#!/bin/sh

usage() {
	echo "Usage: $0 [-nv] [[[user@]host:]dir]

$0 removes old backups and keeps:
- backups for the current day
- 1 backup per past day for the current month
- 1 backup per past month for the current year
- 1 backup per past year

Options:
-n  dry run
-v  verbose

Environment variables:
- BACKUP - backup directory"
}

while getopts :nv opt; do
	case $opt in
	(n) optn=echo ;;
	(v) optv=-t ;;
	(*) usage; exit 1 ;;
	esac
done
shift $((OPTIND -1))

BACKUP=${1:-$BACKUP}
[ "$BACKUP" ] || { usage; exit 1; }

host=${BACKUP%:*} dir=${BACKUP#*:}
ls='ls -r' rm="xargs -r $optv $optn rm -rf"
[ "$host" = "$dir" ] || ls="ssh $host $ls" rm="ssh $host $rm"

echo "# Cleaning backups on $BACKUP"

$ls "$dir" |
awk -v now=$(date +%Y-%m-%d) -v dir="$dir" '
BEGIN {
	yn = substr(now, 1, 4) # Year now
	mn = substr(now, 6, 2) # Month now
	dn = substr(now, 9, 2) # Day now
}
{
	yb = substr($0, 1, 4)  # Year backup
	mb = substr($0, 6, 2)  # Month backup
	db = substr($0, 9, 2)  # Day backup
	$0 = dir "/" $0

	dy = yn - yb
	dm = dy * 12 + mn - mb
	# if (yb != yn) {
	if (dm > 12) {
		if (yb in yearly) print; else yearly[yb] = 1
		next
	}
	# if (mb != mn) {
	dd = dm * 30 + dn - db
	if (dd > 30) {
		if (mb in monthly) print; else monthly[mb] = 1
		next
	}
	if (db != dn) {
		if (db in dayly) print; else dayly[db] = 1
	}
}' | $rm