443 lines
11 KiB
Python
443 lines
11 KiB
Python
"""RSA module
|
|
pri = k[1] //Private part of keys d,p,q
|
|
|
|
Module for calculating large primes, and RSA encryption, decryption,
|
|
signing and verification. Includes generating public and private keys.
|
|
|
|
WARNING: this code implements the mathematics of RSA. It is not suitable for
|
|
real-world secure cryptography purposes. It has not been reviewed by a security
|
|
expert. It does not include padding of data. There are many ways in which the
|
|
output of this module, when used without any modification, can be sucessfully
|
|
attacked.
|
|
"""
|
|
|
|
__author__ = "Sybren Stuvel, Marloes de Boer and Ivo Tamboer"
|
|
__date__ = "2010-02-05"
|
|
__version__ = '1.3.3'
|
|
|
|
# NOTE: Python's modulo can return negative numbers. We compensate for
|
|
# this behaviour using the abs() function
|
|
|
|
from cPickle import dumps, loads
|
|
import base64
|
|
import math
|
|
import os
|
|
import random
|
|
import sys
|
|
import types
|
|
import zlib
|
|
|
|
from rsa._compat import byte
|
|
|
|
# Display a warning that this insecure version is imported.
|
|
import warnings
|
|
warnings.warn('Insecure version of the RSA module is imported as %s, be careful'
|
|
% __name__)
|
|
|
|
def gcd(p, q):
|
|
"""Returns the greatest common divisor of p and q
|
|
|
|
|
|
>>> gcd(42, 6)
|
|
6
|
|
"""
|
|
if p<q: return gcd(q, p)
|
|
if q == 0: return p
|
|
return gcd(q, abs(p%q))
|
|
|
|
def bytes2int(bytes):
|
|
"""Converts a list of bytes or a string to an integer
|
|
|
|
>>> (128*256 + 64)*256 + + 15
|
|
8405007
|
|
>>> l = [128, 64, 15]
|
|
>>> bytes2int(l)
|
|
8405007
|
|
"""
|
|
|
|
if not (type(bytes) is types.ListType or type(bytes) is types.StringType):
|
|
raise TypeError("You must pass a string or a list")
|
|
|
|
# Convert byte stream to integer
|
|
integer = 0
|
|
for byte in bytes:
|
|
integer *= 256
|
|
if type(byte) is types.StringType: byte = ord(byte)
|
|
integer += byte
|
|
|
|
return integer
|
|
|
|
def int2bytes(number):
|
|
"""Converts a number to a string of bytes
|
|
|
|
>>> bytes2int(int2bytes(123456789))
|
|
123456789
|
|
"""
|
|
|
|
if not (type(number) is types.LongType or type(number) is types.IntType):
|
|
raise TypeError("You must pass a long or an int")
|
|
|
|
string = ""
|
|
|
|
while number > 0:
|
|
string = "%s%s" % (byte(number & 0xFF), string)
|
|
number /= 256
|
|
|
|
return string
|
|
|
|
def fast_exponentiation(a, p, n):
|
|
"""Calculates r = a^p mod n
|
|
"""
|
|
result = a % n
|
|
remainders = []
|
|
while p != 1:
|
|
remainders.append(p & 1)
|
|
p = p >> 1
|
|
while remainders:
|
|
rem = remainders.pop()
|
|
result = ((a ** rem) * result ** 2) % n
|
|
return result
|
|
|
|
def read_random_int(nbits):
|
|
"""Reads a random integer of approximately nbits bits rounded up
|
|
to whole bytes"""
|
|
|
|
nbytes = ceil(nbits/8.)
