find files and delete by filename parameter - bash

Question

I have a folder with lots of images. In this folder are subfolders containing high resolution images. Images can be .png, .jpg or .gif.
Some images are duplicates called a.jpg and a.hi.jpg or a.b.c.gif and a.b.c.hi.gif. File names are always different, the will be never a.gif, a.jpg or a.png. I guess i have not to take care of extension.
These are the same images with different resolution.
Now i want to write a script to delete all lower resolution images. But there are files that do not have high resolution like b.png. So i want to delete only if there is a high resolution image too.
I guess i have to do something like this, but can't figure out how exactly.
find . -type f -name "*" if {FILENAME%hi*} =2 --delete smallest else keep file
Could anyone help? Thanks

Answer 1

Something like the following could do the job:
#!/bin/bash
while IFS= read -r -d '' hi
do
d=$(dirname "$hi")
b=$(basename "$hi")
low="${b//.hi./}"
[[ -e "$d/$low" ]] && echo rm -- "$d/$low" #dry run - if satisfied, remove the echo
done < <(find /some/path -type f -name \*.hi.\* -print0)
how it works:
finds all files with .hi. in their names. (not only images, you can extend the find be more restrictive
for all found images
get the directory, where is he
and get the name of the file (without directory)
in the name, remove all occurences of the string .hi. (aka make the "lowres" name
check the existence of the lowres image
delete if exists.

Answer 2

You can use bash extended glob features for this, which you can enable first by
shopt -s extglob
and using the pattern
!(pattern-list)
Matches anything except one of the given patterns.
Now to store the files not containing the string hi
shopt -s extglob
fileList=()
fileList+=( !(*hi*).jpg )
fileList+=( !(*hi*).gif )
fileList+=( !(*hi*).png )
You can print once the array to see if it lists all the files you need as
printf "%s\n" "${fileList[#]}"
and to delete those files do
for eachfile in "${fileList[#]}"; do
rm -v -- "$eachfile"
done
(or) as Benjamin.W suggested in comments below, do
rm -v -- "#{fileList[#]}"

Answer 3

Now i want to write a script to delete all lower resolution images
This script could be used for that:
find /path/to/dir -type f -iname '*.hi.png' -or -iname '*.hi.gif' -or -iname '*.hi.jpg' | while read F; do LOWRES="$(echo "$F" | rev | cut -c7- | rev)$(echo "$F" | rev | cut -c 1-3 | rev)"; if [ -f "$LOWRES" ]; then echo rm -fv -- "$LOWRES"; fi; done
You can run it to see what files will be removed first. If you're ok with results then remove echo before rm command.
Here is the non-one line version, but a script:
#!/bin/sh
find /path/to/dir -type f -iname '*.hi.png' -or -iname '*.hi.gif' -or -iname '*.hi.jpg' |
while read F; do
NAME="$(echo "$F" | rev | cut -c7- | rev)"
EXTENSION="$(echo "$F" | rev | cut -c 1-3 | rev)"
LOWRES="$NAME$EXTENSION"
if [ -f "$LOWRES" ]; then
echo rm -fv -- "$LOWRES"
fi
done