Password Analyser

password_analysis.py

#!/usr/bin/env python
#
# Fecking ugly thing I wrote to parse lulzsec's password releases
# and display some fairly anonymous data.
# Usage: password_analysis.py <filename to process>
# App expects Email | Password
# Easily modified to fit a file format.

import re
import operator
import sys,os
import string
import pprint
from math import sqrt
pp = pprint.PrettyPrinter()
passwords = {}
pass_counts = {}
onecharpass = []
passfile = open(sys.argv[1],'r')
for line in passfile:
        if line.count('|') == 2:
                password, email, f = line.split('|')
                email = email.rstrip().lstrip()
                password = password.rstrip().lstrip()
                if password in passwords:
                        passwords[password] += 1
                else:
                        passwords[password] = 1

# Interesting Things
# Max / Min lengths
# Num passwords containing punctuation
# Avg length
# standard deviation of length
# number of occurrences
pass_count = sorted(passwords.iteritems(), key=operator.itemgetter(1))
for pc in pass_count:
        pass_counts[pc[0]]=pc[1]

print "Top 100 Passwords, and their frequency"
pp.pprint(pass_count[len(pass_count)-100:])

print "Longest password: "
l = ""
for p in passwords:
        if len(p) > len(l):
                l = p
print l, len(l), "chars long"
for p in passwords:
        if len(p) == 1 and p not in onecharpass:
                onecharpass.append(p)
print "Shortest passwords, and their occurrences"
for o in onecharpass:
        print o, "occurred ", pass_counts[o], "times"

t = 0.0
for p in passwords:
        t += len(p)
mean = t/len(passwords.keys())
print "Average password length: ", mean

std = 0.0
for p in passwords:
        std = std + (len(p) - mean)**2
std = sqrt(std/float(len(passwords.keys())-1))

print "Standard Deviation: ", std

punpwds = []
punc = list(string.punctuation)
for f in punc:
        for p in passwords:
                if f in p:
                        punpwds.append(p)
print len(punpwds), "Passwords contain punctuation"
print "Here's a selection:"
print punpwds[:10]