Compare hashes in Python

compare_hashes.py

#!/usr/bin/python3

import zlib
import base64
import random
import string
import hashlib

def crc32(text):
	return zlib.crc32(text.encode())

def adl32(text):
	return zlib.adler32(text.encode())
	
def mix32(text):
	return zlib.crc32(text.encode()) ^ zlib.adler32(text.encode())

def mix64(text):
	return zlib.crc32(text.encode()) << 32 | zlib.adler32(text.encode())

def pyt32(text):
	return hash(text) >> 32

def pyt64(text):
	return hash(text)

def sha32(text):
	return int.from_bytes(hashlib.sha1(text.encode()).digest()[:4], byteorder='big')

def sha64(text):
	return int.from_bytes(hashlib.sha1(text.encode()).digest()[:8], byteorder='big')

def random_string():
	length = random.choice(range(1, 16))
	return ''.join(random.choices(string.ascii_lowercase, k=length))

def random_strings(size):
	strings = set()
	while len(strings) < size:
		strings.add(random_string())
	return strings

size = 10**7 # 10,000,000
strings = random_strings(size)

crc32_hashes = []
adl32_hashes = []
mix32_hashes = []
mix64_hashes = []
pyt32_hashes = []
pyt64_hashes = []
sha32_hashes = []
sha64_hashes = []

for i in strings:
     crc32_hashes.append(crc32(i))
     
for i in strings:
     adl32_hashes.append(adl32(i))
     
for i in strings:
     mix32_hashes.append(mix32(i))

for i in strings:
     mix64_hashes.append(mix64(i))

for i in strings:
     pyt32_hashes.append(pyt32(i))

for i in strings:
     pyt64_hashes.append(pyt64(i))
     
for i in strings:
     sha32_hashes.append(sha32(i))

for i in strings:
     sha64_hashes.append(sha64(i))

print()
print('SIZE: ', size)
print()
print('UNIQUENESS 32:')
print(' - CRC32             ', 1.0 * len(set(crc32_hashes)) / len(crc32_hashes))
print(' - ADL32             ', 1.0 * len(set(adl32_hashes)) / len(adl32_hashes))
print(' - CRC32 ^ ADL32     ', 1.0 * len(set(mix32_hashes)) / len(mix32_hashes))
print(' - PYTHON HASH 32    ', 1.0 * len(set(pyt32_hashes)) / len(pyt32_hashes))
print(' - SHA1 HASH 32      ', 1.0 * len(set(sha32_hashes)) / len(sha32_hashes))
print()
print('UNIQUENESS 64:')
print(' - CRC32<<32 | ADL32 ', 1.0 * len(set(mix64_hashes)) / len(mix64_hashes))
print(' - PYTHON HASH 64    ', 1.0 * len(set(pyt64_hashes)) / len(pyt64_hashes))
print(' - SHA1 HASH 64      ', 1.0 * len(set(sha64_hashes)) / len(sha64_hashes))
print()

OUTPUT:

SIZE:  10000000

UNIQUENESS 32:
 - CRC32              0.9987718
 - ADL32              0.055102
 - CRC32 ^ ADL32      0.9988351
 - PYTHON HASH 32     0.9988358
 - SHA1 HASH 32       0.9988395

UNIQUENESS 64:
 - CRC32<<32 | ADL32  1.0
 - PYTHON HASH 64     1.0
 - SHA1 HASH 64       1.0

TIME:

real	2m15,187s
user	2m9,221s
sys	0m4,342s

LINKS:

• Using both CRC32 and Adler32
• How to hash in python