tests/test_dataset.py

"""The synthetic training dataset (training/make_dataset.py) stays well-formed."""
from server.schema import ActionPlan
from training.make_dataset import SEEDS, to_record


def test_enough_seeds():
    assert len(SEEDS) >= 20


def test_seed_shape_and_validity():
    for now, thread, plan in SEEDS:
        assert isinstance(now, str) and isinstance(thread, str)
        # every target plan must validate against the serving schema
        ActionPlan(**plan)


def test_to_record_builds_chat_triple():
    now, thread, plan = SEEDS[0]
    rec = to_record(now, thread, plan)
    roles = [m["role"] for m in rec["messages"]]
    assert roles == ["system", "user", "assistant"]


def test_covers_hard_cases():
    """Sanity that the expansion includes empties, conflicts, clarifications, images."""
    plans = [p for _, _, p in SEEDS]
    assert any(p["events"] == [] for p in plans)            # no-event chitchat
    assert any(p["conflicts"] for p in plans)               # a conflict case
    assert any(p["proposed_times"] for p in plans)          # alternatives offered
    assert any(p.get("needs_clarification") for p in plans)  # ambiguity
    threads = [t for _, t, _ in SEEDS]
    assert any("[image:" in t for t in threads)             # image-described