def test_error_on_operation_on_graph(graphscope_session):
g = Graph(graphscope_session)
with pytest.raises(RuntimeError, match="Empty graph"):
g.project_to_simple(v_label=0, v_prop=0, e_label=0, e_prop=0)
with pytest.raises(RuntimeError):
property_sssp(g, src=6)
def test_properties_omitted_loader(graphscope_session, student_group_e,
student_v):
graph = Graph(graphscope_session, generate_eid=False)
graph = graph.add_vertices(student_v, "student", [], "student_id")
graph = graph.add_edges(student_group_e, "group", [])
assert len(graph.schema.vertex_properties[0]) == 4
assert len(graph.schema.edge_properties[0]) == 2
def test_error_on_ambigious_default_label(graphscope_session, student_group_e,
student_v, teacher_v):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student")
graph = graph.add_vertices(teacher_v, "teacher")
with pytest.raises(AssertionError, match="ambiguous vertex label"):
graph = graph.add_edges(student_group_e, "group")
def test_v_property_omitted_form_loader(graphscope_session, student_group_e,
student_v):
# properties for v can be omit, all columns will be load,
# the first one used as vid by # default. default vlabel would be '_';
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student")
graph = graph.add_edges(
student_group_e,
"group",
["group_id", "member_size"],
src_field="leader_student_id",
dst_field="member_student_id",
)
assert graph.loaded()
def test_error_on_non_default_and_non_existing_v_label(graphscope_session,
student_group_e,
student_v):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student")
with pytest.raises(ValueError,
match="src label or dst_label not existed in graph"):
graph = graph.add_edges(student_group_e,
"group",
src_label="v",
dst_label="v")
with pytest.raises(ValueError,
match="must be both specified or either unspecified"):
graph = graph.add_edges(student_group_e, src_label="v")
def test_error_on_src_dst_refer_to_same_col(graphscope_session,
student_group_e):
graph = Graph(graphscope_session)
with pytest.raises(AssertionError, match="cannot refer to the same field"):
graph = graph.add_edges(
student_group_e,
"group",
src_field="leader_student_id",
dst_field="leader_student_id",
)
with pytest.raises(AssertionError, match="cannot refer to the same field"):
graph = graph.add_edges(student_group_e,
"group",
src_field=0,
dst_field=0)
def test_multiple_add_vertices_edges(graphscope_session):
prefix = os.path.expandvars("${GS_TEST_DIR}/modern_graph")
graph = Graph(graphscope_session)
graph = graph.add_vertices(Loader(f"{prefix}/person.csv", delimiter="|"),
"person")
graph = graph.add_edges(Loader(f"{prefix}/knows.csv", delimiter="|"),
"knows")
graph = graph.add_vertices(Loader(f"{prefix}/software.csv", delimiter="|"),
"software")
graph = graph.add_edges(
Loader(f"{prefix}/created.csv", delimiter="|"),
"created",
src_label="person",
dst_label="software",
)
assert graph.schema.vertex_labels == ["person", "software"]
assert graph.schema.edge_labels == ["created", "knows"]
graph = graph.add_vertices(Loader(f"{prefix}/person.csv", delimiter="|"),
"person2")
graph = graph.add_edges(
Loader(f"{prefix}/knows.csv", delimiter="|"),
"knows2",
src_label="person2",
dst_label="person2",
)
graph = graph.add_vertices(Loader(f"{prefix}/software.csv", delimiter="|"),
"software2")
graph = graph.add_edges(
Loader(f"{prefix}/created.csv", delimiter="|"),
"created2",
src_label="person2",
dst_label="software2",
)
assert sorted(graph.schema.vertex_labels) == [
"person",
"person2",
"software",
"software2",
]
assert sorted(graph.schema.edge_labels) == [
"created",
"created2",
"knows",
"knows2",
]
def test_load_graph_copy(graphscope_session, arrow_property_graph):
g = arrow_property_graph
g2 = Graph(graphscope_session, g)
assert g.key != g2.key
