support/lettercount.py

   1 import collections
   2 import string
   3 from utilities import sanitise
   4
   5 # def sanitise(text):
   6 #     return [l.lower() for l in text if l in string.ascii_letters]
   7
   8 corpora = ['shakespeare.txt', 'sherlock-holmes.txt', 'war-and-peace.txt']
   9 counts = collections.defaultdict(int)
  10
  11 for corpus in corpora:
  12     text = sanitise(open(corpus, 'r').read())
  13     for letter in text:
  14         counts[letter] += 1
  15
  16 sorted_letters = sorted(counts, key=counts.get, reverse=True)
  17
  18 with open('count_1l.txt', 'w') as f:
  19     for l in sorted_letters:
  20         f.write("{0}\t{1}\n".format(l, counts[l]))
  21
  22