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_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_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_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_serialize_roundtrip(gs_session_distributed, p2p_property_dir):
graph = Graph(gs_session_distributed, 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 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 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_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 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_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_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_remove_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",
)
with pytest.raises(ValueError,
match="Vertex software has usage in relation"):
graph = graph.remove_vertices("software")
with pytest.raises(ValueError, match="label xxx not in vertices"):
graph = graph.remove_vertices("xxx")
with pytest.raises(ValueError, match="label xxx not in edges"):
graph = graph.remove_edges("xxx")
with pytest.raises(ValueError, match="Cannot find edges to remove"):
graph = graph.remove_edges("knows", src_label="xxx", dst_label="xxx")
assert graph.loaded()
with pytest.raises(
ValueError,
match="Remove vertices from a loaded graph doesn't supported yet"):
graph = graph.remove_vertices("person")
with pytest.raises(
ValueError,
match="Remove edges from a loaded graph doesn't supported yet"):
graph = graph.remove_edges("knows")
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 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_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 arrow_property_graph_undirected(graphscope_session):
g = Graph(graphscope_session, directed=False, generate_eid=False)
g = g.add_vertices(f"{new_property_dir}/twitter_v_0", "v0")
g = g.add_vertices(f"{new_property_dir}/twitter_v_1", "v1")
g = g.add_edges(f"{new_property_dir}/twitter_e_0_0_0", "e0", ["weight"],
"v0", "v0")
g = g.add_edges(f"{new_property_dir}/twitter_e_0_1_0", "e0", ["weight"],
"v0", "v1")
g = g.add_edges(f"{new_property_dir}/twitter_e_1_0_0", "e0", ["weight"],
"v1", "v0")
g = g.add_edges(f"{new_property_dir}/twitter_e_1_1_0", "e0", ["weight"],
"v1", "v1")
g = g.add_edges(f"{new_property_dir}/twitter_e_0_0_1", "e1", ["weight"],
"v0", "v0")
g = g.add_edges(f"{new_property_dir}/twitter_e_0_1_1", "e1", ["weight"],
"v0", "v1")
g = g.add_edges(f"{new_property_dir}/twitter_e_1_0_1", "e1", ["weight"],
"v1", "v0")
g = g.add_edges(f"{new_property_dir}/twitter_e_1_1_1", "e1", ["weight"],
"v1", "v1")
yield g
g.unload()
def load_ldbc(sess, prefix, directed=True):
"""Load ldbc dataset as a ArrowProperty Graph.
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, directed=directed)
graph = (graph.add_vertices(
Loader(os.path.join(prefix, "comment_0_0.csv"), delimiter="|"),
"comment",
["creationDate", "locationIP", "browserUsed", "content", "length"],
"id",
).add_vertices(
Loader(os.path.join(prefix, "organisation_0_0.csv"), delimiter="|"),
"organisation",
["type", "name", "url"],
"id",
).add_vertices(
Loader(os.path.join(prefix, "tagclass_0_0.csv"), delimiter="|"),
"tagclass",
["name", "url"],
"id",
).add_vertices(
Loader(os.path.join(prefix, "person_0_0.csv"), delimiter="|"),
"person",
[
"firstName",
"lastName",
"gender",
"birthday",
"creationDate",
"locationIP",
"browserUsed",
],
"id",
).add_vertices(
Loader(os.path.join(prefix, "forum_0_0.csv"), delimiter="|"),
"forum",
["title", "creationDate"],
"id",
).add_vertices(
Loader(os.path.join(prefix, "place_0_0.csv"), delimiter="|"),
"place",
["name", "url", "type"],
"id",
).add_vertices(
Loader(os.path.join(prefix, "post_0_0.csv"), delimiter="|"),
