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)
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)