28. globalformat — Set global ouput format for built-in types.

One of the major shortcomings of Python is the lack of global print options allowing to set the output format for built-in type globally for your application. Especially the default formatting of float values is annoyingly large.

>>> sqrt2 = 2 ** 0.5
>>> print(sqrt2)
1.4142135623730951

To get a shorter output with less decimals one can round the value, or use a format specifier:

>>> print(round(sqrt2, 4))
1.4142
>>> print(f"{sqrt2:.4f}")
1.4142

But this is only convenient when you have to output a single float, but doesn’t work for floats embedded in a container.

>>> cont = (1/2, 1/3, 1/16)
>>> print(cont)
(0.5, 0.3333333333333333, 0.0625)

Luckily, for most numerical work we can (and should) use NumPy, which allows to set global print options for its classes:

>>> import numpy as np
>>> np.set_printoptions(precision=4, floatmode='fixed')
>>> ar = np.array(cont)
>>> print(ar)
[0.5000 0.3333 0.0625]

But even then one falls back to the Python default for a single array element:

>>> print(ar[0], ar[1])
0.5 0.3333333333333333

The solution would be to override float.__repr__, but Python won’t let you:

>>> float.__repr__ = lambda f: f"{f:.4f}"
Traceback (most recent call last):
...
TypeError: cannot set '__repr__' attribute of immutable type 'float'

But since this is Python, there must be some way to overcome this hurdle. And I finally found one: the forbiddenfruit module (https://pypi.org/project/forbiddenfruit/) by Lincoln de Sousa <lincoln@clarete.li> allows to do all kinds of monkeypatching on builtin types. Since we only need a small part of it, we copied the relevant things in this module, thus avoiding that our users need to install that module. Now we can set our own float.__repr__ method, which can take account of a global precision setting.

>>> setfloatformat('.4f')        # set fixed format for all floats
>>> print(cont)
(0.5000, 0.3333, 0.0625)
>>> setfloatformat('')           # reset Python's default
>>> print(cont)
(0.5, 0.3333333333333333, 0.0625)

Or you can use the floatformat context manager to temporarily change the format:

>>> with floatformat('.4f'):
...    print('.4f', cont)
...    with floatformat('10.2e'):
...        print('10.2e', cont)
...    print('back to .4f', cont)
.4f (0.5000, 0.3333, 0.0625)
10.2e (  5.00e-01,   3.33e-01,   6.25e-02)
back to .4f (0.5000, 0.3333, 0.0625)
>>> print('back to default', cont)
back to default (0.5, 0.3333333333333333, 0.0625)

28.1. Classes defined in module globalformat

class globalformat.floatformat(fmt)

Context manager to temporarily change the global float format.

Parameters:

fmt (str) – A format string like in setfloatformat().

Notes

On exit, the global float format is reset to what is was before entering.

28.2. Functions defined in module globalformat

globalformat.setfloatformat(fmt)

Set global default format for float

Parameters:

fmt (str) – A format string to format a single float number. For example: ‘.4f’ will format the float with 4 decimals. An empty string resets the format to Python’s default. An invalid format string raises an Exception.