push_service.py 8.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177
  1. import json
  2. import time
  3. import uuid
  4. from datetime import datetime
  5. from enum import Enum
  6. from concurrent.futures import ThreadPoolExecutor
  7. from threading import Thread
  8. from typing import Optional, Dict
  9. import rocketmq
  10. from rocketmq import ClientConfiguration, Credentials, SimpleConsumer, FilterExpression
  11. from pqai_agent import configs
  12. from pqai_agent.agents.message_push_agent import MessagePushAgent, DummyMessagePushAgent
  13. from pqai_agent.configs import apollo_config
  14. from pqai_agent.logging_service import logger
  15. from pqai_agent.message import MessageType
  16. class TaskType(Enum):
  17. GENERATE = "generate"
  18. SEND = "send"
  19. def generate_task_rmq_message(topic: str, staff_id: str, user_id: str, task_type: TaskType, content: Optional[str] = None) -> rocketmq.Message:
  20. msg = rocketmq.Message()
  21. msg.topic = topic
  22. msg.body = json.dumps({
  23. 'staff_id': staff_id,
  24. 'user_id': user_id,
  25. 'task_type': task_type.value,
  26. # FIXME: 需要支持多模态消息
  27. 'content': content or '',
  28. 'timestamp': int(time.time() * 1000),
  29. }, ensure_ascii=False).encode('utf-8')
  30. msg.tag = task_type.value
  31. return msg
  32. class PushScanThread:
  33. # PushScanThread实际可以是AgentService的一个函数,从AgentService中独立的主要考虑因素为Push后续可能有拆分和扩展
  34. def __init__(self, staff_id: str, agent_service: 'AgentService', mq_topic: str, mq_producer: rocketmq.Producer):
  35. self.staff_id = staff_id
  36. # 需要大量使用AgentService内部的成员
  37. self.service = agent_service
  38. self.rmq_topic = mq_topic
  39. self.rmq_producer = mq_producer
  40. def run(self):
  41. white_list_tags = set(apollo_config.get_json_value('agent_initiate_whitelist_tags'))
  42. first_initiate_tags = set(apollo_config.get_json_value('agent_first_initiate_whitelist_tags', []))
  43. # 合并白名单,减少配置成本
  44. white_list_tags.update(first_initiate_tags)
  45. for staff_user in self.service.user_relation_manager.list_staff_users(staff_id=self.staff_id):
  46. staff_id = staff_user['staff_id']
  47. user_id = staff_user['user_id']
  48. agent = self.service.get_agent_instance(staff_id, user_id)
  49. should_initiate = agent.should_initiate_conversation()
  50. user_tags = self.service.user_relation_manager.get_user_tags(user_id)
  51. if configs.get_env() != 'dev' and not white_list_tags.intersection(user_tags):
  52. should_initiate = False
  53. if should_initiate:
  54. logger.info(f"user[{user_id}], tags{user_tags}: generate a generation task for conversation initiation")
  55. rmq_msg = generate_task_rmq_message(self.rmq_topic, staff_id, user_id, TaskType.GENERATE)
  56. self.rmq_producer.send(rmq_msg)
  57. else:
  58. logger.debug(f"user[{user_id}], do not initiate conversation")
  59. class PushTaskWorkerPool:
  60. def __init__(self, agent_service: 'AgentService', mq_topic: str,
  61. mq_consumer: rocketmq.SimpleConsumer, mq_producer: rocketmq.Producer):
  62. self.agent_service = agent_service
  63. self.generate_executor = ThreadPoolExecutor(max_workers=5)
  64. self.send_executors = {}
  65. self.rmq_topic = mq_topic
  66. self.consumer = mq_consumer
  67. self.producer = mq_producer
  68. self.loop_thread = None
  69. self.is_generator_running = True
  70. self.generate_send_done = False # set by wait_to_finish
  71. self.no_more_generate_task = False # set by self
  72. def start(self):
  73. self.loop_thread = Thread(target=self.process_push_tasks)
  74. self.loop_thread.start()
  75. def process_push_tasks(self):
  76. # RMQ consumer疑似有bug,创建后立即消费可能报NPE
  77. time.sleep(1)
  78. while True:
