PrivacyScanData/truncate/new2/git-birthday-master/git-birthday

#!/usr/bin/env python

import sys
import os
import math
from termcolor import colored

if len(sys.argv) > 1:
    trunc = int(sys.argv[1])
else:
    trunc = 7

def birthday(people,days):
    """The chance of a number of people sharing a birthday
    in a year with a certian number of days."""
    return 1 - math.prod([days - i for i in range(0,people)]) / ( days ** people )

def birthday_threshold(threshold,days):
    """How many people you need for the collision probability
    to reach the threshold, as a fraction of 1."""
    return int(1/2 * (1 + math.sqrt(4*(2 * days * math.log(1/(1 - threshold))) + 1)))

def max_hashes(trunc):
    return 16 ** trunc

commits  =  os.popen('git log --pretty=format:"%H %aI %s"').read().split("\n")
hashes   =  [s.split()[0] for s in commits]
dates    =  [s.split()[1] for s in commits]
subjects =  [s.split(" ", 2)[2] for s in commits]

trunc_hashes = [s[0:trunc] for s in hashes]

dict_hashes = {}
for h in enumerate(trunc_hashes):
    if h[1] not in dict_hashes:
        dict_hashes[h[1]]=[h[0]]
    else:
        dict_hashes[h[1]].append(h[0])

dict_duplicates = dict(filter(lambda elem: len(elem[1]) > 1,dict_hashes.items()))

def printmatch(string,length):
    print(colored(string[0:length], "red") , string[length:],sep="")

print ("For truncation length", colored(trunc,"red") ,"there are", end=" ")
if len(dict_duplicates) == 0:
    print(colored("no","yellow"),"commits with the same hash!")
else:
    print(colored("{:d}".format(len(dict_duplicates)),"cyan"), "commits pairs with the same hash! They are:\n")
    for el in dict_duplicates.items():
        for idx in el[1]:
            print("-"*40)
            print(colored(subjects[idx],"blue"))
            print(colored(dates[idx],"yellow"))
            printmatch(hashes[idx],trunc)
        print("-"*40)
        print("")
    print("There are", len(dict_duplicates), "collisions in total.")
print("The probability of a collision happening at least once is", end=" ")
print(colored("{:.2f}%".format(birthday(len(commits),max_hashes(trunc)) * 100), "green"))
print("For this probability to reach 1%, you'd need", colored("{:d}".format(birthday_threshold(0.01,max_hashes(trunc))),"cyan"), "commits.")
print("For this probability to reach 10%, you'd need", colored("{:d}".format(birthday_threshold(0.1,max_hashes(trunc))),"cyan"), "commits.")
print("For this probability to reach 50%, you'd need", colored("{:d}".format(birthday_threshold(0.50,max_hashes(trunc))),"cyan"), "commits.")