attrs_strict documentation

Background

The purpose of the library is to provide runtime validation for attributes specified in attrs data classes. The types supported are all the builtin types and most of the ones defined in the typing library. For Python 2, the typing module is available through the backport found here.

Getting started

Run pip install attrs-strict to install the latest stable version from PyPi. The source code is hosted on github at https://github.com/bloomberg/attrs-strict. The library currently supports Python2.7, Python3.6 and Python3.7.

Usage and examples

Type enforcement is based on the type attribute set on any field specified in an attrs dataclass. If the type argument is not specified no validation takes place.

from typing import List

import attr

from attrs_strict import type_validator, ContainerError

>>> @attr.s
... class SomeClass(object):
...     list_of_numbers = attr.ib(
...         validator=type_validator(),
...         type=List[int]
...     )
...

>>> sc = SomeClass([1,2,3,4])
>>> sc
SomeClass(list_of_numbers=[1, 2, 3, 4])

>>> SomeClass([1,2,3,'four'])
attrs_strict._error.AttributeTypeError(
  "list_of_numbers must be typing.List[int]"
  "(got four that is a <class 'str'>) in [1, 2, 3, 'four']"
)

Nested type exceptions are validated acordingly, and a backtrace to the initial container is maintained to ease with debugging. This means that if an exception occurs because a nested element doesn’t have the correct type, the representation of the exception will contain the path to the specific element that caused the exception.

from typing import List, Tuple

import attr

from attrs_strict import type_validator, ContainerError

>>> @attr.s
... class SomeClass(object):
...     names = attr.ib(
...        validator=type_validator(), type=List[Tuple[str, str]]
...     )

>>> sc = SomeClass(names=[('Moo', 'Moo'), ('Zoo',123)])
attrs_strict._error.AttributeTypeError(
  "names must be"
  "typing.List[typing.Tuple[str, str]] (got 123 that is a <class 'int'>) in"
  "('Zoo', 123) in [('Moo', 'Moo'), ('Zoo', 123)]"
)

What is currently supported ?

Currently there’s support for builtin types and types specified in the typing module: List, Dict, DefaultDict, Set, Union, Tuple and any combination of them. This means that you can specify nested types like List[List[Dict[int, str]]] and the validation would check if attribute has the specific type.

Callables, TypeVars or Generics are not supported yet but there are plans to support this in the future.

• API Documentation