libmpf.py 文件源码

def to_float(s, strict=False):
    """
    Convert a raw mpf to a Python float. The result is exact if the
    bitcount of s is <= 53 and no underflow/overflow occurs.

    If the number is too large or too small to represent as a regular
    float, it will be converted to inf or 0.0. Setting strict=True
    forces an OverflowError to be raised instead.
    """
    sign, man, exp, bc = s
    if not man:
        if s == fzero: return 0.0
        if s == finf: return math_float_inf
        if s == fninf: return -math_float_inf
        return math_float_inf/math_float_inf
    if sign:
        man = -man
    try:
        if bc < 100:
            return math.ldexp(man, exp)
        # Try resizing the mantissa. Overflow may still happen here.
        n = bc - 53
        m = man >> n
        return math.ldexp(m, exp + n)
    except OverflowError:
        if strict:
            raise
        # Overflow to infinity
        if exp + bc > 0:
            if sign:
                return -math_float_inf
            else:
                return math_float_inf
        # Underflow to zero
        return 0.0