ccs.py

#
# usage: python ccs.py {file path}
#

import sys
import re
import collections
from matplotlib import pyplot as plt

def get_cs(fn):
    r = re.compile(r'\\([a-zA-Z@]+|[^a-zA-Z])')
    return [e for l in [re.findall(r, l) for l in open(fn)] for e in l]

def frequency_ranking(ls):
    cd = collections.Counter(ls)
    return [[k, v] for k,v in cd.most_common()]

if __name__ == '__main__':
    fn = sys.argv[1]
    cls = get_cs(fn)

    ls = frequency_ranking(cls)

    plt.plot([v[1]/len(ls) for v in ls])
    plt.xscale('log')
    plt.yscale('log')
    plt.xlabel('順位')
    plt.ylabel('出現頻度')
    plt.show()