Python random module provides a convenient way for generating pseudo-random numbers in case you need some unpredictable results for your application such as the computer games, a lucky draw system or even the shuffling logic for your music player. Since it provides various functions to generate results in “unpredictable” manner, developers attempted to use this feature to produce random password or authentication token for security purpose without understanding of it’s fundamental implementation. In this article, we will be discussing how the Python random module has been misunderstood and misused for the scenarios which it shall not be used.
Basic usage of Python random module
Let’s take a look at some basic usage of this module. You can use it to generate random integers, float numbers or bytes as per below:
#Generate a random integer between 1 to 10
#2 #generate a random floating point number between 0 to 1 random.random()
#0.3103975786510934#Generate random number between 1 to 2 in uniform distribution
#1.9530600469459607#Generate random number between 1 to 100, with step as 2
random.randrange(1, 100, 2)
#43#Generate random bytes, available in Python version 3.9 random.randbytes(8)
#b'v\xf7\xb2v`\xc8U]' #Generate a integer within 8 bits
And shuffling or sampling the items in a sequence can be achieved easily with below:
slangs = ["Boomer", "Cap", "Basic", "Retweet", "Fit", "Fr", "Clout"]random.shuffle(slangs)
#['Fit', 'Basic', 'Fr', 'Clout', 'Cap', 'Retweet', 'Boomer']
#['Fit', 'Fr', 'Clout', 'Retweet', 'Basic']
You can also use the choice function to choose a random option from a given sequence, for instance:
random.choice(["Boomer", "Cap", "Basic", "Retweet", "Fit", "Fr", "Clout"])
With this function, It’s easy to implement a lucky draw logic where all the participants’ name are passed in and a winner is selected randomly which seems to be the perfect use case of it. The problem comes when developers try to use this function to generate password or security related tokens which they believe it’s random enough and resistant to predication. But it is indeed wrong.
Why the random numbers generated are not random enough?
To answer this question, we will need to take a further look at the implementation of this Python random module. The module uses Mersenne Twister algorithm as the core generator which is designed for modelling and simulation purpose rather than security or cryptography. Some study show it’s not difficult to reconstruct the internal state of the MT to predict the outcome and it can be attacked through this MT randomness.
Actually random module does provide a SystemRandom class which uses the sources from operation system to generate random numbers without relying on the software state and the result is not reproducible. For instance, depends on the actual implementation, some OS uses the atmospheric noise or the exact time of the key presses as the source for generating unpredictable result which is more suitable for cryptography.
You can use it in the same way as the random class except the state is not available:
sys_random = random.SystemRandom()sys_random.random()
#72 sys_random.choice(["Boomer", "Cap", "Basic", "Retweet", "Fit", "Fr", "Clout"])
Unfortunately, this class has been overlooked for many years and some developers continue using the functions from random module for generating password or security tokens which exposed a lot of security vulnerability. To address these concerns, an enhancement proposal raised to add a new secrets module for some common security-related functions, so that people won’t mix up it with the random module.
Python secrets module
Let’s take a look at the functions in secrets module:
#Generate random integer between 0 to 50
#2#Generate random integer with 8 bits
#125#Generate random bytes
#b'\x8a\xb3\xa92)S:\xf2\xac\x90\xaf\xb1\xb3Q\xc5\xfe\x80\xdb\xc2`'#Generate random string in hexadecimal with default 32 bytes
#'cf4f964edf810ca7f6ad6b533038fdcf538c73bd59e11d3340a838e7fd88fdf9'#Generate URL-safe text string
#z9zQQsxcq6SRug import string secrets.choice(string.ascii_letters + string.digits + string.punctuation)
The functions are pretty similar to what we have seen in random module, just that results are generated from os.urandom function which meant to be unpredictable enough for cryptographic applications.
Implementing a strong password generator with Python secrets
With all above said, let’s implement a strong password generator with the secrets module. Assuming we have below requirements on our password:
- Length between 8 to 16
- At least 1 lowercase
- At least 1 uppercase
- At least 1 number
- At least 1 special characters among !#$%&@_~
We can use the choice function to randomly choose 1 character each time for a random iteration between 8 to 16 times, then test if all the requirements are met for the generated password. Below the sample code:
def generate_strong_password(): special_characters = '!#$%&@_~'
password_choices = string.ascii_letters + string.digits + special_characters while True:
password = ''.join(secrets.choice(password_choices) for _ in range(random.randint(8, 16))) if (any(c.islower() for c in password)
and any(c.isupper() for c in password)
and sum(special_characters.find(c) > -1 for c in password) == 1
and any(c.isdigit() for c in password)): break
To check the randomness, you can try the below:
[generate_strong_password() for _ in range(10)] # sample output
In this article, we have reviewed through the most commonly used functions in Python random module for generating pseudo-random results for modelling and simulation. Due to misuse of random module for security related applications from the earlier Python developers, Python has introduced a new secrets module to focus on password, authentication or security related tokens. We have also demonstrated an example on how to use secrets module to implement a strong password generator. Although you can generate a strong password with the combination of the alphanumeric and special characters, you shall never store your password in plaintext or simple encrypted format, a good practice is to always use libraries like hashlib or bcrypt to salt and hash it before storing.
Originally published at https://www.codeforests.com on December 27, 2020.