class ChatServicer(chatbot_pb2_grpc.ChatBotServiceServicer):
def __init__(self):
# 提前加载各种模型
self.classify = Classify()
self.recall = Recall()
self.sort = DnnSort()
self.chatbot = Chatbot()
self.message_manager = MessageManager()
def Chatbot(self, request, context):
user_id = request.user_id
message = request.user_message
create_time = request.create_time
attention, prob = self.classify.predict(message)
if attention == "QA":
# 实现对对话数据的保存
self.message_manager.user_message_pipeline(
user_id,
message,
create_time,
attention,
entity=message["entity"])
recall_list, entity = self.recall.predict(message)
user_response = self.sort.predict(message, recall_list)
else:
# 实现对对话数据的保存
self.message_manager.user_message_pipeline(
user_id,
message,
create_time,
attention,
entity=message["entity"])
user_response = self.chatbot.predict(message)
self.message_manager.bot_message_pipeline(user_id, user_response)
create_time = int(time.time())
return chatbot_pb2.ResponsedMessage(user_response=user_response,
create_time=create_time)
ws.build_vocab()
print(len(ws))
pickle.dump(ws, open(config.chatbot_ws_by_word_input_path, "wb"))
ws = Word_sequence()
for line in open(config.chatbot_target_path, "r",
encoding="utf-8").readlines():
ws.fit(line.strip().split())
ws.build_vocab()
print(len(ws))
pickle.dump(ws, open(config.chatbot_ws_by_word_target_path, "wb"))
def train_seq2seq():
for i in range(10):
train(i)
if __name__ == '__main__':
# save_ws()
# for idx,(input,target,input_length,target_length) in enumerate(dataloader):
# print(input)
# print(target)
# print(input_length)
# print(target_length)
# break
# train_seq2seq()
chatbot = Chatbot()
while True:
print(chatbot.predict(input("请输入：")))
