from deepeval.metrics import ConversationCompletenessMetric
from deepeval.simulator import ConversationSimulator
from deepeval.dataset import ConversationalGolden
from deepeval.test_case import ConversationalTestCase
from deepeval import assert_test
import pytest

from agent_system.agent_handler import model_callback


knowledge_excel = ConversationalGolden(
    scenario="User wants to know the formula for an if statement in excel.",
    expected_outcome="The formula for an if statement in excel is briefly explained.",
    user_description="User is a Excel beginner",
    multimodal=False,
)
knowledge_dvelop = ConversationalGolden(
    scenario="User wants to know how to digitally sign documents.",
    expected_outcome="The process for digitally signing documents is briefly explained.",
    user_description="User is a digital beginner",
    multimodal=False,
)
simulator = ConversationSimulator(model_callback=model_callback, simulator_model="gpt-5.1")  # type: ignore
conv_test_cases = simulator.simulate(conversational_goldens=[knowledge_excel, knowledge_dvelop], max_user_simulations=5)


@pytest.mark.parametrize("test_case", conv_test_cases)
def test_knowledge_questions(test_case: ConversationalTestCase):
    assert_test(test_case=test_case, metrics=[ConversationCompletenessMetric()])