CPyUG Top 100

   1 #!/usr/bin/env python
   2 # -*- coding: UTF-8 -*-
   3 
   4 import re, sys, os
   5 from operator import itemgetter
   6 from collections import defaultdict
   7 
   8 pat =  re.compile('From:\s+?.*<([\da-z_\.-]+@[\da-z\.-]+)>')
   9 mails = defaultdict(int)
  10 for line in open(sys.argv[1], 'rb'):
  11    if line[0] != 'F': continue
  12    mat = pat.match(line)
  13    if mat:
  14        mails[mat.group(1)] += 1
  15 
  16 lst = sorted(mails.iteritems(), key=itemgetter(1), reverse=True)
  17 for t in lst[:100 if 100 < len(lst) else len(lst)]:
  18    print '%s : %d' % (t[0].ljust(16), t[1])