Count character frequencies

charFrequency.py

#!/usr/bin/python
# charFrequency.py (C) 2013 Yrjo Kari-Koskinen <[email protected]>
#
# This program is licensed under Version 2 of the GPL.
#
# Use PYTHONIOENCODING=utf_8 environment variable to read and write utf8 to files.

import sys, unicodedata

def letter(l):
    return unicodedata.category(l)[0] == 'L'

f = {}
sum = 0

for line in sys.stdin.readlines():
    for char in unicode(line.rstrip(), "utf-8"):
        if letter(char):
            sum = sum + 1
            if char in f:
                f[char] = f[char] + 1
            else:
                f[char] = 1

keys = sorted(f, key=f.get, reverse=True)

for key in keys:
    print key, f[key], f[key] * 100 / sum, "%"