How to use Python `secret` module to generate random integer?
Question:
In Python 3.6, new module, secrets, was added.
What is the most efficient way to generate random integer in range [n, m) using this module?
I tried choice(range(n, m)), but I doubt it is the best way.
Answers:
secrets.choice(range(n, m)) should be fine, since range is lazy on Python 3.
n + secrets.randbelow(m-n) is another option. I wouldn’t use it, since it’s less obviously correct.
Since secrets provides access to the SystemRandom class, with the same interface as random.Random, you can also keep your own SystemRandom instance:
my_secure_rng = secrets.SystemRandom()
and do
my_secure_rng.randrange(n, m)
The secret module provides the same interface as random; the underlying random generator has just been changed to SystemRandom which is cryptographically strong.
In short, use it as you would random in circumstances that require a bit more security; I doubt choice suffers from performance so much to warrant your concern.
import secrets
num = secrets.randbelow(50)
print(num)
Not enough rep to comment but to add on to user2357112’s answer if you run a quick execution time test they actually report basically identical.
Best of 3 trials with 1,000,000 loops per trial:
randrange() 1.39906 seconds total time.
choice() 1.39431 seconds total time.
In Python 3.6, new module, secrets, was added.
What is the most efficient way to generate random integer in range [n, m) using this module?
I tried choice(range(n, m)), but I doubt it is the best way.
secrets.choice(range(n, m)) should be fine, since range is lazy on Python 3.
n + secrets.randbelow(m-n) is another option. I wouldn’t use it, since it’s less obviously correct.
Since secrets provides access to the SystemRandom class, with the same interface as random.Random, you can also keep your own SystemRandom instance:
my_secure_rng = secrets.SystemRandom()
and do
my_secure_rng.randrange(n, m)
The secret module provides the same interface as random; the underlying random generator has just been changed to SystemRandom which is cryptographically strong.
In short, use it as you would random in circumstances that require a bit more security; I doubt choice suffers from performance so much to warrant your concern.
import secrets
num = secrets.randbelow(50)
print(num)
Not enough rep to comment but to add on to user2357112’s answer if you run a quick execution time test they actually report basically identical.
Best of 3 trials with 1,000,000 loops per trial:
randrange() 1.39906 seconds total time.
choice() 1.39431 seconds total time.