raw_data_processing/lettercount.py

   1 import collections
   2 import string
   3 from szyfrow.support.utilities import sanitise
   4
   5 corpora = ['shakespeare.txt', 'sherlock-holmes.txt', 'war-and-peace.txt']
   6 counts = collections.Counter()
   7
   8 for corpus in corpora:
   9     text = sanitise(open('szyfrow/support/' + corpus, 'r').read())
  10     counts.update(text)
  11
  12 sorted_letters = sorted(counts, key=counts.get, reverse=True)
  13
  14 with open('szyfrow/support/count_1l.txt', 'w') as f:
  15     for l in sorted_letters:
  16         f.write("{0}\t{1}\n".format(l, counts[l]))
  17
  18