assert g.vineyard_id != g2.vineyard_id
assert str(g.schema) == str(g2.schema)
assert np.all(g.to_numpy("v:v0.id") == g2.to_numpy("v:v0.id"))
g2.unload()
assert not g2.loaded()
# test load from vineyard's graph
g3 = Graph(graphscope_session, vineyard.ObjectID(g.vineyard_id))
assert g3.loaded()
def test_vid_omitted_form_loader(graphscope_session, student_group_e,
student_v):
# vid can be omit, the first column will be used as vid;
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student",
["name", "lesson_nums", "avg_score"])
graph = graph.add_edges(
student_group_e,
"group",
["group_id", "member_size"],
src_label="student",
dst_label="student",
src_field="leader_student_id",
dst_field="member_student_id",
)
assert graph.loaded()
def test_serialize_roundtrip(gs_session_distributed, p2p_property_dir):
graph = gs_session_distributed.load_from(
edges={
"knows": (
Loader("{}/p2p-31_property_e_0".format(p2p_property_dir),
header_row=True),
["src_label_id", "dst_label_id", "dist"],
("src_id", "person"),
("dst_id", "person"),
),
},
vertices={
"person":
Loader("{}/p2p-31_property_v_0".format(p2p_property_dir),
header_row=True),
},
generate_eid=False,
)
graph.serialize("/tmp/serialize")
new_graph = Graph.deserialize("/tmp/serialize", gs_session_distributed)
pg = new_graph.project_to_simple(0, 0, 0, 2)
ctx = graphscope.sssp(pg, src=6)
ret = (ctx.to_dataframe({
"node": "v.id",
"r": "r"
},
vertex_range={
"end": 6
}).sort_values(by=["node"]).to_numpy(dtype=float))
expect = np.array([[1.0, 260.0], [2.0, 229.0], [3.0, 310.0], [4.0, 256.0],
[5.0, 303.0]])
assert np.all(ret == expect)
def test_serialize_roundtrip(gs_session_distributed, p2p_property_dir):
graph = gs_session_distributed.g(generate_eid=False)
graph = graph.add_vertices(f"{p2p_property_dir}/p2p-31_property_v_0",
"person")
graph = graph.add_edges(
f"{p2p_property_dir}/p2p-31_property_e_0",
label="knows",
src_label="person",
dst_label="person",
)
graph.save_to("/tmp/serialize")
new_graph = Graph.load_from("/tmp/serialize", gs_session_distributed)
pg = new_graph.project_to_simple(0, 0, 0, 2)
ctx = graphscope.sssp(pg, src=6)
ret = (ctx.to_dataframe({
"node": "v.id",
"r": "r"
},
vertex_range={
"end": 6
}).sort_values(by=["node"]).to_numpy(dtype=float))
expect = np.array([[1.0, 260.0], [2.0, 229.0], [3.0, 310.0], [4.0, 256.0],
[5.0, 303.0]])
assert np.all(ret == expect)
def load_ogbn_mag(sess, prefix):
"""Load ogbn_mag graph.
The ogbn-mag dataset is a heterogeneous network composed of a subset of the Microsoft Academic Graph (MAG).
See more details here:
https://ogb.stanford.edu/docs/nodeprop/#ogbn-mag
Args:
sess (:class:`graphscope.Session`): Load graph within the session.
prefix (str): Data directory.
directed (bool, optional): Determine to load a directed or undirected graph.
Defaults to True.
Returns:
:class:`graphscope.Graph`: A Graph object which graph type is ArrowProperty
"""
graph = Graph(sess)
graph = (graph.add_vertices(os.path.join(
prefix, "paper.csv"), "paper").add_vertices(
os.path.join(prefix, "author.csv"), "author").add_vertices(
os.path.join(prefix, "institution.csv"),
"institution").add_vertices(
os.path.join(prefix, "field_of_study.csv"),
"field_of_study").add_edges(
os.path.join(prefix,
"author_affiliated_with_institution.csv"),
"affiliated",
src_label="author",
dst_label="institution",
).add_edges(
os.path.join(prefix,
"paper_has_topic_field_of_study.csv"),
"hasTopic",
src_label="paper",
dst_label="field_of_study",
).add_edges(
os.path.join(prefix, "paper_cites_paper.csv"),
"cites",
src_label="paper",
dst_label="paper",
).add_edges(
os.path.join(prefix, "author_writes_paper.csv"),
"writes",
src_label="author",
dst_label="paper",
))
return graph
def test_complete_form_loader(graphscope_session, student_group_e, student_v):
