pymocks

Testing utility decorators and context managers for mocking functions, variables, and HTTP endpoints in Python.

Works with both synchronous and asynchronous test functions. Sync/async is detected automatically. All mocking utilities can be used as decorators or context managers.

Installation

pip install pymocks

Or with uv:

uv add pymocks

Usage

Mocking Functions and Variables

Use Mock with with_mock to monkeypatch module attributes for the duration of a test. Accepts one or more Mock objects. Works as a decorator or context manager:

import my_module
from pymocks import Mock, with_mock


# Mock a function — the replacement must have the same signature
def fake_function(x: int, y: str) -> bool:
    return True


mock = Mock(
    module_where_used=my_module,
    current_value=my_module.some_function,
    new_value=fake_function,
)


@with_mock(mock)
def test_with_mocked_function():
    result = my_module.some_function(1, "a")
    assert result is True


# Mock a variable — the replacement must have the same type
var_mock = Mock(
    module_where_used=my_module,
    current_value=my_module.API_URL,
    new_value="https://mock.example.com",
)


@with_mock(var_mock)
def test_with_mocked_variable():
    assert my_module.API_URL == "https://mock.example.com"

Pass multiple mocks to apply them all at once — they are all reverted together when the scope exits:

@with_mock(mock, var_mock)
def test_with_multiple_mocks():
    result = my_module.some_function(1, "a")
    assert result is True
    assert my_module.API_URL == "https://mock.example.com"

The same works for async tests:

@with_mock(mock)
async def test_async_with_mock():
    result = my_module.some_function(1, "a")
    assert result is True

Or use it as a context manager for more flexible scoping:

def test_with_context_manager():
    with with_mock(mock, var_mock):
        result = my_module.some_function(1, "a")
        assert result is True
    # mocks are reverted here


async def test_async_with_context_manager():
    async with with_mock(mock):
        result = my_module.some_function(1, "a")
        assert result is True

Mocking Classes

Use Mock to replace a class with a subclass. The replacement must be a subclass of the original:

import my_module
from pymocks import Mock, with_mock


class FakeService(my_module.Service):
    def fetch(self) -> str:
        return "fake data"


mock = Mock(
    module_where_used=my_module,
    current_value=my_module.Service,
    new_value=FakeService,
)


@with_mock(mock)
def test_with_mocked_class():
    svc = my_module.Service()
    assert svc.fetch() == "fake data"

Replacing a class with an unrelated class raises TypeError:

class Unrelated:
    pass


# Raises TypeError — Unrelated is not a subclass of Service
Mock(
    module_where_used=my_module,
    current_value=my_module.Service,
    new_value=Unrelated,
)

Signature and Type Validation

Mock validates compatibility between current_value and new_value at construction time:

• Callables: signatures must match exactly (parameter count, names, kinds, annotations, and return annotation)
• Classes: new_value must be a subclass of current_value
• Non-callables: type(current_value) must be the same as type(new_value)
• Mixed: replacing a callable with a non-callable (or vice versa) raises TypeError

def original(x: int) -> str:
    return str(x)


# Signature mismatch — raises TypeError immediately
Mock(
    module_where_used=my_module,
    current_value=original,
    new_value=lambda x: str(x),  # missing annotations
)

# Type mismatch — raises TypeError immediately
Mock(
    module_where_used=my_module,
    current_value="a string",
    new_value=42,
)

Mocking HTTP Endpoints

Use MockAiohttpEndpoint with with_aiohttp_endpoints to mock HTTP calls via aioresponses. Works as a decorator or context manager:

import aiohttp
from pymocks import MockAiohttpEndpoint, with_aiohttp_endpoints

endpoints = (
    MockAiohttpEndpoint(
        url="https://api.example.com/users",
        method="GET",
        json_response={"users": [{"id": 1, "name": "Alice"}]},
    ),
    MockAiohttpEndpoint(
        url="https://api.example.com/users",
        method="POST",
        json_response={"id": 2, "name": "Bob"},
    ),
)


# As a decorator
@with_aiohttp_endpoints(endpoints)
async def test_api_calls():
    async with aiohttp.ClientSession() as session:
        async with session.get("https://api.example.com/users") as resp:
            data = await resp.json()
            assert len(data["users"]) == 1


# As a context manager
async def test_api_calls_ctx():
    async with with_aiohttp_endpoints(endpoints):
        async with aiohttp.ClientSession() as session:
            async with session.get("https://api.example.com/users") as resp:
                data = await resp.json()
                assert len(data["users"]) == 1

