class WordVector(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = WordVector(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def save_vertex(self, vector_id, properties):
"""
保存词向量
:param vector_id: 词向量 ID
:param properties: 词向量属性
:return: 保存结果
"""
return self.g.save_vertex("word_vector", str(vector_id), properties)
def save_edge(self, vector_id, related_id, related_type, properties):
"""
保存词向量关系
:param vector_id: 词向量 ID
:param related_id: 相关联的顶点 ID
:param related_type: 关系类型
:param properties: 关系属性
:return: 保存结果
"""
return self.g.save_edge(related_type, str(related_id), str(vector_id),
properties)
class AddProperty(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = AddProperty(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
self.mock_data()
def mock_data(self):
# 创建一个 person 类型的顶点
dsl = """
g.addV('person')
"""
ret = self.g.exec_dsl(dsl).result().next()
print(ret)
def run(self):
# 给 person 类型的顶点添加 name 和 age 属性
dsl = """
g.V().hasLabel("person").
property('name', name).
property('age', age)
"""
bindings = {"name": "stephen", "age": "25"}
self.g.exec_dsl(dsl, bindings)
Common.get_instance().show_graph()
class Singer(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = Singer(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def save_vertex(self, singer_id, properties):
"""
保存歌手
:param singer_id: 歌手 ID
:param properties: 歌手属性
:return: 保存结果
"""
return self.g.save_vertex("singer", str(singer_id), properties)
def save_edge(self, singer_id, related_id, related_type, properties):
"""
保存歌手关系
:param singer_id: 歌手 ID
:param related_id: 相关联的顶点 ID
:param related_type: 关系类型
:param properties: 关系属性
:return: 保存结果
"""
return self.g.save_edge(related_type, str(singer_id), str(related_id),
properties)
class AddEdge(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = AddEdge(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
self.mock_data()
def mock_data(self):
# 创建两个 person 类型的顶点并设置 id 和 name
dsl = """
g.addV('person').
property(id, uid_1).
property('name', name_1).
addV('person').
property(id, uid_2).
property('name', name_2)
"""
bindings = {
"uid_1": "1",
"name_1": "marko",
"uid_2": "2",
"name_2": "stephen"
}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
def run(self):
# 给两个 person 顶点添加 knows 类型的边
dsl = """
g.V(from_uid).addE('knows').to(V(to_uid)).
property(id, edge_id)
"""
bindings = {"from_uid": "1", "to_uid": "2", "edge_id": "1-knows-2"}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
Common.get_instance().show_graph()
class And(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = And(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
self.mock_data()
def mock_data(self):
# 创建两个 person 类型的顶点
dsl = """
g.addV('person').
property(id, uid_1).
property('age', age_1).
addV('person').
property(id, uid_2).
property('age', age_2)
"""
bindings = {"uid_1": 1, "age_1": 20, "uid_2": 2, "age_2": 25}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
def run(self):
dsl = """
g.V().has('age', inside(start_1, end_1)).
and().
has('age', outside(start_2, end_2)).
values('age')
"""
# (10, 30) 与 (-∞, 15) U (20, +∞) 的交集 => 25
bindings = {'start_1': 10, 'end_1': 30, 'start_2': 15, 'end_2': 20}
vertex_list = self.g.exec_dsl(dsl, bindings).result().next()
print("ans: {}".format(vertex_list))
Common.get_instance().show_graph()
class AddVertex(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = AddVertex(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def run(self):
""" 创建一个 person 类型的顶点并设置 id 和 name """
dsl = "g.addV('person').property(id, uid).property('name', name)"
bindings = {"name": "stephen", "uid": "1"}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
Common.get_instance().show_graph()
class Utils(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = Utils(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
@staticmethod
def format_graph_fields(graph_data):
format_data = {}
for k, v in graph_data.items():
format_data[str(k)] = v[0] if type(v) == list else v
format_data['id'] = format_data['T.id']
format_data['label'] = format_data['T.label']
del format_data['T.id']
del format_data['T.label']
return format_data
def graph_visualization(self, size=100):
"""
将图数据库中的数据可视化
:param size: 起始点的数量（歌曲）
:return:
"""
dsl = """
g.V().hasLabel('song').limit(size).as('SONGS').
project('song', 'user', 'singer').
by(select('SONGS').valueMap(true)).
by(select('SONGS').in('favorite_song').valueMap(true).fold()).
