class-inspector 0.2.0

class inspector package

FunctionInspector

This class is used for inspecting a function, it stores the parameters and any type hints as well as the return annotation. This is then used to generate parametrized test stubs to be filled out and guard conditions for the types as derived from the input type hints.

It can be used to inspect functions that aren't your own or used to aid development.

Example

Example function:

def test_function(param1: float, param2: int, param3: bool) -> float:
    if param3:
        return param1 - param2
    else:
        return param1 + param2

Returned guard conditions:

usage:

func_insp.analyse(test_function)
func_insp.add_guards()

output:

def test_function(param1: float, param2: int, param3: bool) -> float:
    if not all([isinstance(param1, float), isinstance(param2, int), isinstance(param3, bool)]):
        raise TypeError(
            "test_function expects arg types: [float, int, bool], "
            f"received: [{type(param1).__name__}, {type(param2).__name__}, {type(param3).__name__}]"
        )
    if param3:
        return param1 - param2
    else:
        return param1 + param2

Returned parametrized tests:

usage:

func_insp.analyse(test_function)
func_insp.get_test(check_types=True)

output:

@pytest.mark.parametrize(
    "param1, param2, param3, expected_result, expected_context",
    [
        (param1, param2, param3, expected_result, expected_context),
        (param1, param2, param3, None, pytest.raises(TypeError)),
        (param1, param2, param3, None, pytest.raises(TypeError)),
        (param1, param2, param3, None, pytest.raises(TypeError)),
    ]
)
def test_test_function(param1, param2, param3, expected_result, expected_context) -> None:
    with expected_context:
        assert test_function(param1, param2, param3) == expected_result

Compared to previous versions the tests are now combined to pass in the expected context, this could be whatever error a parameter set could raise or a successful assertion with the null_context passed as does_not_raise(). A type error raise test is created for each of the parameters if the check_types argument is True. The match argument can also be used to check the error message is what is expected too.

An example is below:

@pytest.mark.parametrize(
    "param1, param2, param3, expected_result, expected_context",
    [
        (3, 2, True, 1, does_not_raise()),
        ("3", 2, False, 0, pytest.raises(TypeError))
    ]
)
def test_test_function(param1, param2, param3, expected_result, expected_context) -> None:
    with expected_context:
        assert test_function(param1, param2, param3) == expected_result

There is potential for the type testing to use hypothesis @given parameters however that's on the TODO list.

ModuleInspector

A wrapper around FunctionInspector that can be given a module and perform FunctionInspector actions on the group of functions or classes.

ClassInspector

A wrapper around FunctionInspector that can be given a class and perform FunctionInspector actions on the methods of that class.

Custom Validators

Extension of the validators from the attrs library, these allow the use of type checking the constituents of a collection.

Example attrs validator and the problem this is a solution to*:

possible: list = attr.ib(validator=[instance_of(list)])
not_possible: List[int] = attr.ib(validator=[instance_of(List[int])])

This means that we can't validate the members of a collection. However, with the custom validators we can do this:

now_possible: Collection = attr.ib(validator=[validate_generic_of_type(Collection, float)])

Meaning we can pass in any object that implements the following dunder methods [__contains__, __iter__, __len__] AND validate that each member of the collection is a float. This is possible for other types too.

*on closer inspection (ironic given this repo's name) of the attrs API reference this is a solved problem with deep_iterable() and deep_mapping(). attrs: 2, me: 0

AttrGenerator

This is an intermediate step between the ultimate goal of a standard class to attrs convertor, this converts a list of attributes into an attr class with type based validators

Example

from class_inspector.attr_generator import AttrGenerator, AttrMap

attributes = [
        AttrMap("test1", "int", True),
        AttrMap("test2", "float", False),
        AttrMap("test3", "bool", True),
        AttrMap("test4", "str", False),
        AttrMap("test5", "List[int]", True),
        AttrMap("test6", "Dict[str, float]", False),
    ]

at_gen = AttrGenerator("TestClass", attributes)
at_gen.get_attr_class()

output

import attr
from attr.validators import instance_of, deep_iterable, deep_mapping


@attr.define
class TestClass:
    test1: int = attr.ib(validator=[instance_of(int)])
    test2: float = attr.ib(validator=[instance_of(float)], init=False)
    test3: bool = attr.ib(validator=[instance_of(bool)])
    test4: str = attr.ib(validator=[instance_of(str)], init=False)
    test5: list = attr.ib(validator=[deep_iterable(member_validator=instance_of(int), iterable_validator=instance_of(list))])
    test6: dict = attr.ib(validator=[deep_mapping(key_validator=instance_of(str), value_validator=instance_of(float), mapping_validator=instance_of(dict))], init=False)

AttrMap class is a wrapper to avoid a lot of duplicated dictionaries, however this is also valid input:

attributes = [
    {'attr_name': 'test1', 'attr_type': 'int', 'attr_init': True},
    {'attr_name': 'test2', 'attr_type': 'float', 'attr_init': False},
    {'attr_name': 'test3', 'attr_type': 'bool', 'attr_init': True},
    {'attr_name': 'test4', 'attr_type': 'str', 'attr_init': False},
    {'attr_name': 'test5', 'attr_type': 'List[int]', 'attr_init': True},
    {'attr_name': 'test6', 'attr_type': 'Dict[str, float]', 'attr_init': False}
]