Randomness Sources for Python
Current Features
- unified API to randomness sources
- API access to the system's true randomness generator
- a list of randomness providers to choose from
- provider flags which you can filter for
Planned Features
- comprehensive tests for randomness quality (dieharder, ENT, ...)
- more PRNGs (Wichmann-Hill, xorshift, ...)
- more TRNGs (VIA Padlock RNG, external hardware, ...)
- better API to filter and sort providers
- choice between wasteful and conserving usage of random bits, the latter most likely with a mixin
- adapter to provide the Numpy API for all randomness sources, most likely by way of a mixin
Usage
You can instantiate a randomness source directly if you know it's there (or if you handle the exception in case it's not):
from randomness import URandom
try:
from randomness import Random
except ImportError:
pass
random = URandom()
Or you can peruse the list of providers and select one out of those (possibly filtering the list first):
from randomness import PROVIDERS, Flag
# filter randomness providers
my_providers = [provider for provider in PROVIDERS
if provider.flags & Flag.NEVER_BLOCKING]
# dump provider names
for provider in my_providers:
print(provider.name)
# instantiate a randomness context
chosen_provider = my_providers[0]
random = chosen_provider.cls()
From this point forward usage is the same as if you had done
from random import Random; random = Random()
.
# perform a die roll
die_result = random.randint(1, 6)
print(die_result)
# select a random cheese
CHEESE_SHOP_PRODUCTS = ["Tilsit", "Cheddar", "Roquefort", "Gouda"]
cheese = random.choice(CHEESE_SHOP_PRODUCTS)
print(cheese)
# draw lottery numbers
amount_of_numbers = 49
amount_to_draw = 6
series = random.sample(range(1, amount_of_numbers + 1),
amount_to_draw)
print(series)