hackinteach-pytest-approxable

A pytest approxable dict mixin for data class


License
Apache-2.0
Install
pip install hackinteach-pytest-approxable==0.0.3

Documentation

Pytest Approxable Mixin


Example

from pytest_approxable import Approxable
from dataclasses import dataclass
from pytest import approx

@dataclass
class MetalProperties(Approxable):
    name: str
    heat_capacity: float
    conductivity: float

def test_metal_equal():
    mp1 = MetalProperties(name='metal1', heat_capacity=0.5, conductivity=1.)
    mp2 = MetalProperties(name='metal2', heat_capacity=0.499999999, conductivity=0.9999999998)
    assert approx(mp1.approxable_dict, abs=0.1) == mp2.approxable_dict

Problem

When testing codes, we usually approximate floating numbers to avoid decimal points error. To do so, we use pytest.approx as follow

from pytest import approx
def test_approx():
    a = 1.0000000000000001
    b = 1.
    assert approx(a, abs=0.1) == b # this will pass
    assert a == b # this will fail

Moreover, we can also use this with the dictionary derived from a dataclass. For example,

from dataclasses import dataclass, asdict
from pytest import approx

@dataclass
class MetalProperties:
    heat_capacity: float
    conductivity: float

def test_metal_equal():
    mp1 = MetalProperties(heat_capacity=0.5, conductivity=1.)
    mp2 = MetalProperties(heat_capacity=0.499999999, conductivity=0.9999999998)
    assert approx(asdict(mp1), abs=0.1) == asdict(mp2)

Now the problem arises when we have non-number fields in the class

@dataclass
class MetalProperties:
    name: str
    heat_capacity: float
    conductivity: float

def test_metal_equal():
    mp1 = MetalProperties(name='metal1', heat_capacity=0.5, conductivity=1.)
    mp2 = MetalProperties(name='metal2', heat_capacity=0.499999999, conductivity=0.9999999998)
    assert approx(asdict(mp1), abs=0.1) == asdict(mp2)

Above code will fails since attribute name can't be approximated

##Progress

  • tests
  • supports nested object