runtime-type-checker

This package performs type-check at runtime with help of type annotations.

How to use this package

There are two ways to perform type checks using this package.

I provide a few simple examples here. For a complete overview, have a look at the package's unit tests.

1- the check_type function

You can check an object against a type or an annotation via the check_type function.

The function returns None if the check was successful or raises a TypeError in case of error.

Note that this function does not check recursively for e.g. the attributes of a class.

from typing import List, Sequence, Optional, Mapping
from dataclasses import dataclass
from runtime_type_checker import check_type


check_type("a", str)  # OK
check_type(["a"], List[str])  # OK
check_type(["a", 1], Sequence[str])  # raises TypeError


@dataclass
class Foo:
    a: int
    b: Optional[Mapping[str, int]] = None


check_type(Foo(1), Foo)  # OK
check_type(Foo(1), int)  # raises TypeError

2- The check_types decorator

You can also type-check classes upon instance creation and functions or methods upon call through the check_types decorator:

from typing import Optional, Mapping
from dataclasses import dataclass
from runtime_type_checker import check_types

def run_typed(f):
  return check_types(dataclass(f))

@check_types
@dataclass
class Foo:
    a: int
    b: Optional[Mapping[str, int]] = None


Foo(1)              # returns an instance of foo
Foo(0, {"a": "b"})  # raises TypeError


@check_types
def bar(a: bool, **options: str) -> str:
    return options.get("b", "missing") if a else "unknown"

bar(True, b="1")  # returns "1"
bar(True, c=1)    # raises TypeError

Package features and short-comings

1- Features

• simplicity: there's only one function and one decorator to keep in mind.
• robustness: this package relies on the typing-inspect for the heavy lifting. This package is maintained by core contributors to the typing module, which means very little hacks on my side to work with older versions of python.

2- Short-comings

• coverage: I don't offer coverage for all features of type annotations: for example Protocol, Generators, IO are not currently supported. Generics are not really well handled.