|
|
randomdata = os.urandom(nbytes)
|
|
return bytes2int(randomdata)
|
|
|
|
def ceil(x):
|
|
"""ceil(x) -> int(math.ceil(x))"""
|
|
|
|
return int(math.ceil(x))
|
|
|
|
def randint(minvalue, maxvalue):
|
|
"""Returns a random integer x with minvalue <= x <= maxvalue"""
|
|
|
|
# Safety - get a lot of random data even if the range is fairly
|
|
# small
|
|
min_nbits = 32
|
|
|
|
# The range of the random numbers we need to generate
|
|
range = maxvalue - minvalue
|
|
|
|
# Which is this number of bytes
|
|
rangebytes = ceil(math.log(range, 2) / 8.)
|
|
|
|
# Convert to bits, but make sure it's always at least min_nbits*2
|
|
rangebits = max(rangebytes * 8, min_nbits * 2)
|
|
|
|
# Take a random number of bits between min_nbits and rangebits
|
|
nbits = random.randint(min_nbits, rangebits)
|
|
|
|
return (read_random_int(nbits) % range) + minvalue
|
|
|
|
def fermat_little_theorem(p):
|
|
"""Returns 1 if p may be prime, and something else if p definitely
|
|
is not prime"""
|
|
|
|
a = randint(1, p-1)
|
|
return fast_exponentiation(a, p-1, p)
|
|
|
|
def jacobi(a, b):
|
|
"""Calculates the value of the Jacobi symbol (a/b)
|
|
"""
|
|
|
|
if a % b == 0:
|
|
return 0
|
|
result = 1
|
|
while a > 1:
|
|
if a & 1:
|
|
if ((a-1)*(b-1) >> 2) & 1:
|
|
result = -result
|
|
b, a = a, b % a
|
|
else:
|
|
if ((b ** 2 - 1) >> 3) & 1:
|
|
result = -result
|
|
a = a >> 1
|
|
return result
|
|
|
|
def jacobi_witness(x, n):
|
|
"""Returns False if n is an Euler pseudo-prime with base x, and
|
|
True otherwise.
|
|
"""
|
|
|
|
j = jacobi(x, n) % n
|
|
f = fast_exponentiation(x, (n-1)/2, n)
|
|
|
|
if j == f: return False
|
|
return True
|
|
|
|
def randomized_primality_testing(n, k):
|
|
"""Calculates whether n is composite (which is always correct) or
|
|
prime (which is incorrect with error probability 2**-k)
|
|
|
|
Returns False if the number if composite, and True if it's
|
|
probably prime.
|
|
"""
|
|
|
|
q = 0.5 # Property of the jacobi_witness function
|
|
|
|
# t = int(math.ceil(k / math.log(1/q, 2)))
|
|
t = ceil(k / math.log(1/q, 2))
|
|
for i in range(t+1):
|
|
x = randint(1, n-1)
|
|
if jacobi_witness(x, n): return False
|
|
|
|
return True
|
|
|
|
def is_prime(number):
|
|
"""Returns True if the number is prime, and False otherwise.
|
|
|
|
>>> is_prime(42)
|
|
0
|
|
>>> is_prime(41)
|
|
1
|
|
"""
|
|
|
|
"""
|
|
if not fermat_little_theorem(number) == 1:
|
|
# Not prime, according to Fermat's little theorem
|
|
return False
|
|
"""
|
|
|
|
if randomized_primality_testing(number, 5):
|
|
# Prime, according to Jacobi
|
|
return True
|
|
|
|
# Not prime
|
|
return False
|
|
|
|
|
|
def getprime(nbits):
|
|
"""Returns a prime number of max. 'math.ceil(nbits/8)*8' bits. In
|
|
other words: nbits is rounded up to whole bytes.
|
|
|
|
>>> p = getprime(8)
|
|
>>> is_prime(p-1)
|
|
0
|
|
>>> is_prime(p)
|
|
1
|
|
>>> is_prime(p+1)
|
|
0
|
|
"""
|
|
|
|
nbytes = int(math.ceil(nbits/8.))
