Let's assume you backup a file every hour on AWS S3 and you want to clean up versions that are older than a week. However for the older versions, you want to leave one version per day just in case if you need them. Following script does that for you.

delete_all_aws_s3_object_versions.sh

#!/bin/bash

deleteBefore=`date --date="1 week ago" +%F`
bucket=$1
fileToDelete=$2
fileName='aws_delete.json'
rm $fileName
echo "Removing all versions of $fileToDelete from $bucket"
versionsToDelete=`aws s3api list-object-versions --bucket "$bucket" --prefix "$fileToDelete" --query "Versions[?(LastModified<'$deleteBefore' && (contains(LastModified, 'T0') || contains(LastModified, 'T1') || contains(LastModified, 'T20') || contains(LastModified, 'T21') || contains(LastModified, 'T22')))].{Key: Key, VersionId: VersionId}"`
cat << EOF > $fileName
{"Objects":$versionsToDelete, "Quiet":true}
EOF
aws s3api delete-objects --bucket "$bucket" --delete file://$fileName
# s3api delete-objects can handle upto 1000 records
echo "Delete successsful"
# We leave the aws_delete.json file, in case you want to later see what happened.

If the obj list is greater than 1000, the operation times out. Better to handle the objects in loop of some kind

For deleting all objects with the above command, use this snippet to loop through all object list 1000 objects at a time:
#Get old version Objects
echo "Old version objects under this prefix:"
cat "all-objects-$bucket.json" | jq '.[] | select(.IsLatest | not)' | jq -s '.' > "old-objects-$bucket.json"
no_of_obj=$(cat old-objects-$bucket.json | jq 'length')
i=0
while [ $i -lt $no_of_obj ]
do
next=$((i+999))
oldversions=$(cat "old-objects-$bucket.json" | jq '.[] | {Key,VersionId}' | jq -s '.' | jq .[$i:$next])
cat << EOF > deleted-files-start-index-$i.json
{"Objects":$oldversions, "Quiet":true}
EOF
echo "Deleting records from $i - $next"
aws s3api delete-objects --bucket "$bucket" --delete file://deleted-files-start-index-$i.json
let i=i+1000
done