Shell script to delete all files of certain type in different folders, except 10 latest

Question

First of all, please excuse my poor english.

I am trying to write a shell script to clean up some files of certain type in my backup folders.

I need to remove files of certain type, except 10 newest in each folder.

Folder structure looks like this:

Root folder
│
├── folder_1
│   ├── file_1.txt
│   └── file_2.txt
│   └── file_n.txt
│
├── folder_2
│   ├── file_1.txt
│   └── file_2.txt
│   └── file_n.txt
│
├── folder_n
│   ├── file_1.txt
│   └── file_2.txt
│   └── file_n.txt
│

I've used script from David Foerster as a basis, but can't figure out on how to make it work in different folders separately without manually writing each folder name in script.

Currently, script looks like this:

find /volume1/rootfolder/ -type f -name *.txt -printf '%T@ %p\0' |
sort --zero-terminated --reverse --numeric-sort --field-separator=' ' --key 1,1 |
gawk -F ' ' -v RS='\0' -v ORS='\0' -v retain_count=10 \
  'BEGIN{ maxage = systime() - retain_younger_days * 24 * 3600; }
  (NR > retain_count) && (int($1) < maxage) { print(substr($0, length($1) + 2)); }' |
xargs -r0 -- rm --

But the issue is that it removes all files in all folders, except 10 latest. So, overall i have just 10 files in all folders, not 10 in each folder.

Please, help me figure out, how to make script to process each folder separately.

Answer 1

You divide it into two parts. First, you build a directory structure, and in this case, we only need directories containing files. In the second part, we sort the newest file first and skip those with tail.

#!/bin/bash
while IFS= read -rd ''; do
    find "$REPLY" 

    -maxdepth 1 -type f -name '*.txt' -printf '%T@\t%p\0' | sort -zk1rn | cut -zf2- | tail -zn +$((10+1)) | xargs -r -0 echo rm
done < <( 

    find $PWD/data 

    -mindepth 2 -type f -printf %h\0 | sort -zu 

)

There is also the possibility to use globstar with a for-loop:

#!/bin/bash
shopt -s globstar nullglob
for a in data///; do
    find "$a" 

    -maxdepth 1 -type f -name '.txt' -printf '%T@\t%p\0' | sort -zk1rn | cut -zf2- | tail -zn +$((10+1)) | xargs -r -0 echo rm
done

Lastly, we make use of an array of counters to keep track of the file count:

#!/bin/bash
find $PWD/data -mindepth 1 -type f -name '*.txt' -printf '%T@\t%p\0' 

    | sort -zk1rn | cut -zf2- 

    | awk -F / -vRS='\0' -vORS='\0' '{a=$0; NF=NF-1} b[$0]++ >= 10 {print a}' | xargs -r -0 echo rm