from goodnews.briefs import _select_diverse


def row(id, source, topic):
    # _select_diverse reads id, source_name, topic; plain dicts support [] access.
    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", "health"),
        row(4, "C", "environment"),
    ]
    assert [r["id"] for r in _select_diverse(rows, limit=3)] == [1, 3, 4]


def test_caps_a_topic_when_alternatives_exist():
    rows = [
        row(1, "A", "science"), row(2, "B", "science"),
        row(3, "C", "science"), row(4, "D", "science"),
        row(5, "E", "community"), row(6, "F", "animals"), row(7, "G", "culture"),
    ]
    selected = _select_diverse(rows, limit=5, max_per_topic=2)
    topics = [r["topic"] for r in selected]
    assert len(selected) == 5
    assert topics.count("science") == 2  # capped, even though 4 were available
    assert {"community", "animals", "culture"} <= set(topics)


def test_relaxes_cap_when_only_one_topic_available():
    rows = [row(i, f"S{i}", "science") for i in range(1, 6)]
    selected = _select_diverse(rows, limit=5)
    assert len(selected) == 5  # all science: cap relaxed because nothing else exists


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