Python Type Refinement with isinstance and TypeIs

Refine union types using built-in isinstance checks and TypeIs predicates.

What It Does and Why It Matters

isinstance(value, SomeType) narrows a union in the true branch. A TypeIs-returning function allows application-specific type predicates to tell a checker the precise resulting type in both true and false branches, subject to soundness constraints. It must return a real boolean and accurately describe the object being tested.

TypeIs was added in Python 3.13; static narrowing still depends on checker support.

Annotations guide static tools; by default they do not validate external input or change the return value at runtime.

Step-by-Step Example

Start with the complete example below. Read the inputs and predict what the final print statement will show before you run it.

from typing import TypeIs

def is_text(value: str | int) -> TypeIs[str]:
    return isinstance(value, str)

def describe(value: str | int) -> str:
    if is_text(value):
        return value.upper()
    return str(value + 1)
print(describe("hello"))
print(describe(3))

How the Code Works

The imports make these standard-library tools available: typing.
The definition of is_text, describe creates reusable behavior; its body runs only when called or instantiated as appropriate.
The print calls display the results. Compare these results with the method’s documented return type and the input used.
The predicate returns a runtime Boolean. The annotation tells a static checker what type it may assume within the checked branch.

Expected output:

HELLO
4

Each displayed line comes from the corresponding calculation or check. The output helps verify the return values before you adapt the example to your own input.

What to Notice

Pay attention to the accepted input types, the returned value and the boundary behavior. Predict what changes if you replace the example input with an empty, zero, negative or missing value, where that value is meaningful for this operation.

If the task uses a library function, inspect its documented exceptions and return type before reusing it inside a larger function.

Where You Would Use This

This helps maintain larger projects where functions and data shapes cross module boundaries. Check the code with a static analyzer, and still validate data received from files, users or remote APIs. Annotation constructs do not normally enforce types automatically at runtime.

Practice Exercise

Change the union to str | int | None and handle all three cases.

Before running your edited example, write down the result you expect. Then test at least one valid boundary or deliberately invalid value, using appropriate exception handling where necessary.

Reference: official Python documentation for the complete API contract and version notes.

Common Beginner Mistakes

Writing a TypeIs predicate that returns True for values outside the declared target type.
Overlooking the documented condition: it must return a real boolean and accurately describe the object being tested.

Key Takeaways

Refine union types using built-in isinstance checks and TypeIs predicates.
isinstance(value, SomeType) narrows a union in the true branch.
Change the union to str | int | None and handle all three cases.
Annotations guide static tools; by default they do not validate external input or change the return value at runtime.