"post",
[
"imageFile",
"creationDate",
"locationIP",
"browserUsed",
"language",
"content",
"length",
],
"id",
).add_vertices(
Loader(os.path.join(prefix, "tag_0_0.csv"), delimiter="|"),
"tag",
["name", "url"],
"id",
).add_edges(
Loader(os.path.join(prefix, "comment_replyOf_comment_0_0.csv"),
delimiter="|"),
"replyOf",
src_label="comment",
dst_label="comment",
).add_edges(
Loader(os.path.join(prefix, "comment_replyOf_post_0_0.csv"),
delimiter="|"),
"replyOf",
src_label="comment",
dst_label="post",
).add_edges(
Loader(os.path.join(prefix, "place_isPartOf_place_0_0.csv"),
delimiter="|"),
"isPartOf",
src_label="place",
dst_label="place",
).add_edges(
Loader(
os.path.join(prefix, "tagclass_isSubclassOf_tagclass_0_0.csv"),
delimiter="|",
),
"isSubclassOf",
src_label="tagclass",
dst_label="tagclass",
).add_edges(
Loader(os.path.join(prefix, "forum_hasTag_tag_0_0.csv"),
delimiter="|"),
"hasTag",
src_label="forum",
dst_label="tag",
).add_edges(
Loader(os.path.join(prefix, "comment_hasTag_tag_0_0.csv"),
delimiter="|"),
"hasTag",
src_label="comment",
dst_label="tag",
).add_edges(
Loader(os.path.join(prefix, "post_hasTag_tag_0_0.csv"), delimiter="|"),
"hasTag",
src_label="post",
dst_label="tag",
).add_edges(
Loader(os.path.join(prefix, "person_knows_person_0_0.csv"),
delimiter="|"),
"knows",
["creationDate"],
src_label="person",
dst_label="person",
).add_edges(
Loader(os.path.join(prefix, "forum_hasModerator_person_0_0.csv"),
delimiter="|"),
"hasModerator",
src_label="forum",
dst_label="person",
).add_edges(
Loader(os.path.join(prefix, "person_hasInterest_tag_0_0.csv"),
delimiter="|"),
"hasInterest",
src_label="person",
dst_label="tag",
).add_edges(
Loader(os.path.join(prefix, "post_isLocatedIn_place_0_0.csv"),
delimiter="|"),
"isLocatedIn",
src_label="post",
dst_label="place",
).add_edges(
Loader(os.path.join(prefix, "comment_isLocatedIn_place_0_0.csv"),
delimiter="|"),
"isLocatedIn",
src_label="comment",
dst_label="place",
).add_edges(
Loader(
os.path.join(prefix, "organisation_isLocatedIn_place_0_0.csv"),
delimiter="|",
),
"isLocatedIn",
src_label="organisation",
dst_label="place",
).add_edges(
Loader(os.path.join(prefix, "person_isLocatedIn_place_0_0.csv"),
delimiter="|"),
"isLocatedIn",
src_label="person",
dst_label="place",
).add_edges(
Loader(os.path.join(prefix, "tag_hasType_tagclass_0_0.csv"),
delimiter="|"),
"hasType",
src_label="tag",
dst_label="tagclass",
).add_edges(
Loader(os.path.join(prefix, "post_hasCreator_person_0_0.csv"),
delimiter="|"),
"hasCreator",
src_label="post",
dst_label="person",
).add_edges(
Loader(os.path.join(prefix, "comment_hasCreator_person_0_0.csv"),
delimiter="|"),
"hasCreator",
src_label="comment",
dst_label="person",
).add_edges(
Loader(os.path.join(prefix, "forum_containerOf_post_0_0.csv"),
delimiter="|"),
"containerOf",
src_label="forum",
dst_label="post",
).add_edges(
Loader(os.path.join(prefix, "forum_hasMember_person_0_0.csv"),
delimiter="|"),
"hasMember",
["joinDate"],
src_label="forum",
dst_label="person",
).add_edges(
Loader(
os.path.join(prefix, "person_workAt_organisation_0_0.csv"),
delimiter="|",
),
"workAt",
["workFrom"],
src_label="person",
dst_label="organisation",
).add_edges(
Loader(os.path.join(prefix, "person_likes_comment_0_0.csv"),
delimiter="|"),
"likes",
["creationDate"],
src_label="person",
dst_label="comment",
).add_edges(
Loader(os.path.join(prefix, "person_likes_post_0_0.csv"),
delimiter="|"),
"likes",
["creationDate"],
src_label="person",
dst_label="post",
).add_edges(
Loader(
os.path.join(prefix, "person_studyAt_organisation_0_0.csv"),
delimiter="|",
),
"studyAt",
["classYear"],
src_label="person",
dst_label="organisation",
))
return graph
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()
