FileChecker.py

starts = ['binaries', 'Source', 'Config', 'Content']

import os
import os.path as path

def get_file_data(filepath):
	if path.isfile(filepath):
		sz = path.getsize(filepath)
		#print("For %s is %d"%(filepath, sz))
		return filepath, sz
	print ("ERROR")
	return 0, 0

def get_dir_data(d):
	r = []
	if path.isdir(d):
		for p in os.listdir(d):
			pp = d + '/' + p
			if path.isdir(pp):
				r += get_dir_data(pp)
			elif path.isfile(pp):
				r.append(get_file_data(pp))
			else:
				continue

		return r

	print ("ERROR")
	return r

r = []
for s in starts:
	print("Started %s"%s)
	t = get_dir_data(s)
	r += t

def mean(x): return len(x) and sum(x)/len(x)

fs = [len, sum, max, min, mean]
szs = [x[1] for x in r]
for f in fs:
	print(f, f(szs))

print ("max file is %s", max(r, key=lambda x:x[1])[0])

mxs = sorted(r, key=lambda x:x[1], reverse=True)
print("Maxes")
for x in mxs[:20]:
	print ('%s\t\t\tis\t%f Mb'%(x[0], x[1]/1024/1024))
sm = sum(map(lambda x: x[1], mxs[:20]))/1024/1024
print("SUM is %f"%sm)