# Trouve-KA — tests de la recherche hybride (RRF, corps kNN)
# Author: Simon-Pierre Boucher
# Contact: contact@spboucher.ai

from trouveka.ranking import build_knn_body, rrf_fuse


def _hit(_id: str, **extra):
    return {"_id": _id, "_source": {"url": f"https://x.qc/{_id}"}, **extra}


def test_rrf_doc_present_dans_les_deux_gagne():
    lexical = [_hit("a"), _hit("b"), _hit("c")]
    knn = [_hit("b"), _hit("d")]
    fused = rrf_fuse(lexical, knn)
    assert fused[0]["_id"] == "b"  # présent dans les deux listes
    assert {h["_id"] for h in fused} == {"a", "b", "c", "d"}


def test_rrf_conserve_le_highlight_lexical():
    lexical = [_hit("a", highlight={"body": ["<em>x</em>"]})]
    knn = [_hit("a")]
    fused = rrf_fuse(lexical, knn)
    assert "highlight" in fused[0]


def test_rrf_listes_vides():
    assert rrf_fuse([], []) == []
    assert [h["_id"] for h in rrf_fuse([_hit("a")], [])] == ["a"]


def test_knn_body_filtres_appliques():
    body = build_knn_body([0.1] * 4, filters=[{"term": {"language": "fr"}}], size=30)
    knn = body["query"]["knn"]["embedding"]
    assert knn["k"] == 30
    assert knn["filter"]["bool"]["filter"][0]["term"]["language"] == "fr"


def test_knn_body_sans_filtre():
    body = build_knn_body([0.1] * 4, filters=[], size=10)
    assert "filter" not in body["query"]["knn"]["embedding"]