# a complete form for loading from ev files.
# types are inferred from Loader.
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student",
["name", "lesson_nums", "avg_score"],
"student_id")
graph = graph.add_edges(
student_group_e,
"group",
["group_id", "member_size"],
src_label="student",
dst_label="student",
src_field="leader_student_id",
dst_field="member_student_id",
)
assert graph.loaded()
def test_Load_complex_graph_variants(
graphscope_session,
score_e,
student_group_e_df,
student_v_array,
teacher_v_oss,
lesson_v_mars,
):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v_array, "student",
["name", "lesson_nums", "avg_score"],
"student_id")
graph = graph.add_vertices(teacher_v_oss, "teacher",
["student_num", "score", "email", "tel"])
graph = graph.add_vertices(lesson_v_mars, "lesson")
graph = graph.add_edges(
score_e,
"score",
["score", "score_id"],
src_label="student",
dst_label="lesson",
src_field="studnet_id",
dst_field="subject",
)
graph = graph.add_edges(
student_group_e_df,
"group",
["member_size"],
src_label="student",
dst_label="student",
src_field="leader_studnet_id",
dst_field="member_student_id",
)
assert graph.schema is not None
def test_multi_src_dst_edge_loader(graphscope_session, student_group_e,
teacher_group_e, student_v, teacher_v):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student",
["name", "lesson_nums", "avg_score"],
"student_id")
graph = graph.add_vertices(teacher_v, "teacher",
["student_num", "score", "email", "tel"],
"teacher_id")
graph = graph.add_edges(
student_group_e,
"group",
["group_id", "member_size"],
src_label="student",
dst_label="student",
src_field="leader_student_id",
dst_field="member_student_id",
)
graph = graph.add_edges(
teacher_group_e,
"group",
["group_id", "member_size"],
src_label="teacher",
dst_label="teacher",
src_field="leader_teacher_id",
dst_field="member_teacher_id",
)
assert graph.loaded()
def test_error_on_transform_graph():
g = Graph()
ag = gs.to_appendable(g)
with pytest.raises(AssertionError,
match="expect source graph is immutable"):
agg = gs.to_appendable(ag)
g2 = gs.to_immutable(ag)
with pytest.raises(AssertionError,
match="expect source graph is appendable"):
g3 = gs.to_immutable(g2)
def __call__(self, graph: Graph, *args, **kwargs):
kwargs_extend = dict(app_class=self.java_app_class, **kwargs)
if not hasattr(graph, "graph_type"):
raise InvalidArgumentError("Missing graph_type attribute in graph object.")
if (
self.java_app_type.find("simple") != -1
and graph.graph_type == graph_def_pb2.ARROW_PROPERTY
):
graph = graph._project_to_simple()
app_ = graph.session._wrapper(JavaAppDagNode(graph, self))
return app_(*args, **kwargs_extend)
def test_loader_with_specified_data_type(graphscope_session, student_group_e,
student_v):
graph = Graph(graphscope_session, oid_type="string", generate_eid=False)
graph = graph.add_vertices(
student_v,
"student",
["name", ("lesson_nums", "int"), ("avg_score", "float")],
"student_id",
)
graph = graph.add_edges(student_group_e, "group",
["group_id", ("member_size", "int")])
assert graph.schema.vertex_properties == [{
"name": 21,
"lesson_nums": 11,
"avg_score": 18,
"student_id": 21
}]
assert graph.schema.edge_properties == [{
"group_id": 21,
"member_size": 11
}]
def test_add_vertices_edges(graphscope_session):
prefix = os.path.expandvars("${GS_TEST_DIR}/modern_graph")
graph = Graph(graphscope_session)
graph = graph.add_vertices(Loader(f"{prefix}/person.csv", delimiter="|"),
"person")
graph = graph.add_edges(Loader(f"{prefix}/knows.csv", delimiter="|"),
"knows")
assert graph.schema.vertex_labels == ["person"]
assert graph.schema.edge_labels == ["knows"]
with pytest.raises(ValueError,
