rsync-newest

#!/bin/bash
set -eu

sync() {
    local host=${1:?No host given}; shift
    local src=${1:?No src dir given}; shift
    local dst=${1:?No dst dir given}; shift
    local bytes=${1:?No byte capacity given}; shift
    local tmpdir

    tmpdir=$(mktemp -dt rsync-newest.XXXXXX) || return 1
    trap 'rm -rf "$tmpdir"' RETURN

    ssh "$host" 'bash -s' "$src" >"$tmpdir/metadata" <<'EOF' || return 1
find "$1" -type f -printf '%Ts\t%s\t%P\n'
EOF

    sort -k 1nr -k 2n "$tmpdir/metadata" | awk -F $'\t' -v cap="$bytes" '{
        total_size += $2
        if (total_size > cap) {
            exit 0
        }
        print $3
    }' | sort >"$tmpdir/new" &&
    find "$dst" -type f -printf "%P\n" | sort >"$tmpdir/existing" &&
    comm -13 "$tmpdir/new" "$tmpdir/existing" >"$tmpdir/remove" &&
    comm -23 "$tmpdir/new" "$tmpdir/existing" >"$tmpdir/copy" ||
    return 1

    {
        sed 's/^/x /' "$tmpdir/remove"
        sed 's/^/> /' "$tmpdir/copy"
    } | less

    read -p "Continue? [yN] "
    [[ "$REPLY" == y ]] || exit 0

    (
        cd "$dst"
        xargs -d $'\n' rm <"$tmpdir/remove"
        find . -type d -empty -exec rm -d {} +
    )
    rsync -tiz --files-from="$tmpdir/copy" -R "$host:$src" "$dst"
}

usage="\
usage: rsync-newest <host:src> <dst> <capacity>

capacity is parsed by numfmt, so can use SI or IEC suffixes like K, M, G, Ki,
Mi, Gi, etc.
"
while getopts "h" opt; do
    case "$opt" in
    h) echo "$usage"; exit 0;;
    *) exit 1;;
    esac
done
shift $((OPTIND-1))
[[ $# -eq 3 ]] || {
    echo "$usage" >&2
    exit 1
}
raw_src=$1; shift
dst=$1; shift
raw_capacity=$1; shift

host=${raw_src%%:*}
src=${raw_src#*:}
capacity=$(numfmt --from=auto "$raw_capacity")

sync "$host" "$src" "$dst" "$capacity"