Testing

This section describes testing practices for applications based on Argenta. Examples are based on the actual public API.

Unit Testing Handlers

Handlers in Argenta are regular functions. They are convenient to test as pure functions without starting the entire application cycle. unittest or pytest are recommended.

Usage example:

import io
from contextlib import redirect_stdout

from argenta import Router, Command, Response
from argenta.command import InputCommand


router = Router(title="Demo")


@router.command(Command("PING", description="Ping command"))
def ping(response: Response):
    print("PONG")


def test_ping_prints_pong():
    # Call handler
    with redirect_stdout(io.StringIO()) as stdout:
        router.finds_appropriate_handler(InputCommand.parse("PING"))
    assert "PONG" in stdout.getvalue()

---

Testing with Dependency Injection (DI)

If a handler needs dependencies, use dishka and Argenta integration:

Usage example:

import io
from contextlib import redirect_stdout

from argenta.command import InputCommand
from dishka import Provider, make_container, Scope

from argenta import Router, Response
from argenta.di.integration import setup_dishka, FromDishka


class Service:
    def hello(self) -> str:
        return "world"


def get_service() -> Service:
    return Service()


router = Router(title="DI")


@router.command("HELLO")
def hello(response: Response, service: FromDishka[Service]) -> None:
    print(f"hello {service.hello()}")


class _FakeApp:
    # Minimal stub for setup_dishka; app object is not used in unit tests
    registered_routers = [router]


def test_hello_uses_service():
    provider = Provider(scope=Scope.APP)
    provider.provide(get_service)

    container = make_container(provider)
    setup_dishka(app=_FakeApp(), container=container, auto_inject=True)

    # Call handler
    with redirect_stdout(io.StringIO()) as stdout:
        router.finds_appropriate_handler(InputCommand.parse("HELLO"))

    assert "hello world" in stdout.getvalue()

---

Integration Testing of the Application

For higher-level tests, assemble App and Router and call handlers through command parsing, bypassing the infinite input loop. This provides behavior close to reality without the need to simulate stdin.

Usage example:

import io
from contextlib import redirect_stdout

from argenta import App, Router, Command, Response
from argenta.command import InputCommand


def test_simple_app() -> None:
    app = App(override_system_messages=True, repeat_command_groups_printing=False)
    router = Router(title="App")

    @router.command(Command("HELP", description="Show help"))
    def help_cmd(response: Response):
        print("Available commands: HELP")

    app.include_router(router)

    with redirect_stdout(io.StringIO()) as stdout:
        router.finds_appropriate_handler(InputCommand.parse("HELP"))

    assert "Available commands:" in stdout.getvalue()

---

E2E Testing of the Loop

Full execution of the start_polling loop can be covered through a subprocess with passing strings to stdin. This is heavier and usually not required. If still necessary, an example is below.

Danger

Important: Always pass the exit command string trigger as the last element in the side_effects list when patching input.

Otherwise, the application under test will wait for the next command input and will not be able to terminate correctly.

Usage example:

import sys
from unittest.mock import patch
import pytest
from pytest import CaptureFixture

from argenta import App, Orchestrator, Router, Command, Response


@pytest.fixture(autouse=True)
def patched_argv():
    with patch.object(sys, "argv", ["program.py"]):
        yield


def test_input_incorrect_command(capsys: CaptureFixture[str]):
    router = Router()
    orchestrator = Orchestrator()

    @router.command(Command("test"))
    def test(response: Response) -> None:
        print("test command")

    app = App(override_system_messages=True, printer=print)
    app.include_router(router)
    app.set_unknown_command_handler(lambda command: print(f"Unknown command: {command.trigger}"))

    with patch("builtins.input", side_effect=["help", "q"]):
        orchestrator.start_polling(app)

    output = capsys.readouterr().out
    assert "\nUnknown command: help\n" in output

---

Testing Tips

1. Isolate tests: Each test should be independent of others.

2. Mocks for external integrations: Replace databases, HTTP clients, etc. with stubs and dishka providers.

3. Cover error scenarios: Incorrect flags, unknown commands, empty input.

4. Minimize formatting dependency: Compare key output fragments, not the entire block.

5. Measure coverage: Use pytest-cov.