|
|
|
|
while True:
|
|
integer = read_random_int(nbits)
|
|
|
|
# Make sure it's odd
|
|
integer |= 1
|
|
|
|
# Test for primeness
|
|
if is_prime(integer): break
|
|
|
|
# Retry if not prime
|
|
|
|
return integer
|
|
|
|
def are_relatively_prime(a, b):
|
|
"""Returns True if a and b are relatively prime, and False if they
|
|
are not.
|
|
|
|
>>> are_relatively_prime(2, 3)
|
|
1
|
|
>>> are_relatively_prime(2, 4)
|
|
0
|
|
"""
|
|
|
|
d = gcd(a, b)
|
|
return (d == 1)
|
|
|
|
def find_p_q(nbits):
|
|
"""Returns a tuple of two different primes of nbits bits"""
|
|
|
|
p = getprime(nbits)
|
|
while True:
|
|
q = getprime(nbits)
|
|
if not q == p: break
|
|
|
|
return (p, q)
|
|
|
|
def extended_euclid_gcd(a, b):
|
|
"""Returns a tuple (d, i, j) such that d = gcd(a, b) = ia + jb
|
|
"""
|
|
|
|
if b == 0:
|
|
return (a, 1, 0)
|
|
|
|
q = abs(a % b)
|
|
r = long(a / b)
|
|
(d, k, l) = extended_euclid_gcd(b, q)
|
|
|
|
return (d, l, k - l*r)
|
|
|
|
# Main function: calculate encryption and decryption keys
|
|
def calculate_keys(p, q, nbits):
|
|
"""Calculates an encryption and a decryption key for p and q, and
|
|
returns them as a tuple (e, d)"""
|
|
|
|
n = p * q
|
|
phi_n = (p-1) * (q-1)
|
|
|
|
while True:
|
|
# Make sure e has enough bits so we ensure "wrapping" through
|
|
# modulo n
|
|
e = getprime(max(8, nbits/2))
|
|
if are_relatively_prime(e, n) and are_relatively_prime(e, phi_n): break
|
|
|
|
(d, i, j) = extended_euclid_gcd(e, phi_n)
|
|
|
|
if not d == 1:
|
|
raise Exception("e (%d) and phi_n (%d) are not relatively prime" % (e, phi_n))
|
|
|
|
if not (e * i) % phi_n == 1:
|
|
raise Exception("e (%d) and i (%d) are not mult. inv. modulo phi_n (%d)" % (e, i, phi_n))
|
|
|
|
return (e, i)
|
|
|
|
|
|
def gen_keys(nbits):
|
|
"""Generate RSA keys of nbits bits. Returns (p, q, e, d).
|
|
|
|
Note: this can take a long time, depending on the key size.
|
|
"""
|
|
|
|
while True:
|
|
(p, q) = find_p_q(nbits)
|
|
(e, d) = calculate_keys(p, q, nbits)
|
|
|
|
# For some reason, d is sometimes negative. We don't know how
|
|
# to fix it (yet), so we keep trying until everything is shiny
|
|
if d > 0: break
|
|
|
|
return (p, q, e, d)
|
|
|
|
def gen_pubpriv_keys(nbits):
|
|
"""Generates public and private keys, and returns them as (pub,
|
|
priv).
|
|
|
|
The public key consists of a dict {e: ..., , n: ....). The private
|
|
key consists of a dict {d: ...., p: ...., q: ....).