by(select('SONGS').in('create_song').valueMap(true).fold())
"""
bindings = {"size": size}
callback = self.g.query_dsl(dsl, bindings)
relation = {"vertex": [], "edge": []}
for ret in callback.result():
for data in ret:
song = self.format_graph_fields(data["song"])
relation["vertex"].append(song)
user_list = []
for user in data["user"]:
user = self.format_graph_fields(user)
user_list.append(user)
relation["edge"].append({
"from": user["id"],
"to": song["id"],
"type": "favorite_song"
})
relation["vertex"].extend(user_list)
singer_list = []
for singer in data["singer"]:
singer = self.format_graph_fields(singer)
singer_list.append(singer)
relation["edge"].append({
"from": singer["id"],
"to": song["id"],
"type": "create_song"
})
relation["vertex"].extend(singer_list)
result = {
"type": "force",
"categories": [{
"name": "用户"
}, {
"name": "歌曲"
}, {
"name": "歌手"
}],
"nodes": [],
"links": []
}
category_index = {"user": 0, "song": 1, "singer": 2}
vertex_index = {}
for i, vertex in enumerate(relation["vertex"]):
result["nodes"].append({
"name": vertex["name"],
"value": 1,
"category": category_index[vertex["label"]]
})
vertex_index[vertex["id"]] = i
for edge in relation["edge"]:
result["links"].append({
"source": vertex_index[edge["from"]],
"target": vertex_index[edge["to"]]
})
return json.dumps(result)
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
self.mock_data()
class Common(object):
_instances = {}
WAITING_TIME = 1
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = Common(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def show_graph(self):
time.sleep(Common.WAITING_TIME)
vertex_list = self._get_vertex()
edge_list = self._get_edge()
data = {}
if vertex_list:
data["vertex"] = vertex_list
if edge_list:
data["edge"] = edge_list
print(json.dumps(data, indent=4, sort_keys=True))
print("-" * 50)
def _get_vertex(self):
dsl = "g.V().valueMap(true)"
callback = self.g.exec_dsl(dsl)
result = []
for ret in callback.result():
for vertex in ret:
data = self._format_graph_fields(vertex)
result.append(data)
return result
def _get_edge(self):
dsl = "g.E().valueMap(true)"
callback = self.g.exec_dsl(dsl)
result = []
for ret in callback.result():
for edge in ret:
data = self._format_graph_fields(edge)
result.append(data)
return result
@staticmethod
def _format_graph_fields(graph_data):
format_data = {}
for k, v in graph_data.items():
format_data[str(k)] = v[0] if type(v) == list else v
format_data['id'] = format_data['T.id']
format_data['label'] = format_data['T.label']
del format_data['T.id']
del format_data['T.label']
return format_data
class User(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = User(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
def find_similar_user(self, uid, top_n=3):
"""
找到与当前用户兴趣相近的用户
:param uid: 用户 ID
:return: 兴趣相近的用户信息
"""
dsl = """
g.V(uid).as('v1').
out('favorite_song').in('favorite_song').dedup().where(neq('v1')).as('v2').
project('v1', 'v2', 'v1n', 'v2n').
by(select('v1')).
by(select('v2')).
by(select('v1').out('favorite_song').fold()).
by(select('v2').out('favorite_song').fold()).
as('q1').
project('v1', 'v2', 'i', 'u').
by(select('v1')).
by(select('v2')).
by(select('v1n').as('n').
select('q1').select('v2n').unfold().
where(within('n')).
count()).
by(union(select('v1n').
select('q1').select('v2n')).unfold().
dedup().count()).
project('v1', 'v2', 'jaccard').
by(select('v1').valueMap(true)).
by(select('v2').valueMap(true)).
by(math('i/u')).order().by(select('jaccard'), desc).
limit(top_n)
"""
bindings = {"uid": uid, "top_n": top_n}
callback = self.g.query_dsl(dsl, bindings)
result = []
for ret in callback.result():
for data in ret:
v2_vertex = Utils.format_graph_fields(data["v2"])
result.append({
"uid": v2_vertex["id"],
"name": v2_vertex["name"],
"jaccard_score": round(data["jaccard"], 5)
})
return result
def save_vertex(self, uid, properties):
"""
保存用户
:param uid: 用户 ID
:param properties: 用户属性
:return: 保存结果
"""
return self.g.save_vertex("user", str(uid), properties)
def save_edge(self, uid, related_id, related_type, properties):
"""
保存用户关系
:param uid: 用户 ID
:param related_id: 相关联的顶点 ID
:param related_type: 关系类型
:param properties: 关系属性
:return: 保存结果
"""
return self.g.save_edge(related_type, str(uid), str(related_id),
properties)
class Aggregate(object):
_instances = {}
@classmethod
def get_instance(cls, graph_tag='default'):
if graph_tag not in cls._instances:
cls._instances[graph_tag] = Aggregate(graph_tag)
return cls._instances[graph_tag]
def __init__(self, graph_tag):
self.g = Graph(graph_tag)
self.mock_data()
def mock_data(self):
# 创建两个 person 类型的顶点和两个 song 类型的顶点
dsl = """
g.addV('person').
property(id, uid_1).
property('name', name_1).
addV('person').
property(id, uid_2).
property('name', name_2).
addV('song').
property(id, song_id_1).
property('name', song_name_1).
addV('song').
property(id, song_id_2).
property('name', song_name_2)
"""
bindings = {
"uid_1": "u1",
"name_1": "marko",
"uid_2": "u2",
"name_2": "stephen",
"song_id_1": "s1",
"song_name_1": "song-A",
"song_id_2": "s2",
"song_name_2": "song-B"
}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
# 添加三条边： marko-like->song-A；stephen-like-song-A；stephen-like-song-B
dsl = """
g.V(uid_1).addE('like').to(V(song_id_1)).
property(id, edge_id_1).
V(uid_2).addE('like').to(V(song_id_1)).
property(id, edge_id_2).
V(uid_2).addE('like').to(V(song_id_2)).
property(id, edge_id_3)
"""
bindings = {
"uid_1": "u1",
"name_1": "marko",
"uid_2": "u2",
"name_2": "stephen",
"song_id_1": "s1",
"song_name_1": "song-A",
"song_id_2": "s2",
"song_name_2": "song-B",
"edge_id_1": "u1-like-song-A",
"edge_id_2": "u2-like-song-A",
"edge_id_3": "u2-like-song-B"
}
ret = self.g.exec_dsl(dsl, bindings).result().next()
print(ret)
def run(self):
# 找到与 marko 有共同兴趣的人还喜欢哪些 marko 没喜欢的歌曲
dsl = """
g.V(uid_1).out('like').aggregate('x').
in('like').out('like').
where(without('x')).
values('name')
"""
bindings = {"uid_1": "u1"}
self.g.exec_dsl(dsl, bindings)
Common.get_instance().show_graph()