Asserting Request Data

Use assert_request on a MockAiohttpEndpoint to inspect the URL, headers, body, and params sent by the code under test. The callback receives a single MockedAiohttpRequest — mirroring the assert_request(request) shape used by the httpx backend:

from pymocks import MockAiohttpEndpoint, MockedAiohttpRequest, with_aiohttp_endpoints

def check_request(request: MockedAiohttpRequest) -> None:
    assert (request.headers or {})["Authorization"] == "Bearer token123"
    assert request.json == {"name": "alice"}

endpoint = MockAiohttpEndpoint(
    url="https://api.example.com/users",
    method="POST",
    json_response={"id": 1},
    assert_request=check_request,
)


@with_aiohttp_endpoints((endpoint,))
async def test_post_sends_correct_data():
    async with aiohttp.ClientSession() as session:
        await session.post(
            "https://api.example.com/users",
            json={"name": "alice"},
            headers={"Authorization": "Bearer token123"},
        )

MockedAiohttpRequest is a frozen dataclass with the following fields:

When assert_request is None (the default), the endpoint returns the configured response without any assertion callback.

Mocking httpx Endpoints

Use MockHttpxEndpoint with with_httpx_endpoints to mock httpx calls (both httpx.Client and httpx.AsyncClient) via pytest-httpx. Works as a decorator or context manager, sync or async:

import httpx
from pymocks import MockHttpxEndpoint, with_httpx_endpoints

endpoints = (
    MockHttpxEndpoint(
        url="https://api.example.com/users",
        method="GET",
        json_response={"users": [{"id": 1, "name": "Alice"}]},
    ),
    MockHttpxEndpoint(
        url="https://api.example.com/users",
        method="POST",
        json_response={"id": 2, "name": "Bob"},
        status_code=201,
    ),
)


# Sync decorator
@with_httpx_endpoints(endpoints)
def test_api_calls():
    with httpx.Client() as client:
        data = client.get("https://api.example.com/users").json()
        assert len(data["users"]) == 1


# Async context manager
async def test_api_calls_ctx():
    async with (
        with_httpx_endpoints(endpoints),
        httpx.AsyncClient() as client,
    ):
        response = await client.post(
            "https://api.example.com/users", json={"name": "Bob"},
        )
        assert response.status_code == 201

Supported methods: GET, POST, PUT, DELETE, PATCH, HEAD, OPTIONS.

The assert_request callback for an httpx endpoint receives the raw httpx.Request, so you can read headers, URL, and body directly:

import httpx
from pymocks import MockHttpxEndpoint, with_httpx_endpoints

def check_request(request: httpx.Request) -> None:
    assert request.headers["Authorization"] == "Bearer token123"
    assert request.read() == b'{"name":"alice"}'

endpoint = MockHttpxEndpoint(
    url="https://api.example.com/users",
    method="POST",
    json_response={"id": 1},
    assert_request=check_request,
)


@with_httpx_endpoints((endpoint,))
def test_post_sends_correct_data():
    with httpx.Client() as client:
        client.post(
            "https://api.example.com/users",
            json={"name": "alice"},
            headers={"Authorization": "Bearer token123"},
        )

Every registered endpoint must be called at least once or the block raises an AssertionError on exit — this matches pytest-httpx's default assertion behavior.

API Reference

Mock[T_mocked]

A dataclass that defines a monkeypatch specification. Validates compatibility on construction.

MockAiohttpEndpoint

A frozen dataclass defining an HTTP endpoint mock for aiohttp (via aioresponses).

MockHttpxEndpoint

A frozen dataclass defining an HTTP endpoint mock for httpx (via pytest-httpx).

MockedAiohttpRequest

A frozen dataclass passed to MockAiohttpEndpoint.assert_request. Wraps the per-request data aioresponses supplies into a single argument, matching the assert_request(request) shape of the httpx backend.

with_mock(*mocks) / with_aiohttp_endpoints(endpoints) / with_httpx_endpoints(endpoints)

All three can be used as decorators or context managers (sync and async):

# Decorator — single mock
@with_mock(mock)
def test_decorated(): ...

# Decorator — multiple mocks
@with_mock(mock, var_mock)
def test_decorated(): ...

# Sync context manager
with with_mock(mock, var_mock):
    ...

# Async context manager
async with with_mock(mock):
    ...

When used as decorators, sync/async is detected automatically.

Requirements

• Python >= 3.12
• pytest
• aioresponses
• pytest-httpx (for httpx endpoint mocking)

License

MIT