3 minute read

Password cracking through Bruteforcing may take a long time and most of the users usually use common English words or names, and numbers as their passwords. Therefore, Dictionary attacks can be quite useful to crack the passwords.

A dictionary is a simple txt file that may contain from a few thousands to a few millions of common words or phrases (includes numbers as well).

If you have a stolen user credential database, you might be able to crack the passwords by matching all dictionary words against the hashed passwords!

A Hash is typically a one-way function that creates a unique digest from an input string. For example, if your password is hello_there, the output hash digest would look like the following

Algorithm Output Hash
MD5 290e1c9cc54995453b810dfb15b853a1
SHA-1 f976f0ad02501f0a95bfcf3c1081e0759b508d47
SHA-256 299d2e40d6b7026b6029b8ff4cff0ad0fbfe14b20d704a609a2631cada32fbc1

Here, MD5, SHA-1, and SHA-256 are widely used hashing algorithms to convert a string into a one-way output.

The term one-way means you cannot retrive the string from the hashed output. It is important to hash the passwords because we do not want to keep passwords in plain sight.

If you are logging into your social media account (e.g., Facebook/Instagram) or your laptop, you have to input your username and password. However, the systems maintain a user database containing your name and the hashed password. For example,

user_name password_hash
mrx 678cfd979de6de0ec70d08b0b7a4b6aad645802abe2504aed9c4d1ca3da101c5
guest_user 3809f08dc16f01a7c9393eab3146e38e7ffd7d19b0aa5a3754aa2f7780fc4b77
other b712430498d0b31595b86c34a939b4dcdde5050a4e8a143e99037a6e6984a68f

Now, suppose, you have stolen/found a user credential database like this. However, because you cannot calculate the string, you need to match different guessed strings after hashing.

Requirements

  1. Stolen User Credential Database (two columns: username, password_hash)
  2. A dictionary file (each line contains a dictionary word)
  3. Attacker should know which algorithm is being used. Although it’s not a big deal. The attacker can try all algorithms one after another. In this post we will use SHA-256

Create a simple Python Program

First, let’s import the necessary modules

import pandas as pd
from hashlib import sha256

Now, we need to write a function that returns True if the hashed dictionary word matches the database password

def dictionary_attack(dictionary_word,target_hash):
    pass_bytes = dictionary_word.encode('utf-8')
    pass_hash = sha256(pass_bytes)
    digest = pass_hash.hexdigest()
    if digest == target_hash:
        return True

Suppose, we want to check whether we can crack the password of the first user. Therefore, we need to get the first password_hash from the user database, and pass all the dictionary words one after another.

if __name__ == "__main__":
    # read user DB and the dictionary using pandas.read_csv
    dictionary = pd.read_csv("password_dictionary", names=['passwords'])
    users = pd.read_csv("users")
    
    # match all words from the dictionary until it matches/ends
    for test_word in dictionary["passwords"]:
        if dictionary_attack(test_word,users["password_hash"][0]) == True:
            print("Matched Password: ", test_word)
            break
        else:
            continue

Now, within a few seconds, we get a matched word from our dictionary: 2midrash.

Well, if you want to compromise all user password, you can add a new for loop that iterates over the indices of all users. Or for any range of user, you can create a function like this and call it in the main function.

# inputs an string and returns the sha256 digest
def create_hash(word):
    pass_bytes = word.encode('utf-8')
    pass_hash = sha256(pass_bytes)
    digest = pass_hash.hexdigest()
    return digest

# inputs the number of users and returns nothing
# intended for finding passwords for multiple users
def find_multiple_users(num_user):
    for i in range(num_user):
        check_pass = users["password_hash"][i]
        for test_word in dictionary["passwords"]:
            if create_hash(test_word) == check_pass:
                print("Found Matched Password:", test_word, "for user", users["username"][i])
                break

However, the code we used above is enefficient while calculating for multiple users. Can you guess, why?

Because we are creating hash each time we match against the user database passwords. So, if we try 20000 dictionary words againt 20 user passwords, the number of hashing calculation would be 20000*20.

To avoid the additional computation, we can precompute the hash and keep those values in a dictionary. The new dictionary will have the dictionary words as keys and the hashed outputs will be the values. Therefore, we can add an extra function to the code as follows

# Returns the dictionary of password and corresponding digests
def hash_dictionary():
    global hash_dict
    hash_dict = {}
    for test_word in dictionary["passwords"]:
        hash_dict[test_word] = create_hash(test_word)
    return hash_dict

You can check the code and files and get some idea about different experiment settings. The four experiments are as follows:

  • Experiment 1: Time taken to match a dictionary word for the first user
  • Experiment 2: Approximate Avg. time for all users
  • Experiment 3: approx time if used the Hash dictionary (precomputed hash)
  • Experiment 4: Adding Salt to user password to avoid duplicates

That’s all for today, cheers!

Leave a comment