hashcmp.py

import hashlib
import time

def files_are_equal(contents1, contents2):
    for b1, b2 in zip(contents1, contents2):
        if b1 != b2:
            return False

    return True

def content(filename):
    with open(filename, 'rb') as f:
        return f.read()

content1 = content('largefile')
content2 = content('largefile2')
content3 = content('largefile3')

start = time.time()
print("One and two are equal: {}".format(
      files_are_equal(content1, content2)))
print("One and three are equal: {}".format(
      files_are_equal(content1, content3)))
end = time.time()
print("Content comparison took: {}".format(end - start))

start = time.time()
hash1 = hashlib.md5(content1).hexdigest()
hash2 = hashlib.md5(content2).hexdigest()
hash3 = hashlib.md5(content3).hexdigest()
print("One and two are equal: {}".format(hash1 == hash2))
print("One and three are equal: {}".format(hash1 == hash3))
end = time.time()
print("Hash comparison took: {}".format(end - start))