du (disk usage) but for s3 (via ls-bucket)

s3du

#!/usr/bin/env python

import sys
import argparse

parser = argparse.ArgumentParser(description="du your s3")
parser.add_argument("--max-depth", type=int, help="max depth to report")
parser.add_argument("--raw", action="store_true", help="do not pretty print output")

args = parser.parse_args()


class Summary:
    def __init__(self):
        self._sizes = {}
        self._ts = {}

    def record_size(self, path, size):
        if path in self._sizes:
            self._sizes[path] = self._sizes[path] + size
        else:
            self._sizes[path] = size

    def record_ts(self, path, ts):
        if path not in self._ts or ts > self._ts[path]:
            self._ts[path] = ts

    def print(self, raw_output=True):
        for k, v in [x for x in self._sizes.items() if x[1] > 0]:
            if not raw_output:
                try:
                    import humanize
                    print("{} {} {}".format(k, humanize.naturalsize(v, binary=True), self._ts[k]))
                except:
                    print("{} {:,} {}".format(k, v, self._ts[k]))
            else:
                print("{} {} {}".format(k, v, self._ts[k]))


summary = Summary()

def parse_date(date, time):
    return "{}T{}Z".format(date, time)

for line in sys.stdin:
    size, key, date, time, tz = line.strip().split()
    paths = key.split("/")[0:-1]

    if args.max_depth > 0 and len(paths) > args.max_depth:
        path = "/".join(paths[0:args.max_depth])
    elif len(paths) > 0:
        path = "/".join(paths)
    else:
        path = "/"

    summary.record_size(path, int(size))
    summary.record_ts(path, parse_date(date, time))

summary.print(args.raw)