway. It's just a different point of view, I'm assuming that a directory without flac files doesn't match the request.

[cYrus]

A directory called -e... evil!

[Cristian Ciupitu]

To avoid any issues with glob substitutions you should use -name '*.flac'.