def test_hyperedges_direct_manual_shaping():
h1 = cugraph.hypergraph(
hyper_df,
direct=True,
EDGES={"aa": ["cc"], "cc": ["cc"]},
)
assert len(h1["edges"]) == 6
h2 = cugraph.hypergraph(
hyper_df,
direct=True,
EDGES={"aa": ["cc", "bb", "aa"], "cc": ["cc"]},
)
assert len(h2["edges"]) == 12
def test_hyper_to_pa_all_direct():
hg = cugraph.hypergraph(simple_df, ["id", "a1", "🙈"], direct=True)
nodes_arr = hg["graph"].nodes().to_arrow()
assert len(hg["graph"].nodes()) == 9
assert len(nodes_arr) == 9
edges_err = hg["graph"].edges().to_arrow()
assert len(hg["graph"].edges()) == 9
assert len(edges_err) == 9
def test_skip_na_hyperedge():
nans_df = cudf.DataFrame.from_pandas(
pd.DataFrame({
"x": ["a", "b", "c"],
"y": ["aa", None, "cc"]
}))
expected_hits = ["a", "b", "c", "aa", "cc"]
skip_attr_h_edges = cugraph.hypergraph(nans_df,
drop_edge_attrs=True)["edges"]
assert_eq(len(skip_attr_h_edges), len(expected_hits))
default_h_edges = cugraph.hypergraph(nans_df)["edges"]
assert_eq(len(default_h_edges), len(expected_hits))
def test_complex_df():
complex_df = pd.DataFrame({
"src": [0, 1, 2, 3],
"dst": [1, 2, 3, 0],
"colors": [1, 1, 2, 2],
"bool": [True, False, True, True],
"char": ["a", "b", "c", "d"],
"str": ["a", "b", "c", "d"],
"ustr": [u"a", u"b", u"c", u"d"],
"emoji": ["😋", "😋😋", "😋", "😋"],
"int": [0, 1, 2, 3],
"num": [0.5, 1.5, 2.5, 3.5],
"date_str": [
"2018-01-01 00:00:00",
"2018-01-02 00:00:00",
"2018-01-03 00:00:00",
"2018-01-05 00:00:00",
],
"date": [
dt.datetime(2018, 1, 1),
dt.datetime(2018, 1, 1),
dt.datetime(2018, 1, 1),
dt.datetime(2018, 1, 1),
],
"time": [
pd.Timestamp("2018-01-05"),
pd.Timestamp("2018-01-05"),
pd.Timestamp("2018-01-05"),
pd.Timestamp("2018-01-05"),
],
})
for c in complex_df.columns:
try:
complex_df[c + "_cat"] = complex_df[c].astype("category")
except Exception:
# lists aren't categorical
# print('could not make categorical', c)
pass
complex_df = cudf.DataFrame.from_pandas(complex_df)
cugraph.hypergraph(complex_df)
def test_drop_na_direct():
df = cudf.DataFrame.from_pandas(
pd.DataFrame({"a": ["a", None, "a"], "i": [1, 1, None]})
)
hg = cugraph.hypergraph(df, dropna=True, direct=True)
assert len(hg["graph"].nodes()) == 2
assert len(hg["graph"].edges()) == 1
def test_drop_na_hyper():
df = cudf.DataFrame.from_pandas(
pd.DataFrame({"a": ["a", None, "c"], "i": [1, 2, None]})
)
hg = cugraph.hypergraph(df, dropna=True)
assert len(hg["graph"].nodes()) == 7
assert len(hg["graph"].edges()) == 4
def test_hyper_to_pa_mixed():
df = cudf.DataFrame.from_pandas(pd.DataFrame({
"x": ["a", "b", "c"],
"y": [1, 2, 3]
}))
hg = cugraph.hypergraph(df)
nodes_arr = hg["graph"].nodes().to_arrow()
assert len(nodes_arr) == 9
edges_err = hg["graph"].edges().to_arrow()
assert len(edges_err) == 6
def test_skip_drop_na_direct():
df = cudf.DataFrame.from_pandas(pd.DataFrame({
"a": ["a", None, "b"],
"b": ["a", "b", "c"],
"c": [1, 2, 3]
}))
hg = cugraph.hypergraph(df, SKIP=["c"], dropna=True, direct=True)
assert len(hg["graph"].nodes()) == 4
assert len(hg["graph"].edges()) == 2
def test_skip_hyper():
df = cudf.DataFrame.from_pandas(pd.DataFrame({
"a": ["a", None, "b"],
"b": ["a", "b", "c"],
"c": [1, 2, 3]
}))
hg = cugraph.hypergraph(df, SKIP=["c"], dropna=False)
assert len(hg["graph"].nodes()) == 9
assert len(hg["graph"].edges()) == 6
def test_hyperedges_direct_categories():
h = cugraph.hypergraph(
hyper_df,
direct=True,
categories={
"aa": "N",
"bb": "N",
"cc": "N",
},
)
assert_eq(len(h["edges"]), 9)
assert_eq(len(h["nodes"]), 6)
def test_hyper_to_pa_na():
df = cudf.DataFrame.from_pandas(pd.DataFrame({
"x": ["a", None, "c"],
"y": [1, 2, None]
}))
