askdu.py

#!/usr/bin/env python
from __future__ import division, print_function
import sys
import os
import heapq
import time

try:
    input = raw_input
except NameError:
    pass

items = []

t = 0
total_bytes = 0
total_files = 0

ignored = set()

def add_files(path, dev=None):
    global t
    global total_bytes
    global total_files
    if time.time() > t:
        t = time.time() + 0.2
        sys.stdout.write("\rScanning (%s files totaling %s seen so far)..." % (total_files, size_to_hr(total_bytes)))
        sys.stdout.flush()
    total_files += 1
    try:
        statinfo = os.lstat(path)
    except OSError:
        return 0
    if dev is None:
        dev = statinfo.st_dev
    elif statinfo.st_dev != dev:
        return 0
    size = statinfo.st_size
    total_bytes += size
    if statinfo.st_mode & int('170000', 8) == int('40000', 8):
        # directory
        try:
            for subpath in os.listdir(path):
                size += add_files(os.path.join(path, subpath), dev)
        except OSError:
            pass
    heapq.heappush(items, (-size, path))
    return size

def add_items():
    if len(sys.argv) >= 2:
        rootdir = sys.argv[1]
    else:
        default_rootdir = os.path.abspath(os.sep)
        rootdir = input("Enter path to examine [%s]: " % default_rootdir)
        if not rootdir:
            rootdir = default_rootdir
    print('')
    add_files(rootdir)
    print('')

def size_to_hr(size):
    if size > 1048576:
        if size > 1073741824:
            return "%0.2fG" % (size / 1073741824)
        else:
            return "%0.2fM" % (size / 1048576)
    else:
        if size > 1024:
            return "%0.2fK" % (size / 1024)
    if size == 1:
        return "1 byte"
    return "%s bytes" % size


add_items()

while items:

    size, path = heapq.heappop(items)

    head = tail = os.path.normpath(path)
    while tail:
        head, tail = os.path.split(head)
        if head in ignored:
            break
    else:
        size = -size

        print()
        print("%s\t%s" % (size_to_hr(size), path))
        while True:
            try:
                i = input("Expand/Ignore/# of parent levels to ignore/Quit ?").lower()
            except EOFError:
                sys.exit(0)
            if i == 'e':
                break
            elif i == 'i':
                ignored.add(os.path.normpath(path))
                break
            elif i.isdigit():
                for n in range(int(i)):
                    path, _tail = os.path.split(path)
                ignored.add(os.path.normpath(path))
                break
            elif i == 'q':
                sys.exit(0)
                break
            # else continue