  79. msgs = self.consumer.receive(1, 300)
  80. if not msgs:
  81. # 没有生成任务在执行且没有消息,才可退出
  82. if self.generate_send_done:
  83. if not self.no_more_generate_task:
  84. logger.debug("no message received, there should be no more generate task")
  85. self.no_more_generate_task = True
  86. continue
  87. else:
  88. if self.is_generator_running:
  89. logger.debug("Waiting for generator threads to finish")
  90. continue
  91. else:
  92. break
  93. else:
  94. continue
  95. msg = msgs[0]
  96. task = json.loads(msg.body.decode('utf-8'))
  97. msg_time = datetime.fromtimestamp(task['timestamp'] / 1000).strftime("%Y-%m-%d %H:%M:%S")
  98. logger.debug(f"recv message:{msg_time} - {task}")
  99. if task['task_type'] == TaskType.GENERATE.value:
  100. self.generate_executor.submit(self.handle_generate_task, task, msg)
  101. elif task['task_type'] == TaskType.SEND.value:
  102. staff_id = task['staff_id']
  103. if staff_id not in self.send_executors:
  104. self.send_executors[staff_id] = ThreadPoolExecutor(max_workers=1)
  105. self.send_executors[staff_id].submit(self.handle_send_task, task, msg)
  106. else:
  107. logger.error(f"Unknown task type: {task['task_type']}")
  108. self.consumer.ack(msg)
  109. logger.info("PushGenerateWorkerPool stopped")
  110. def wait_to_finish(self):
  111. self.generate_send_done = True
  112. while not self.no_more_generate_task:
  113. #FIXME(zhoutian): condition variable should be used to replace time sleep
  114. time.sleep(1)
  115. self.generate_executor.shutdown(wait=True)
  116. self.is_generator_running = False
  117. self.loop_thread.join()
  118. def handle_send_task(self, task: Dict, msg: rocketmq.Message):
  119. try:
  120. staff_id = task['staff_id']
  121. user_id = task['user_id']
  122. agent = self.agent_service.get_agent_instance(staff_id, user_id)
  123. # 二次校验是否需要发送
  124. if not agent.should_initiate_conversation():
  125. logger.debug(f"user[{user_id}], do not initiate conversation")
  126. self.consumer.ack(msg)
  127. return
  128. content = task['content']
  129. recent_dialogue = agent.dialogue_history[-10:]
  130. agent_voice_whitelist = set(apollo_config.get_json_value("agent_voice_whitelist"))
  131. if len(recent_dialogue) < 2 or staff_id not in agent_voice_whitelist:
  132. message_type = MessageType.TEXT
  133. else:
  134. message_type = self.agent_service.response_type_detector.detect_type(
  135. recent_dialogue[:-1], recent_dialogue[-1], enable_random=True)
  136. response = agent.generate_response(content)
  137. if response:
  138. self.agent_service.send_response(staff_id, user_id, response, message_type, skip_check=True)
  139. else:
  140. logger.debug(f"agent[{staff_id}] generate empty response")
  141. self.consumer.ack(msg)
  142. except Exception as e:
  143. logger.error(f"Error processing message sending: {e}")
  144. self.consumer.ack(msg)
  145. def handle_generate_task(self, task: Dict, msg: rocketmq.Message):
  146. try:
  147. staff_id = task['staff_id']
  148. user_id = task['user_id']
  149. main_agent = self.agent_service.get_agent_instance(staff_id, user_id)
  150. push_agent = DummyMessagePushAgent()
  151. message_to_user = push_agent.generate_message(
  152. context=main_agent.get_prompt_context(None),
  153. dialogue_history=main_agent.dialogue_history
  154. )
  155. rmq_message = generate_task_rmq_message(self.rmq_topic, staff_id, user_id, TaskType.SEND, message_to_user)
  156. logger.debug(f"send message: {rmq_message.body.decode('utf-8')}")
  157. self.producer.send(rmq_message)
  158. self.consumer.ack(msg)
  159. except Exception as e:
  160. logger.error(f"Error processing message generation: {e}")
  161. # FIXME: 是否需要ACK
  162. self.consumer.ack(msg)