push_service.py 8.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181
  1. import json
  2. import time
  3. import traceback
  4. import uuid
  5. from datetime import datetime
  6. from enum import Enum
  7. from concurrent.futures import ThreadPoolExecutor
  8. from threading import Thread
  9. from typing import Optional, Dict
  10. import rocketmq
  11. from rocketmq import ClientConfiguration, Credentials, SimpleConsumer, FilterExpression
  12. from pqai_agent import configs
  13. from pqai_agent.agents.message_push_agent import MessagePushAgent, DummyMessagePushAgent
  14. from pqai_agent.configs import apollo_config
  15. from pqai_agent.logging_service import logger
  16. from pqai_agent.message import MessageType
  17. class TaskType(Enum):
  18. GENERATE = "generate"
  19. SEND = "send"
  20. def generate_task_rmq_message(topic: str, staff_id: str, user_id: str, task_type: TaskType, content: Optional[str] = None) -> rocketmq.Message:
  21. msg = rocketmq.Message()
  22. msg.topic = topic
  23. msg.body = json.dumps({
  24. 'staff_id': staff_id,
  25. 'user_id': user_id,
  26. 'task_type': task_type.value,
  27. # FIXME: 需要支持多模态消息
  28. 'content': content or '',
  29. 'timestamp': int(time.time() * 1000),
  30. }, ensure_ascii=False).encode('utf-8')
  31. msg.tag = task_type.value
  32. return msg
  33. class PushScanThread:
  34. # 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
  35. def __init__(self, staff_id: str, agent_service: 'AgentService', mq_topic: str, mq_producer: rocketmq.Producer):
  36. self.staff_id = staff_id
  37. # 需要大量使用AgentService内部的成员
  38. self.service = agent_service
  39. self.rmq_topic = mq_topic
  40. self.rmq_producer = mq_producer
  41. def run(self):
  42. white_list_tags = set(apollo_config.get_json_value('agent_initiate_whitelist_tags', []))
  43. first_initiate_tags = set(apollo_config.get_json_value('agent_first_initiate_whitelist_tags', []))
  44. # 合并白名单,减少配置成本
  45. white_list_tags.update(first_initiate_tags)
  46. for staff_user in self.service.user_relation_manager.list_staff_users(staff_id=self.staff_id):
  47. staff_id = staff_user['staff_id']
  48. user_id = staff_user['user_id']
  49. agent = self.service.get_agent_instance(staff_id, user_id)
  50. should_initiate = agent.should_initiate_conversation()
  51. user_tags = self.service.user_relation_manager.get_user_tags(user_id)
  52. if configs.get_env() != 'dev' and not white_list_tags.intersection(user_tags):
  53. should_initiate = False
  54. if should_initiate:
  55. logger.info(f"user[{user_id}], tags{user_tags}: generate a generation task for conversation initiation")
  56. rmq_msg = generate_task_rmq_message(self.rmq_topic, staff_id, user_id, TaskType.GENERATE)
  57. self.rmq_producer.send(rmq_msg)
  58. else:
  59. logger.debug(f"user[{user_id}], do not initiate conversation")
  60. class PushTaskWorkerPool:
  61. def __init__(self, agent_service: 'AgentService', mq_topic: str,
  62. mq_consumer: rocketmq.SimpleConsumer, mq_producer: rocketmq.Producer):
  63. self.agent_service = agent_service
  64. self.generate_executor = ThreadPoolExecutor(max_workers=5)
  65. self.send_executors = {}
  66. self.rmq_topic = mq_topic
  67. self.consumer = mq_consumer
  68. self.producer = mq_producer
  69. self.loop_thread = None
  70. self.is_generator_running = True
  71. self.generate_send_done = False # set by wait_to_finish
  72. self.no_more_generate_task = False # set by self
  73. def start(self):
  74. self.loop_thread = Thread(target=self.process_push_tasks)
  75. self.loop_thread.start()
  76. def process_push_tasks(self):
  77. # RMQ consumer疑似有bug,创建后立即消费可能报NPE
  78. time.sleep(1)
  79. while True:
  80. msgs = self.consumer.receive(1, 300)
  81. if not msgs:
  82. # 没有生成任务在执行且没有消息,才可退出
