Human Readable Sizes

Brandon Rozek

March 15, 2021

When playing with large and small values, it is useful to convert them to a different unit in scientific notation. Let’s look at bytes.

size_categories = ["B", "KB", "MB", "GB", "TB"]

You can figure out how to best represent it by seeing how many of the base (in this case 1000) fits into the value. $$ category = \lfloor \frac{\log{(size_{bytes})}}{\log{(base)}} \rfloor $$ You’ll want to make sure that you don’t overflow in the number of categories you have

category_num = min(category_num, len(size_categories))

You can then get its category representation by $$ size = \frac{size_{bytes}}{(2^{category})} $$ We can wrap this all up info a nice python function

def humanReadableBytes(num_bytes: int) -> str:
    base = 1000
    # Zero Case
    if num_bytes == 0:
        return "0"
    size_categories = ["B", "KB", "MB", "GB", "TB"]
    category_num = int(math.log(num_bytes) / math.log(base))
    # Make sure it doesn't overflow
    category_num = min(category_num, len(size_categories) - 1)
    return "{:.2f} ".format(num_bytes / (base ** category_num)) + \