Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time
Branch Status
Master: Build Status Appveyor Build Status Coverage Status Requirements Status PyPI version
Dev: Build Status Appveyor Build Status Coverage Status Requirements Status is a Python 3.5+ library for integration testing and data validation through configurable and optional runtime type hint enforcement. It uses the standard type hinting syntax (defined in PEP 484).

NOTICE: Python versions 3.5.2 and earlier (3.5.0-3.5.2) are now deprecated. Only Python versions 3.5.3+ would be supported. Deprecated versions will no longer be officially supported in version 0.4.x.


  • Supports most of simple and nested types
  • Supports Callables, TypeVars and Generics
  • Supports invariant, covariant, contravariant and bivariant type checking
    • Default mode is invariant - the type has to match exactly, which is better suitable for testing but differs from Python's normal covariant type checking (a subclass can be used wherever a parent class is expected).
  • Can be applied to both functions and classes (in this case it will be applied to all methods of the class)
  • Highly configurable
    • Global on/off switch
    • Group configuration
    • Local override of groups
    • Type checking mode selection
    • Dynamic reconfiguration


Stable 0.3.x - Stable and ready for every day use version

pip install enforce

Dev current - "Bleeding edge" features that, while are fairly consistent, may change.

pip install git+


Type enforcement is done using decorators around functions that you desire to be checked. By default, this decorator will ensure that any variables passed into the function call matches its declaration (invariantly by default). This includes integers, strings, etc. as well as lists, dictionaries, and more complex objects. Currently, the type checking is eager.

Note, eager means that for a large nested structure, every item in that structure will be checked. This may be a nightmare for performance! See caveats for more details.

You can also apply the runtime_validation decorator around a class, and it will enforce the types of every method in that class.

Note: this is a development feature and is not as thoroughly tested as the function decorators.


Basic type hint enforcement

>>> import enforce
>>> @enforce.runtime_validation
... def foo(text: str) -> None:
...     print(text)
>>> foo('Hello World')
Hello World
>>> foo(5)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/william/.local/lib/python3.5/site-packages/enforce/", line 106, in universal
    _args, _kwargs = enforcer.validate_inputs(parameters)
  File "/home/william/.local/lib/python3.5/site-packages/enforce/", line 69, in validate_inputs
    raise RuntimeTypeError(exception_text)
  The following runtime type errors were encountered:
       Argument 'text' was not of type <class 'str'>. Actual type was <class 'int'>.

Callable Support

def foo(a: typing.Callable[[int, int], str]) -> str:
    return a(5, 6)

def bar(a: int, b: int) -> str:
    return str(a * b)

class Baz:
    def __call__(self, a: int, b: int) -> str:
      return bar(a, b)


TypeVar and Generics

T = typing.TypeVar('T', int, str)

class Sample(typing.Generic[T]):
    def get(self, data: T) -> T:
        return data

def foo(data: Sample[int], arg: int) -> int:
    return data.get(arg)

def bar(data: T, arg: int) -> T:
    return arg

sample_good = Sample[int]()
sample_bad = Sample()

with self.assertRaises(TypeError):
    sample = Sample[list]()

foo(sample_good, 1)

with self.assertRaises(RuntimeTypeError):
    foo(sample_bad, 1)

bar(1, 1)

with self.assertRaises(RuntimeTypeError):
    bar('str', 1)

Class Decorator

Applying this decorator to a class will automatically apply the decorator to every method in the class.

class DoTheThing(object):
    def __init__(self):
        self.do_the_stuff(5, 6.0)

    def do_the_stuff(self, a: int, b: float) -> str:
        return str(a * b)

NamedTuple supports typed NamedTuples.

MyNamedTuple = typing.NamedTuple('MyNamedTuple', [('param', int)])

# Optionally making a NamedTuple typed
# It will now enforce its type signature
# and will throw exceptions if there is a type mismatch
# MyNamedTuple(param='str') will now throw an exception
MyNamedTuple = runtime_validation(MyNamedTuple)

# This function now accepts only NamedTuple arguments
def foo(data: MyNamedTuple):
    return data.param


You can assign functions to groups, and apply options on the group level.

'None' leaves previous value unchanged.

All available global settings:

default_options = {
    # Global on/off switch
    'enabled': None,
    # Group related settings
    'groups': {
        # Dictionary of type {<name: str>: <status: bool>}
        # Sets the status of specified groups
        # Enable - True, disabled - False, do not change - None
        'set': {},
        # Sets the status of all groups to False before updating
        'disable_previous': False,
        # Sets the status of all groups to True before updating
        'enable_previous': False,
        # Deletes all the existing groups before updating
        'clear_previous': False,
        # Updating the default group status - default group is not affected by other settings
        'default': None
    # Sets the type checking mode
    # Available options: 'invariant', 'covariant', 'contravariant', 'bivariant' and None
    'mode': None
# Basic Example
def foo(a: List[str]):

foo(1)	# No exception as the 'best_group' was not explicitly enabled
# Group Configuration
enforce.config({'groups': {'set': {'best_group': True}}}) # Enabling group 'best_group'

with self.assertRaises(RuntimeTypeError):

    'groups': {
        'set': {
            'foo': True
        'disable_previous': True,
        'default': False
    })  # Disable everything but the 'foo' group

# Using foo's settings
def test1(a: str): return a

# Using foo's settings but locally overriding it to stay constantly enabled
@runtime_validation(group='foo', enabled=False)
def test2(a: str): return a

# Using bar's settings - deactivated group -> no type checking is performed
def test3(a: str): return a

# Using bar's settings but overriding locally -> type checking enabled
@runtime_validation(group='bar', enabled=True)
def test4(a: str): return a

with self.assertRaises(RuntimeTypeError):
with self.assertRaises(RuntimeTypeError):


enforce.config({'enabled': False})  # Disables


enforce.config({'enabled': True})  # Re-enables

enforce.config(reset=True) # Resets global settings to their default state


Currently, iterators, generators and coroutines type checks are not supported (mostly). However, it is still possible to check if an object is iterable.

We are still working on the best approach for lazy type checking (checking list items only when accessed) and lazy type evaluation (accepting strings as type hints).

Currently, the type checker will examine every object in a list. This means that for large structures performance can be a nightmare.

Class decorators are not as well tested, and you may encounter a bug or two. Please report an issue if you do find one and we'll try to fix it as quickly as possible.


0.3.4 - 11.06.2017

  • Further improved exception messages and their consistency
  • General bug fixes

0.3.3 - 23.04.2017

  • Improved support for Dictionaries
  • Fixed some thread safety issues

0.3.2 - 29.01.2017

  • Added support for Python 3.5.3 and 3.6.0
  • Added support for NamedTuple
  • Added support for Set
  • New exception message generation system
  • Fixed failing nested lists type checking

0.3.1 - 17.09.2016

  • Added support for Callable classes (classes with __call__ method are now treated like any other Callable object)
  • Fixed bugs in processing callables without specified return type


Please check out our active issues on our Github page to see what work needs to be done, and feel free to create a new issue if you find a bug.

Actual development is done in the 'dev' branch, which is merged to master at milestones.