match="src label and dst label cannot be None"):
graph = graph.add_edges(Loader(f"{prefix}/knows.csv", delimiter="|"),
"created")
with pytest.raises(ValueError,
match="src label or dst_label not existed in graph"):
graph = graph.add_edges(
Loader(f"{prefix}/created.csv", delimiter="|"),
"created",
src_label="person",
dst_label="software",
)
graph = graph.add_vertices(Loader(f"{prefix}/software.csv", delimiter="|"),
"software")
with pytest.raises(ValueError,
match="Cannot add new relation to existed graph"):
graph = graph.add_edges(
Loader(f"{prefix}/knows.csv", delimiter="|"),
"knows",
src_label="software",
dst_label="software",
)
graph = graph.add_edges(
Loader(f"{prefix}/created.csv", delimiter="|"),
"created",
src_label="person",
dst_label="software",
)
assert graph.schema.vertex_labels == ["person", "software"]
assert graph.schema.edge_labels == ["knows", "created"]
def arrow_property_graph_lpa(graphscope_session):
g = Graph(graphscope_session, generate_eid=False)
g = g.add_vertices(f"{property_dir}/lpa_dataset/lpa_3000_v_0", "v0")
g = g.add_vertices(f"{property_dir}/lpa_dataset/lpa_3000_v_1", "v1")
g = g.add_edges(f"{property_dir}/lpa_dataset/lpa_3000_e_0", "e0",
["weight"], "v0", "v1")
yield g
g.unload()
def test_error_on_duplicate_labels(graphscope_session, student_group_e,
student_v):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v, "student")
with pytest.raises(ValueError,
match="Label student already existed in graph"):
graph = graph.add_vertices(student_v, "student")
graph = graph.add_edges(student_group_e, "group")
with pytest.raises(ValueError, match="already existed in graph"):
graph = graph.add_edges(student_group_e, "group")
def test_serialize_roundtrip(p2p_property_dir):
gs_image, gie_manager_image = get_gs_image_on_ci_env()
sess = graphscope.session(
num_workers=2,
k8s_gs_image=gs_image,
k8s_gie_graph_manager_image=gie_manager_image,
k8s_coordinator_cpu=0.5,
k8s_coordinator_mem="2500Mi",
k8s_vineyard_cpu=0.1,
k8s_vineyard_mem="512Mi",
k8s_engine_cpu=0.1,
k8s_engine_mem="1500Mi",
k8s_vineyard_shared_mem="2Gi",
k8s_volumes=get_k8s_volumes(),
)
graph = sess.load_from(
edges={
"knows": (
Loader("{}/p2p-31_property_e_0".format(p2p_property_dir),
header_row=True),
["src_label_id", "dst_label_id", "dist"],
("src_id", "person"),
("dst_id", "person"),
),
},
vertices={
"person":
Loader("{}/p2p-31_property_v_0".format(p2p_property_dir),
header_row=True),
},
generate_eid=False,
)
graph.serialize("/tmp/serialize")
graph.unload()
new_graph = Graph.deserialize("/tmp/serialize", sess)
pg = new_graph.project_to_simple(0, 0, 0, 2)
ctx = graphscope.sssp(pg, src=6)
ret = (ctx.to_dataframe({
"node": "v.id",
"r": "r"
},
vertex_range={
"end": 6
}).sort_values(by=["node"]).to_numpy(dtype=float))
expect = np.array([[1.0, 260.0], [2.0, 229.0], [3.0, 310.0], [4.0, 256.0],
[5.0, 303.0]])
assert np.all(ret == expect)
def test_load_complex_graph(
graphscope_session,
score_e,
student_teacher_e,
teacher_lesson_e,
student_v,
teacher_v,
lesson_v,
):
graph = Graph(graphscope_session, oid_type="string")
graph = graph.add_vertices(student_v, "student",
["name", "lesson_nums", "avg_score"],
"student_id")
graph = graph.add_vertices(teacher_v, "teacher",
["student_num", "score", "email", "tel"])
graph = graph.add_vertices(lesson_v, "lesson")
graph = graph.add_edges(
score_e,
"score",
["score", "score_id"],
src_label="student",
dst_label="lesson",
src_field="student_id",
dst_field="subject",
)
graph = graph.add_edges(
student_teacher_e,
"student_teacher",
["teaching_score"],
src_label="student",
dst_label="teacher",
src_field="student_id",
dst_field="teacher_id",
)
graph = graph.add_edges(
