Python 3.x rounding half up

Question

I know that questions about rounding in python have been asked multiple times already, but the answers did not help me. I'm looking for a method that is rounding a float number half up and returns a float number. The method should also accept a parameter that defines the decimal place to round to. I wrote a method that implements this kind of rounding. However, I think it does not look elegant at all.

def round_half_up(number, dec_places):
    s = str(number)

    d = decimal.Decimal(s).quantize(
        decimal.Decimal(10) ** -dec_places,
        rounding=decimal.ROUND_HALF_UP)

    return float(d)

I don't like it, that I have to convert float to a string (to avoid floating point inaccuracy) and then work with the decimal module. Do you have any better solutions?

Edit: As pointed out in the answers below, the solution to my problem is not that obvious as correct rounding requires correct representation of numbers in the first place and this is not the case with float. So I would expect that the following code

def round_half_up(number, dec_places):

    d = decimal.Decimal(number).quantize(
        decimal.Decimal(10) ** -dec_places,
        rounding=decimal.ROUND_HALF_UP)

    return float(d)

(that differs from the code above just by the fact that the float number is directly converted into a decimal number and not to a string first) to return 2.18 when used like this: round_half_up(2.175, 2) But it doesn't because Decimal(2.175) will return Decimal('2.17499999999999982236431605997495353221893310546875'), the way the float number is represented by the computer. Suprisingly, the first code returns 2.18 because the float number is converted to string first. It seems that the str() function conducts an implicit rounding to the number that was initially meant to be rounded. So there are two roundings taking place. Even though this is the result that I would expect, it is technically wrong.

Answer 1

Rounding is surprisingly hard to do right, because you have to handle floating-point calculations very carefully. If you are looking for an elegant solution (short, easy to understand), what you have like like a good starting point. To be correct, you should replace decimal.Decimal(str(number)) with creating the decimal from the number itself, which will give you a decimal version of its exact representation:

d = Decimal(number).quantize(...)...

Decimal(str(number)) effectively rounds twice, as formatting the float into the string representation performs its own rounding. This is because str(float value) won't try to print the full decimal representation of the float, it will only print enough digits to ensure that you get the same float back if you pass those exact digits to the float constructor.

If you want to retain correct rounding, but avoid depending on the big and complex decimal module, you can certainly do it, but you'll still need some way to implement the exact arithmetics needed for correct rounding. For example, you can use fractions:

import fractions, math

def round_half_up(number, dec_places=0):
    sign = math.copysign(1, number)
    number_exact = abs(fractions.Fraction(number))
    shifted = number_exact * 10**dec_places
    shifted_trunc = int(shifted)
    if shifted - shifted_trunc >= fractions.Fraction(1, 2):
        result = (shifted_trunc + 1) / 10**dec_places
    else:
        result = shifted_trunc / 10**dec_places
    return sign * float(result)

assert round_half_up(1.49) == 1
assert round_half_up(1.5) == 2
assert round_half_up(1.51) == 2
assert round_half_up(2.49) == 2
assert round_half_up(2.5) == 3
assert round_half_up(2.51) == 3

Note that the only tricky part in the above code is the precise conversion of a floating-point to a fraction, and that can be off-loaded to the as_integer_ratio() float method, which is what both decimals and fractions do internally. So if you really want to remove the dependency on fractions, you can reduce the fractional arithmetic to pure integer arithmetic; you stay within the same line count at the expense of some legibility:

def round_half_up(number, dec_places=0):
    sign = math.copysign(1, number)
    exact = abs(number).as_integer_ratio()
    shifted = (exact[0] * 10**dec_places), exact[1]
    shifted_trunc = shifted[0] // shifted[1]
    difference = (shifted[0] - shifted_trunc * shifted[1]), shifted[1]
    if difference[0] * 2 >= difference[1]:  # difference >= 1/2
        shifted_trunc += 1
    return sign * (shifted_trunc / 10**dec_places)

Note that testing these functions brings to spotlight the approximations performed when creating floating-point numbers. For example, print(round_half_up(2.175, 2)) prints 2.17 because the decimal number 2.175 cannot be represented exactly in binary, so it is replaced by an approximation that happens to be slightly smaller than the 2.175 decimal. The function receives that value, finds it smaller than the actual fraction corresponding to the 2.175 decimal, and decides to round it down. This is not a quirk of the implementation; the behavior derives from properties of floating-point numbers and is also present in the round built-in of Python 3 and 2.

Answer 2

I don't like it, that I have to convert float to a string (to avoid floating point inaccuracy) and then work with the decimal module. Do you have any better solutions?

Yes; use Decimal to represent your numbers throughout your whole program, if you need to represent numbers such as 2.675 exactly and have them round to 2.68 instead of 2.67.

There is no other way. The floating point number which is shown on your screen as 2.675 is not the real number 2.675; in fact, it is very slightly less than 2.675, which is why it gets rounded down to 2.67:

>>> 2.675 - 2
0.6749999999999998

It only shows in string form as '2.675' because that happens to be the shortest string such that float(s) == 2.6749999999999998. Note that this longer representation (with lots of 9s) isn't exact either.

However you write your rounding function, it is not possible for my_round(2.675, 2) to round up to 2.68 and also for my_round(2 + 0.6749999999999998, 2) to round down to 2.67; because the inputs are actually the same floating point number.

So if your number 2.675 ever gets converted to a float and back again, you have already lost the information about whether it should round up or down. The solution is not to make it float in the first place.