  83. if self.generate_send_done:
  84. if not self.no_more_generate_task:
  85. logger.debug("no message received, there should be no more generate task")
  86. self.no_more_generate_task = True
  87. continue
  88. else:
  89. if self.is_generator_running:
  90. logger.debug("Waiting for generator threads to finish")
  91. continue
  92. else:
  93. break
  94. else:
  95. continue
  96. msg = msgs[0]
  97. task = json.loads(msg.body.decode('utf-8'))
  98. msg_time = datetime.fromtimestamp(task['timestamp'] / 1000).strftime("%Y-%m-%d %H:%M:%S")
  99. logger.debug(f"recv message:{msg_time} - {task}")
  100. if task['task_type'] == TaskType.GENERATE.value:
  101. self.generate_executor.submit(self.handle_generate_task, task, msg)
  102. elif task['task_type'] == TaskType.SEND.value:
  103. staff_id = task['staff_id']
  104. if staff_id not in self.send_executors:
  105. self.send_executors[staff_id] = ThreadPoolExecutor(max_workers=1)
  106. self.send_executors[staff_id].submit(self.handle_send_task, task, msg)
  107. else:
  108. logger.error(f"Unknown task type: {task['task_type']}")
  109. self.consumer.ack(msg)
  110. logger.info("PushGenerateWorkerPool stopped")
  111. def wait_to_finish(self):
  112. self.generate_send_done = True
  113. while not self.no_more_generate_task:
  114. #FIXME(zhoutian): condition variable should be used to replace time sleep
  115. time.sleep(1)
  116. self.generate_executor.shutdown(wait=True)
  117. self.is_generator_running = False
  118. self.loop_thread.join()
  119. def handle_send_task(self, task: Dict, msg: rocketmq.Message):
  120. try:
  121. staff_id = task['staff_id']
  122. user_id = task['user_id']
  123. agent = self.agent_service.get_agent_instance(staff_id, user_id)
  124. # 二次校验是否需要发送
  125. if not agent.should_initiate_conversation():
  126. logger.debug(f"user[{user_id}], do not initiate conversation")
  127. self.consumer.ack(msg)
  128. return
  129. content = task['content']
  130. recent_dialogue = agent.dialogue_history[-10:]
  131. agent_voice_whitelist = set(apollo_config.get_json_value("agent_voice_whitelist"))
  132. # FIXME(zhoutian): 不应该再由agent控制,或者agent和API共享同一配置
  133. if len(recent_dialogue) < 2 or staff_id not in agent_voice_whitelist:
  134. message_type = MessageType.TEXT
  135. else:
  136. message_type = self.agent_service.response_type_detector.detect_type(
  137. recent_dialogue, content, enable_random=True)
  138. response = agent.generate_response(content)
  139. if response:
  140. self.agent_service.send_response(staff_id, user_id, response, message_type, skip_check=True)
  141. else:
  142. logger.debug(f"agent[{staff_id}] generate empty response")
  143. self.consumer.ack(msg)
  144. except Exception as e:
  145. fmt_exc = traceback.format_exc()
  146. logger.error(f"Error processing message sending: {e}, {fmt_exc}")
  147. self.consumer.ack(msg)
  148. def handle_generate_task(self, task: Dict, msg: rocketmq.Message):
  149. try:
  150. staff_id = task['staff_id']
  151. user_id = task['user_id']
  152. main_agent = self.agent_service.get_agent_instance(staff_id, user_id)
  153. push_agent = MessagePushAgent()
  154. message_to_user = push_agent.generate_message(
  155. context=main_agent.get_prompt_context(None),
  156. dialogue_history=main_agent.dialogue_history
  157. )
  158. rmq_message = generate_task_rmq_message(self.rmq_topic, staff_id, user_id, TaskType.SEND, message_to_user)
  159. logger.debug(f"send message: {rmq_message.body.decode('utf-8')}")
  160. self.producer.send(rmq_message)
  161. self.consumer.ack(msg)
  162. except Exception as e:
  163. fmt_exc = traceback.format_exc()
  164. logger.error(f"Error processing message generation: {e}, {fmt_exc}")
  165. # FIXME: 是否需要ACK
  166. self.consumer.ack(msg)