#!/usr/bin/env bash

set -e

! read -rd '' AWK_PREFILTER <<'EOF'
/(\/|^)[0-9]{4}-[0-9]{2}-[0-9]{2}-[0-9]{6}$/ {
	a[$0]
}

END {
	n = asorti(a)
	for (i = n; i >= 1; --i)
		print a[i]
}
EOF

! read -rd '' AWK_FILTER <<'EOF'
function flr(n, s) {
	return int(n / s) * s
}

function mstart(t) {
	return mktime(strftime("%Y %m 01 00 00 00", t, 1), 1)
}

BEGIN {
	n = split("h d w M", arr)
	for (i = 1; i <= n; ++i) {
		b = arr[i]

		if (!match(opt_retain, "([*0-9]+)(\\^?)" b, md))
			continue

		ret[b] = md[1]
		newest_in[b] = (md[2] == "^")
		buckets[++n_buckets] = b
	}

	split("sun mon tue wed thu fri sat", arr)
	for (i in arr)
		arr2[arr[i]] = i - 1
	wshift = (11 - arr2[opt_week_start])*24*60*60
}

NR == 1 {
	++keep[$0]
}

{
	t = mktime(gensub(/(.*\/|^)(.+)-(..)-(..)-(..)(..)(..)$/,
		"\\2 \\3 \\4 \\5 \\6 \\7", 1), 1)
	if (t < 0) {
		++keep[$0]
		next
	}

	bt["h"] = flr(t, 60*60)
	bt["d"] = flr(t, 24*60*60)
	bt["w"] = flr(t + wshift, 7*24*60*60) - wshift
	bt["M"] = mstart(t)

	for (i = 1; i <= n_buckets; ++i) {
		b = buckets[i]

		if (b == "M" && "w" in ret) {
			if (newest_in["w"]) {
				if (mstart(bt["w"] + 6*24*60*60) > bt["M"])
					next
			} else if (bt["w"] < bt["M"]) {
				next
			}
		}

		if (newest_in[b] && bt[b] == last[b, b] && (i == 1 ||
				bt[buckets[i-1]] != last[b, buckets[i-1]]))
			next

		if ((b, bt[b]) in bkeep || ret[b] == "*" || bc[b]++ < ret[b])
			bkeep[b, bt[b]] = $0

		for (j in buckets)
			last[b, buckets[j]] = bt[buckets[j]]
	}
}

END {
	for (i in bkeep)
		++keep[bkeep[i]]

	n = asorti(keep)
	for (i = 1; i <= n; ++i)
		print keep[i]
}
EOF

run() {
	echo "* ${@}"
	"${@}"
}

current() {
	local str="${1}" interval="${2}"
	local s=0 val d

	while [[ "${interval}" =~ ([0-9]+)(.?) ]]; do
		val="${BASH_REMATCH[1]}"
		case "${BASH_REMATCH[2]}" in
			d) ((s += val * 24*60*60)) ;;
			h) ((s += val * 60*60)) ;;
			m) ((s += val * 60)) ;;
			*) ((s += val)) ;;
		esac

		interval="${interval#*${BASH_REMATCH[0]}}"
	done
	((s < 1)) && return 1

	d="${str:0:10} ${str:11:2}:${str:13:2}:${str:15:2}"
	(($(date -u +%s) / s == $(date -ud "${d}" +%s) / s))
}

backup() {
	local dest_dir="${1}" src_path="${2}"
	local retain="${3}" interval="${4}" week_start="${5}"
	local dest_path snaps keep skip i ki

	mapfile -td $'\0' snaps < <(shopt -s nullglob;
		printf '%s\0' "${dest_dir}/"* | awk -v 'RS=\0' -v 'ORS=\0' \
			-- "${AWK_PREFILTER}")
	[[ -n "${snaps}" ]] && current "${snaps[0]##*/}" "${interval}" && \
		skip=1
	printf '[%d] [%s] %s => %s%s\n' "$((++dcount))" "${retain}" \
		"${src_path}" "${dest_dir}" "${skip:+ [SKIPPED]}"
	[[ -n "${skip}" ]] && return

	dest_path="${dest_dir}/$(date -u '+%Y-%m-%d-%H%M%S')"
	run rsync -ac --mkpath ${snaps[0]:+"--link-dest=${snaps[0]}"} \
		"${src_path}" "${dest_path}"
	snaps+=("${dest_path}")
	mapfile -td $'\0' snaps < <(
		printf '%s\0' "${snaps[@]}" | awk -v 'RS=\0' -v 'ORS=\0' \
			-- "${AWK_PREFILTER}")

	mapfile -td $'\0' keep < <(
		printf '%s\0' "${snaps[@]}" | awk -v 'RS=\0' -v 'ORS=\0' \
			-v "opt_retain=${retain}" \
			-v "opt_week_start=${week_start}" \
			-- "${AWK_FILTER}")
	if [[ -z "${keep}" ]]; then
		echo "WARNING: dirs to keep shouldn't be empty; skipping" >&2
		return
	fi

	for ((i = ${#snaps[@]} - 1, ki = 0; i >= 0; --i)); do
		if [[ "${snaps[i]}" == "${keep[ki]}" ]]; then
			((++ki))
		else
			run rm -r "${snaps[i]}"
		fi
	done
}

declare -A def_opts=(
	['retain']='*h' # e.g. 24h7d4w12M or 24^h10d*M
	['interval']='0'
	['week_start']='mon'
)
declare -A opts

while IFS='=' read -r f1 f2; do
	[[ -z "${f1}" ]] || [[ "${f1}" == '#'* ]] && continue

	if [[ "${f1}" =~ ^\[(.*)]$ ]]; then
		base_path="${BASH_REMATCH[1]}"
		declare -A opts="$(declare -p def_opts | sed '1s/[^=]*=//')"
	elif [[ "${f1}" == '$'* ]]; then
		opt="${f1:1}"
		[[ "${f2}" == '-' ]] && opts["${opt}"]= ||
			opts["${opt}"]="${f2:-"${def_opts["${opt}"]}"}"
	else
		backup "${base_path}/${f1}" "${f2}" \
			"${opts['retain']}" \
			"${opts['interval']}" \
			"${opts['week_start']}"
	fi
done < "${1}"