COMP6441-birthday-attack/enforcer.py

from multiprocessing import Pool
from hashlib import sha256
import time
import sys

# Function to calculate the hash for the given file content with proper newline handling
def calculate_hash(file_lines_with_newline, num_chars):
    # Join file lines with '\n' to ensure correct structure
    file_content = "".join(file_lines_with_newline)
    return sha256(file_content.encode()).hexdigest()[-num_chars:]

# Function to generate a modified fake file by adding spaces based on the bit pattern
def modify_and_hash(args):
    fake_og, num_chars, bit_pattern, real_hash = args
    
    # Modify the fake file based on the bit pattern
    fake_modified = [
        line.rstrip() + " " * ((bit_pattern >> idx) & 1) + "\n"
        for idx, line in enumerate(fake_og)
    ]
    
    # Calculate the hash for the modified fake file
    fake_hash = calculate_hash(fake_modified, num_chars)
    
    # Return if the hashes match
    return (fake_hash == real_hash, fake_hash, fake_modified)

def main(real_file, fake_file, num_chars):
    # Read the original real file and retain newline characters
    with open(real_file, 'r') as f:
        real_og = f.readlines()  # This keeps newlines intact
    
    # Read the original fake file and retain newline characters
    with open(fake_file, 'r') as f:
        fake_og = f.readlines()  # This keeps newlines intact

    # Calculate the hash of the real file (unmodified)
    real_hash = calculate_hash(real_og, num_chars)
    print(f"Real file hash: {real_hash}")

    hash_counter = 0  # To track the number of hashes calculated
    start_time = time.time()  # Start the timer

    # Use multiprocessing Pool
    with Pool() as pool:
        i = 0
        batch_size = 100
        found_collision = False

        while not found_collision:
            # Prepare a batch of bit patterns to process in parallel
            bit_patterns = [(fake_og, num_chars, pattern, real_hash) for pattern in range(i, i + batch_size)]
            
            # Process the batch in parallel
            results = pool.map(modify_and_hash, bit_patterns)

            # Update the hash counter
            hash_counter += batch_size

            # Display progress on the same line
            elapsed_time = time.time() - start_time
            sys.stdout.write(f"\rProcessed {hash_counter} hashes in {elapsed_time:.2f} seconds.")
            sys.stdout.flush()

            # Check the results to see if a collision was found
            for match, fake_hash, fake_modified in results:
                if match:
                    elapsed_time = time.time() - start_time
                    print(f"\nCollision found! The fake file's hash matches the real file's hash: {real_hash}")
                    print(f"Total hashes processed: {hash_counter} in {elapsed_time:.2f} seconds.")

                    # Write the modified fake file to the output
                    with open(f"{fake_file}.out", 'w') as f_out:
                        f_out.writelines(fake_modified)
                    
                    found_collision = True
                    break

            i += batch_size

if __name__ == "__main__":
    if len(sys.argv) != 4:
        print(f"Usage: {sys.argv[0]} <real_file> <fake_file> <num_chars>")
        sys.exit(1)

    real_file = sys.argv[1]
    fake_file = sys.argv[2]
    num_chars = int(sys.argv[3])

    main(real_file, fake_file, num_chars)