benchmark.py

import hashlib
from datetime import datetime
import os

def benchmark_hash_function():
    for i in range(1, 65536, 1000):
        content = b"a" * i

        begin = datetime.now()
        hash_function = hashlib.md5()
        hash_function.update(content)
        hash_result = hash_function.hexdigest()
        time_taken = datetime.now() - begin

        print (f"{i},{time_taken.total_seconds() * 1000}")

def benchmark_file_hash():
    hash_function = hashlib.md5()
    for i in range(2, 50):
        content = b"a" * 65536 * i
        with open(f"benchmark-file-{i}.tmp", "w+b") as f:
            f.write(content)

    for j in range(2, 50):
        with open(f"benchmark-file-{j}.tmp", "rb") as f:
            begin = datetime.now()
            f.seek(0, os.SEEK_END)
            size = f.tell()
            hash_function = hashlib.md5()
            f.seek(0, os.SEEK_SET)
            data = f.read(65536)        
            hash_function.update(data)
            f.seek(-65536, os.SEEK_END)
            data = f.read(65536)
            hash_function.update(data)
            hash_result = hash_function.hexdigest()
            time_taken = datetime.now() - begin
            print (f"{size},{time_taken.total_seconds() * 1000}")

benchmark_file_hash()