teacher_lesson_e,
"teacher_lesson",
["times"],
src_label="teacher",
dst_label="lesson",
src_field="teacher_id",
dst_field="lesson",
)
assert graph.schema is not None
def p2p_property_graph_undirected(graphscope_session):
g = Graph(graphscope_session, directed=False, generate_eid=False)
g = g.add_vertices(f"{property_dir}/p2p-31_property_v_0", "person")
g = g.add_edges(
f"{property_dir}/p2p-31_property_e_0",
label="knows",
src_label="person",
dst_label="person",
)
yield g
g.unload()
def test_error_on_operation_on_graph(graphscope_session):
g = Graph(graphscope_session.session_id)
with pytest.raises(RuntimeError,
match="The graph is not registered in remote"):
g.project_to_simple(v_label=0, v_prop=0, e_label=0, e_prop=0)
with pytest.raises(RuntimeError,
match="The graph is not registered in remote"):
g.unload()
with pytest.raises(AssertionError):
property_sssp(g, src=6)
def test_errors_on_files(graphscope_session, one_column_file,
double_type_id_file, empty_file):
with pytest.raises(AnalyticalEngineInternalError,
match="Object not exists"):
Graph(graphscope_session,
vineyard.ObjectName("non_exist_vy_name"))._ensure_loaded()
return
graph = graphscope_session.g()
with pytest.raises(AnalyticalEngineInternalError, match="IOError"):
non_existing_file = "file:///abc"
graph.add_edges(Loader(non_existing_file))._ensure_loaded()
with pytest.raises(AnalyticalEngineInternalError,
match="Index out of range: 1"):
graph.add_edges(one_column_file)._ensure_loaded()
with pytest.raises(AnalyticalEngineInternalError,
match="CSV conversion error to int64"):
graph.add_edges(double_type_id_file)._ensure_loaded()
with pytest.raises(AnalyticalEngineInternalError, match="End Of File"):
graph.add_edges(empty_file)._ensure_loaded()
def test_remove_vertices_edges(graphscope_session):
prefix = os.path.expandvars("${GS_TEST_DIR}/modern_graph")
graph = (Graph(graphscope_session).add_vertices(
Loader(f"{prefix}/person.csv", delimiter="|"),
"person").add_edges(Loader(f"{prefix}/knows.csv", delimiter="|"),
"knows"))
another_graph = graph.add_vertices(
Loader(f"{prefix}/software.csv", delimiter="|"), "software").add_edges(
Loader("{prefix}/created.csv", delimiter="|"),
"created",
src_label="person",
dst_label="software",
)
another_graph = another_graph.remove_edges("created")
another_graph = another_graph.remove_vertices("software")
assert graph.schema.vertex_labels == another_graph.schema.vertex_labels
assert graph.schema.edge_labels == another_graph.schema.edge_labels
def process_add_edges(graph, edges):
e_labels = normalize_parameter_edges(edges)
# Configurations inherited from input graph
# directed, oid_type, generate_eid
# CHECK:
# 1. edge's src/dst labels must existed in vertex_labels
# 2. label name not in existed edge labels
vertex_labels = graph.schema.vertex_labels
edge_labels = graph.schema.edge_labels
check_edge_validity(edges, vertex_labels)
for edge in edges:
check_argument(
edge.label not in edge_labels,
f"Duplicate label name with existing edge labels: {edge.label}",
)
config = assemble_op_config(e_labels, [], graph._directed,
graph.schema.oid_type, graph._generate_eid)
op = dag_utils.add_edges(graph, attrs=config)
graph_def = op.eval()
return Graph(graph.session_id, graph_def)
def test_load_from_pandas(graphscope_session, student_group_e_df,
student_v_df):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v_df)
graph = graph.add_edges(student_group_e_df)
assert graph.loaded()
def test_load_from_numpy(graphscope_session, student_group_e_array,
student_v_array):
graph = Graph(graphscope_session)
graph = graph.add_vertices(student_v_array)
graph = graph.add_edges(student_group_e_array)
assert graph.loaded()