hg = cugraph.hypergraph(df, dropna=False)
print(hg["graph"].nodes())
nodes_arr = hg["graph"].nodes().to_arrow()
assert len(hg["graph"].nodes()) == 9
assert len(nodes_arr) == 9
edges_err = hg["graph"].edges().to_arrow()
assert len(hg["graph"].edges()) == 6
assert len(edges_err) == 6
def test_drop_edge_attrs(categorical_metadata):
h = cugraph.hypergraph(simple_df,
columns=["id", "a1", "🙈"],
drop_edge_attrs=True,
categorical_metadata=categorical_metadata)
assert len(h.keys()) == len(
["entities", "nodes", "edges", "events", "graph"])
edges = cudf.DataFrame.from_pandas(pd.DataFrame({
"event_id": [
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
],
"edge_type": [
"a1", "a1", "a1", "id", "id", "id", "🙈", "🙈", "🙈"
],
"attrib_id": [
"a1::1",
"a1::2",
"a1::3",
"id::a",
"id::b",
"id::c",
"🙈::æski ēˈmōjē",
"🙈::😋",
"🙈::s",
],
}))
if categorical_metadata:
edges = edges.astype({"edge_type": "category"})
assert_frame_equal(edges, h["edges"], check_dtype=False)
for (k, v) in [
("entities", 9), ("nodes", 12), ("edges", 9), ("events", 3)
]:
assert len(h[k]) == v
def make_and_shape_hypergraph(df, **kwargs):
hyper = cugraph.hypergraph(df, **kwargs)
del hyper["events"]
del hyper["entities"]
SOURCE = kwargs.get("SOURCE", "src")
TARGET = kwargs.get("TARGET", "dst")
NODEID = kwargs.get("NODEID", "node_id")
EVENTID = kwargs.get("EVENTID", "event_id")
CATEGORY = kwargs.get("CATEGORY", "category")
nodes = hyper["nodes"][[NODEID, CATEGORY]]
edges = hyper["edges"][[SOURCE, TARGET]]
# Create graph
graph, nodes, edges = from_cudf_edgelist(edges, SOURCE, TARGET)
nodes["name"] = nodes["node"]
# Add vis components
nodes = annotate_nodes(graph, nodes, edges)
edges = annotate_edges(graph, nodes, edges)
return graph, nodes, edges
def test_drop_edge_attrs_direct(categorical_metadata):
h = cugraph.hypergraph(
simple_df,
["id", "a1", "🙈"],
direct=True,
drop_edge_attrs=True,
EDGES={
"id": ["a1"],
"a1": ["🙈"]
},
categorical_metadata=categorical_metadata,
)
assert_eq(len(h.keys()),
len(["entities", "nodes", "edges", "events", "graph"]))
edges = cudf.DataFrame.from_pandas(
pd.DataFrame({
"event_id": [
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
],
"edge_type":
["a1::🙈", "a1::🙈", "a1::🙈", "id::a1", "id::a1", "id::a1"],
"src": ["a1::1", "a1::2", "a1::3", "id::a", "id::b", "id::c"],
"dst":
["🙈::æski ēˈmōjē", "🙈::😋", "🙈::s", "a1::1", "a1::2", "a1::3"],
}))
if categorical_metadata:
edges = edges.astype({"edge_type": "category"})
assert_eq(edges, h["edges"])
for (k, v) in [("entities", 9), ("nodes", 9), ("edges", 6), ("events", 0)]:
assert_eq(len(h[k]), v)
def test_hyperedges_direct():
h = cugraph.hypergraph(hyper_df, direct=True)
assert_eq(len(h["edges"]), 9)
assert_eq(len(h["nodes"]), 9)
def test_hyperedges(categorical_metadata):
h = cugraph.hypergraph(simple_df,
categorical_metadata=categorical_metadata)
assert_eq(len(h.keys()),
len(["entities", "nodes", "edges", "events", "graph"]))
edges = pd.DataFrame({
"event_id": [
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
"event_id::0",
"event_id::1",
"event_id::2",
],
"edge_type": [
"a1",
"a1",
"a1",
"a2",
"a2",
"a2",
"id",
"id",
"id",
"🙈",
"🙈",
"🙈",
],
"attrib_id": [
"a1::1",
"a1::2",
"a1::3",
"a2::red",
"a2::blue",
"a2::green",
"id::a",
"id::b",
"id::c",
"🙈::æski ēˈmōjē",
"🙈::😋",
"🙈::s",
],
"id": ["a", "b", "c"] * 4,
"a1": [1, 2, 3] * 4,
"a2": ["red", "blue", "green"] * 4,
"🙈": ["æski ēˈmōjē", "😋", "s"] * 4,
})
if categorical_metadata:
edges = edges.astype({"edge_type": "category"})
assert_eq(edges, h["edges"])
for (k, v) in [("entities", 12), ("nodes", 15), ("edges", 12),
("events", 3)]:
assert_eq(len(h[k]), v)
