gostyle_old, commit old
[gostyle.git] / pdb-gtl / stats.py
blobb51ccbe4f0207ac60cc762d616554aaf4868faa9
1 #!/usr/bin/python
3 import os, random, shutil, numpy
5 merg= 'rawpat_files_merged'
6 test=merg+'_test'
8 def list_dir(raw):
9 import os, random, shutil
10 ranks = os.listdir(raw)
11 for rank in ranks:
12 plays = os.listdir(raw + '/'+ rank)
13 for play in plays:
14 yield raw + '/' + rank + '/' + play
16 def dir_stats(dr):
17 num = 0
18 sizes = []
20 for f in list_dir(dr):
21 num += 1
22 size = os.stat(f).st_size
23 sizes += [size]
25 return (num, numpy.array(sizes).mean(), sum(sizes))
27 num, mean, size = dir_stats(merg)
28 numt, meant, sizet = dir_stats(test)
30 print "Test set X Train Set"
31 print "Mean filesize:", meant, " X ", mean
32 print "Number of files:", numt, " X ", num
33 print "Total size:", sizet, " X ", size