typecheck-runtime 0.1.0

Tiny runtime checker for typing annotations (dict[str,str], list[T], unions, literals) with an Option-like API.

typecheck-runtime

Tiny runtime checker for typing annotations with an Option-like API.

py-typecheck solves a very specific problem:

I have a value (object or Any) and I want to check if the value refers to specific typing annotation.

Does not parse data. Does not transforms data. Just checks structure and returns refered value or None if validation failed.

Installation

pip install typecheck-runtime

Python ≥ 3.10

---

Core idea

Main function is checked

from py_typecheck import checked

Acts like Option[T] or Rust-like Some(T, None):

• returns value if matches the type
• returns None if validation failed

No exceptions, no side-efects.

---

Basic usage

value: object = {"a": 1}

if (d := checked(value, dict[str, int])) is not None:
    print(d["a"] + 1)

Important

• always compare against None when using checked
• do not rely on truthiness (0, False, [] are valid values!)
• bool is treated as not matching int (so True wont pass as int)

---

Supported typing constructs

py-typecheck supports common runtime-validable constructs from typing:

Primitive types

checked(1, int)          # 1
checked("x", int)        # None
checked(True, int)       # None

Note: bool is not considered an int in this library.

Union (X | Y, typing.Union)

checked(1, int | str)      # 1
checked("x", int | str)    # "x"
checked(1.5, int | str)    # None

List / Set / FrozenSet

checked([1, 2], list[int])           # [1, 2]
checked({1, 2}, set[int])            # {1, 2}
checked(frozenset({1}), frozenset[int])

Nested structures work as expected:

checked([[1, 2], [3]], list[list[int]])

Dict

checked({"a": 1}, dict[str, int])
checked({"a": "x"}, dict[str, int])   # None

Supports unions and nesting:

checked({"a": [1, 2]}, dict[str, list[int]])

Tuple

Fixed-length:

checked((1, "x"), tuple[int, str])

Variadic:

checked((1, 2, 3), tuple[int, ...])

Literal

from typing import Literal

checked(True, Literal[True])    # True
checked(False, Literal[True])   # None

Any

from typing import Any

checked(object(), Any)   # always matches

---

is_type

There is also a boolean helper:

from py_typecheck import is_type

if is_type(value, dict[str, int]):
    ...

• It returns only True / False.
• For most code paths, checked is preferred, because it avoids type confusion and double-checking mistakes.

---

Design principles

• Explicit is better than clever
• No exceptions for control flow
• No implicit casting
• No data mutation
• No dependency on dataclasses or models
• One function, one responsibility

This is not a validator framework. This is a runtime structural check with a safe return value.

Non-goals

py-typecheck intentionally does not:

• coerce types ("1" → 1)
• fill defaults
• validate constraints (ranges, regexes, etc.)
• replace pydantic or attrs

If you want parsing + validation + coercion → use Pydantic. If you want “is this already the right shape?” → use this.

---

Comparison

Tool	Purpose
isinstance	Runtime type only
typing	Static type checking
pydantic	Parsing + validation + coercion
py-typecheck	Runtime structural check only

---

Development status

• Fully typed (basedpyright strict)
• Tested against Python 3.10 – 3.13
• CI + lint + coverage
• Small surface area, easy to audit

---

License MIT