How to write bash script that deletes the oldest file in a directory if there are greater than N files in the directory?

Question

Wondering how would one go about writing a script that:

Counts the number of files in a directory.

If that number is greater than a specified number N, delete the oldest file in the directory.

For bonus feelings of good will: Only included .zip files in the count/delete.

This feels a little like we may be doing school or real work for @Ryan. Perhaps you can demonstrate what you have come up with so far. — George M, May 14 '12 at 20:53
@uther Not for work, for a world backup on a personal Minecraft server. If you don't want to help, that's OK. So far I just have a script that copies files to a location, but it will fill up the drive unless I know how to do the script requested above. — MetaGuru, May 14 '12 at 20:54
It's not that I don't want to help, otherwise I wouldn't spend time here. What you are asking for is fairly trivial. It would go a long way to see you make an effort to write a script first and ask for critique/improvements instead of asking someone to write the script for you from scratch. I want you to learn something. — George M, May 14 '12 at 20:58
Why all the down-votes? BTW, possibly [relevant reading](http://mywiki.wooledge.org/BashFAQ/004). — jw013, May 14 '12 at 21:28
When you mouse over the downvote button it says: "This question does not show any research effort". If that doesn't apply here, when does it apply? — Mikel, May 14 '12 at 22:52
Oh, sorry, I voted to close for the wrong duplicate. I should have picked [Shell script for moving oldest files?](http://unix.stackexchange.com/q/22674) or [Bash, remove oldest files](http://unix.stackexchange.com/q/35792) — Gilles 'SO- stop being evil', May 14 '12 at 23:12

score 1 · Answer 1 · edited Feb 02 '21 at 14:23

1

N=50
[[ $( ls | wc -l ) -gt $N ]] && ls -tr | tail -n1 | tr \\n \\0 | xargs -0 echo rm

Barely tested, but I think this comes close. If you are happy with the text output, you can remove the echo to set it live.

ls list files
wc count them
[[ ... -gt ... ]] if
&& then
ls -tr list files in reverse age
tail -n1 show only last line (replace 1 as needed)
tr \\n \\0 make \0 the delimiter between filenames (only one here)
xargs -0 echo rm -- append every argument (delimited by \0s) found to echo rm --
echo rm -- give you the opportunity to check if the result is really what you want
rm -- remove file (put -- before filenames to handle filenames starting with -)

edited Feb 02 '21 at 14:23

Ned64

8,486
9
48
86

answered May 14 '12 at 21:03

jippie

13,756
10
44
64

1

This will break on filenames containing newlines. – Chris Down May 14 '12 at 21:46
1

This will break on filenames containing any kind of whitespace or `\'"`. – Gilles 'SO- stop being evil' May 14 '12 at 23:11
Schoolbook assignments don't have newlines and quotes in filenames. ;o) – jippie May 15 '12 at 06:28
@jippie this is for my minecraft server backups :) – MetaGuru May 15 '12 at 15:44
Same for minecraft server backups ;o) (just guessing) – jippie May 15 '12 at 19:07

score 1 · Answer 2 · answered May 14 '12 at 21:51

For example, to only perform the actions if there are more than 50 files in the folder:

shopt -s dotglob nullglob
for file in *; do
    [[ -f $file ]] && files+=( "$file" )
done
if (( ${#files[@]} > 50 )); then
    IFS= read -r -d $'\0' line < <(find . -maxdepth 1 -type f -printf '%T@ %p\0' 2>/dev/null | sort -z -n)
    rm "${line#* }"
fi

How to write bash script that deletes the oldest file in a directory if there are greater than N files in the directory?

2 Answers2