grandiso hints doesn't check _is_edge_attr_match and _is_node_attr_match.

[khoale88]

As hints are given, grandiso skip _is_edge_attr_match and _is_edge_attr_match. In the following test case, the result should be either
x and for A and B or None and None for A and B, not x and z. This is due to the relation label [r:XY].

def test_hint():
    host = nx.DiGraph()
    host.add_node("x")
    host.add_node("y")
    host.add_node("z")
    host.add_edge("x", "y", __labels__={"XY"}, name="XY")
    host.add_edge("x", "z", __labels__={"XZ"}, name="XZ")

    qry = """
    MATCH (A) -[r:XY]-> (B)
    RETURN ID(A), ID(B)
    """

    res = GrandCypher(host).run(qry, hints=[{"A": "x", "B": "z"}])
    assert res == {'ID(A)': ['x'], 'ID(B)': ['z']}

Is this the design in grandiso?

[khoale88]

This is partially solved in the subquery commit.

There is another case, if no edge between hint nodes, but the motif need an edge there.

def test_hint_weird(self):
        host = nx.DiGraph()
        host.add_node("x", name="x")
        host.add_node("y", name="y")

        qry = """
        MATCH (A) --> (B)
        RETURN ID(A), ID(B)
        """
        gc = GrandCypher(host)
        gc._doublecheck_hint_result = True
        res = gc.run(qry, hints=[{"A": "x", "B": "y"}])
        assert res == {'ID(A)': ["x"], 'ID(B)': ["y"]}

And this is the error

    def _aggregate_edge_labels(edges: Dict) -> Dict:
        """
        Aggregate '__labels__' attributes from edges into a single set.
        """
        aggregated = {"__labels__": set()}
        for edge_id, attrs in edges.items():
>           if "__labels__" in attrs and attrs["__labels__"]:
E           TypeError: argument of type 'NoneType' is not iterable

[khoale88]

I solved the second case as well, let's see if we have other corner cases

[j6k4m8]

My guiding principle for hints is that if the user specifies a hint, to assume that those vertices have already been matched and no longer need to be checked at all. But that of course falls apart in the two circumstances you mentioned! I will check out the behavior in the subquery PR, thank you!!!