|
|
"""
|
|
|
|
(p, q, e, d) = gen_keys(nbits)
|
|
|
|
return ( {'e': e, 'n': p*q}, {'d': d, 'p': p, 'q': q} )
|
|
|
|
def encrypt_int(message, ekey, n):
|
|
"""Encrypts a message using encryption key 'ekey', working modulo
|
|
n"""
|
|
|
|
if type(message) is types.IntType:
|
|
return encrypt_int(long(message), ekey, n)
|
|
|
|
if not type(message) is types.LongType:
|
|
raise TypeError("You must pass a long or an int")
|
|
|
|
if message > 0 and \
|
|
math.floor(math.log(message, 2)) > math.floor(math.log(n, 2)):
|
|
raise OverflowError("The message is too long")
|
|
|
|
return fast_exponentiation(message, ekey, n)
|
|
|
|
def decrypt_int(cyphertext, dkey, n):
|
|
"""Decrypts a cypher text using the decryption key 'dkey', working
|
|
modulo n"""
|
|
|
|
return encrypt_int(cyphertext, dkey, n)
|
|
|
|
def sign_int(message, dkey, n):
|
|
"""Signs 'message' using key 'dkey', working modulo n"""
|
|
|
|
return decrypt_int(message, dkey, n)
|
|
|
|
def verify_int(signed, ekey, n):
|
|
"""verifies 'signed' using key 'ekey', working modulo n"""
|
|
|
|
return encrypt_int(signed, ekey, n)
|
|
|
|
def picklechops(chops):
|
|
"""Pickles and base64encodes it's argument chops"""
|
|
|
|
value = zlib.compress(dumps(chops))
|
|
encoded = base64.encodestring(value)
|
|
return encoded.strip()
|
|
|
|
def unpicklechops(string):
|
|
"""base64decodes and unpickes it's argument string into chops"""
|
|
|
|
return loads(zlib.decompress(base64.decodestring(string)))
|
|
|
|
def chopstring(message, key, n, funcref):
|
|
"""Splits 'message' into chops that are at most as long as n,
|
|
converts these into integers, and calls funcref(integer, key, n)
|
|
for each chop.
|
|
|
|
Used by 'encrypt' and 'sign'.
|
|
"""
|
|
|
|
msglen = len(message)
|
|
mbits = msglen * 8
|
|
nbits = int(math.floor(math.log(n, 2)))
|
|
nbytes = nbits / 8
|
|
blocks = msglen / nbytes
|
|
|
|
if msglen % nbytes > 0:
|
|
blocks += 1
|
|
|
|
cypher = []
|
|
|
|
for bindex in range(blocks):
|
|
offset = bindex * nbytes
|
|
block = message[offset:offset+nbytes]
|
|
value = bytes2int(block)
|
|
cypher.append(funcref(value, key, n))
|
|
|
|
return picklechops(cypher)
|
|
|
|
def gluechops(chops, key, n, funcref):
|
|
"""Glues chops back together into a string. calls
|
|
funcref(integer, key, n) for each chop.
|
|
|
|
Used by 'decrypt' and 'verify'.
|
|
"""
|
|
message = ""
|
|
|
|
chops = unpicklechops(chops)
|
|
|
|
for cpart in chops:
|
|
mpart = funcref(cpart, key, n)
|
|
message += int2bytes(mpart)
|
|
|
|
return message
|
|
|
|
def encrypt(message, key):
|
|
"""Encrypts a string 'message' with the public key 'key'"""
|
|
|
|
return chopstring(message, key['e'], key['n'], encrypt_int)
|
|
|
|
def sign(message, key):
|
|
"""Signs a string 'message' with the private key 'key'"""
|
|
|
|
return chopstring(message, key['d'], key['p']*key['q'], decrypt_int)
|
|
|
|
def decrypt(cypher, key):
|
|
"""Decrypts a cypher with the private key 'key'"""
|
|
|
|
return gluechops(cypher, key['d'], key['p']*key['q'], decrypt_int)
|
|
|
|
def verify(cypher, key):
|
|
"""Verifies a cypher with the public key 'key'"""
|
|
|
|
return gluechops(cypher, key['e'], key['n'], encrypt_int)
|
|
|
|
# Do doctest if we're not imported
|
|
if __name__ == "__main__":
|
|
import doctest
|
|
doctest.testmod()
|
|
|
|
__all__ = ["gen_pubpriv_keys", "encrypt", "decrypt", "sign", "verify"]
|
|
|