Source code for ABXpy.misc.type_fitting

import numpy as np


def fit_integer_type(n, is_signed=True):
    """Returns the minimal space needed to store integers of maximal value n"""

    if is_signed:
        m = 1
        types = [np.int8, np.int16, np.int32, np.int64]
    else:
        m = 0
        types = [np.uint8, np.uint16, np.uint32, np.uint64]

    if n < 2 ** (8 - m):
        return types[0]
    elif n < 2 ** (16 - m):
        return types[1]
    elif n < 2 ** (32 - m):
        return types[2]
    elif n < 2 ** (64 - m):
        return types[3]
    else:
        raise ValueError(
            'Values are too big to be represented by 64 bits integers!')