from goodnews.briefs import _select_diverse


def row(id, source, topic):
    return {"id": id, "source_name": source, "topic": topic}


def test_prefers_distinct_sources_best_first():
    rows = [
        row(1, "A", "science"),
        row(2, "A", "science"),   # same source as #1 — skipped while others remain
        row(3, "B", "community"),
        row(4, "C", "environment"),
    ]
    assert [r["id"] for r in _select_diverse(rows, limit=3)] == [1, 3, 4]


def test_at_most_one_health_when_alternatives_exist():
    rows = [
        row(1, "A", "health"), row(2, "B", "health"),
        row(3, "C", "science"), row(4, "D", "community"),
        row(5, "E", "animals"), row(6, "F", "environment"),
    ]
    topics = [r["topic"] for r in _select_diverse(rows, limit=5)]
    assert len(topics) == 5
    assert topics.count("health") == 1


def test_science_plus_health_capped_at_two():
    rows = [
        row(1, "A", "science"), row(2, "B", "science"), row(3, "C", "science"),
        row(4, "D", "health"), row(5, "E", "community"),
        row(6, "F", "animals"), row(7, "G", "culture"),
    ]
    topics = [r["topic"] for r in _select_diverse(rows, limit=5)]
    assert len(topics) == 5
    assert topics.count("science") + topics.count("health") <= 2
    # …which means the rest are the gentler lanes
    assert sum(t in {"community", "animals", "culture", "environment"} for t in topics) >= 3


def test_relaxes_caps_to_fill_on_thin_days():
    rows = [row(i, f"S{i}", "science") for i in range(1, 6)]  # only science available
    assert len(_select_diverse(rows, limit=5)) == 5


def test_backfills_repeating_source_when_needed():
    rows = [row(1, "A", "science"), row(2, "A", "science"), row(3, "A", "science")]
    assert len(_select_diverse(rows, limit=2)) == 2


def test_never_exceeds_limit():
    rows = [row(i, f"S{i}", "science") for i in range(20)]
    assert len(_select_diverse(rows, limit=5)) == 5