Delete all files except those that contain terms listed in a file

Question

I need to remove all files from a directory that are not partially matched to a list in a text file.

I can't figure out the matching. The underscore _ can be used to delimite the match. 1 would match with 1_ , but not 11 .

$ cat /files/keep_list.txt  
1
22
333
4444
$ ls /files/    
1_a.gif         # don't delete
1_b.gif         # don't delete
1_.jpeg         # don't delete
2_c.gif         # delete
2_d.gif         # delete
2_.jpeg         # delete
22_.jpeg        # don't delete
23_.jpeg        # delete
333_abc.gif     # don't delete
4445_123.jpeg   # delete

Answer 1

I'd fill GLOBIGNORE with globs that would match files to be kept and run rm *_* to get rid of everything else that has an underscore in its name.

#!/bin/bash -
mapfile -t parts <keep_list.txt

for part in "${parts[@]}"; do
  printf -v GLOBIGNORE '%s:%q_*:*_%q[_.]*' \
    "$GLOBIGNORE" "$part"{,}
done

echo rm *_*

Drop echo if the output looks good.

Answer 2

Not a bash solution, but in python (which should also be available on any linux system):

deleteButKeepSome.py:

import os

match = ['1', '22', '333', '4444'] # put the keep_list in here

for root, dirs, files in os.walk('/files'):
    for file in files:
        if file[:file.find('_')] not in match:
            os.remove(f'{root}/{file}')

To run:

$ python deleteButKeepSome.py

Be careful to execute it in the correct folder!