AgentCoreService/pqai_agent/agent_service.py

#! /usr/bin/env python
# -*- coding: utf-8 -*-
# vim:fenc=utf-8

import re
import signal
import sys
import time
from typing import Dict, List, Optional, Union
import logging
from datetime import datetime, timedelta
import threading
import traceback

import apscheduler.triggers.cron
import rocketmq
from apscheduler.schedulers.background import BackgroundScheduler
from rocketmq import FilterExpression
from sqlalchemy.orm import sessionmaker

from pqai_agent import configs, push_service
from pqai_agent.abtest.utils import get_abtest_info
from pqai_agent.agent_config_manager import AgentConfigManager
from pqai_agent.agents.message_reply_agent import MessageReplyAgent
from pqai_agent.configs import apollo_config
from pqai_agent.exceptions import NoRetryException
from pqai_agent.logging import logger
from pqai_agent import chat_service
from pqai_agent.chat_service import CozeChat, ChatServiceType
from pqai_agent.dialogue_manager import DialogueManager, DialogueState, DialogueStateCache
from pqai_agent.history_dialogue_service import HistoryDialogueDatabase
from pqai_agent.push_service import PushScanThread, PushTaskWorkerPool
from pqai_agent.rate_limiter import MessageSenderRateLimiter
from pqai_agent.response_type_detector import ResponseTypeDetector
from pqai_agent.service_module_manager import ServiceModuleManager
from pqai_agent.toolkit import get_tools
from pqai_agent.user_manager import UserManager, UserRelationManager
from pqai_agent.message_queue_backend import MessageQueueBackend, AliyunRocketMQQueueBackend
from pqai_agent.user_profile_extractor import UserProfileExtractor
from pqai_agent.mq_message import MessageType, MqMessage, MessageChannel
from pqai_agent.utils.agent_abtest_utils import get_agent_abtest_config
from pqai_agent.utils.db_utils import create_ai_agent_db_engine


class AgentService:
    def __init__(
        self,
        receive_backend: MessageQueueBackend,
        send_backend: MessageQueueBackend,
        human_backend: MessageQueueBackend,
        user_manager: UserManager,
        user_relation_manager: UserRelationManager,
        chat_service_type: ChatServiceType = ChatServiceType.OPENAI_COMPATIBLE
    ):
        self.config = configs.get()

        self.receive_queue = receive_backend
        self.send_queue = send_backend
        self.human_queue = human_backend

        # 核心服务模块
        self.agent_state_cache = DialogueStateCache()
        self.user_manager = user_manager
        self.user_relation_manager = user_relation_manager
        self.user_profile_extractor = UserProfileExtractor()
        self.response_type_detector = ResponseTypeDetector()
        self.agent_registry: Dict[str, DialogueManager] = {}
        self.history_dialogue_db = HistoryDialogueDatabase(self.config['database']['ai_agent'])
        self.agent_db_engine = create_ai_agent_db_engine()
        self.agent_db_session_maker = sessionmaker(bind=self.agent_db_engine)

        chat_config = self.config['chat_api']['openai_compatible']
        self.text_model_name = chat_config['text_model']
        self.multimodal_model_name = chat_config['multimodal_model']
        self.text_model_client = chat_service.OpenAICompatible.create_client(self.text_model_name)
        self.multimodal_model_client = chat_service.OpenAICompatible.create_client(self.multimodal_model_name)

        coze_config = configs.get()['chat_api']['coze']
        coze_oauth_app = CozeChat.get_oauth_app(
            coze_config['oauth_client_id'], coze_config['private_key_path'], str(coze_config['public_key_id']),
            account_id=coze_config.get('account_id', None)
        )
        self.coze_client = CozeChat(
            base_url=chat_service.COZE_CN_BASE_URL,
            auth_app=coze_oauth_app
        )
        self.chat_service_type = chat_service_type

        # 定时任务调度器
        self.scheduler = None
        self.scheduler_mode = self.config.get('system', {}).get('scheduler_mode', 'local')
        self.scheduler_queue = None
        self.msg_scheduler_thread = None
        self.running = False
        self.process_threads = []
        self._sigint_cnt = 0

        # Push相关
        self.push_task_producer = None
        self.push_generate_task_consumer = None
        self.push_send_task_consumer = None
        self._init_push_task_queue()
        self.next_push_disabled = True
        self._resume_unfinished_push_task()

        self.send_rate_limiter = MessageSenderRateLimiter()

        # Agent配置和实验相关
        self.service_module_manager = ServiceModuleManager(self.agent_db_session_maker)
        self.agent_config_manager = AgentConfigManager(self.agent_db_session_maker)

    def setup_initiative_conversations(self, schedule_params: Optional[Dict] = None):
        if not schedule_params:
            schedule_params = {'hour': '8,16,20'}
        self.scheduler.add_job(
            self._check_initiative_conversations,
            apscheduler.triggers.cron.CronTrigger(**schedule_params)
        )

    def setup_scheduler(self):
        self.scheduler = BackgroundScheduler()
        if self.scheduler_mode == 'mq':
            logging.info("setup event message scheduler with MQ")
            mq_conf = self.config['mq']
            topic = mq_conf['scheduler_topic']
            self.scheduler_queue = AliyunRocketMQQueueBackend(
                mq_conf['endpoints'],
                mq_conf['instance_id'],
                topic,
                has_consumer=True, has_producer=True,
                group_id=mq_conf['scheduler_group'],
                topic_type='DELAY',
                await_duration=5
            )
            self.msg_scheduler_thread = threading.Thread(target=self.process_scheduler_events)
            self.msg_scheduler_thread.start()
        # 定时更新模块配置任务
        self.scheduler.add_job(self.service_module_manager.refresh_configs, 'interval',
                               seconds=60, id='refresh_module_configs')
        self.scheduler.add_job(self.agent_config_manager.refresh_configs, 'interval',
                               seconds=60, id='refresh_agent_configs')
        self.scheduler.start()

    def process_scheduler_events(self):
        while self.running:
            msg = self.scheduler_queue.consume()
            if msg:
                try:
                    self.process_scheduler_event(msg)
                    self.scheduler_queue.ack(msg)
                except Exception as e:
                    logger.error("Error processing scheduler event: {}".format(e))
            time.sleep(1)
        logger.info("Scheduler event processing thread exit")

    def process_scheduler_event(self, msg: MqMessage):
        if msg.type == MessageType.AGGREGATION_TRIGGER:
            # 延迟触发的消息，需放入接收队列以驱动Agent运转
            self.receive_queue.produce(msg)
        else:
            logger.warning(f"Unknown message type: {msg.type}")

    def get_agent_instance(self, staff_id: str, user_id: str) -> DialogueManager:
        """获取Agent实例"""
        agent_key = 'agent_{}_{}'.format(staff_id, user_id)
        if agent_key not in self.agent_registry:
            self.agent_registry[agent_key] = DialogueManager(
                staff_id, user_id, self.user_manager, self.agent_state_cache, self.agent_db_session_maker)
        agent = self.agent_registry[agent_key]
        agent.refresh_profile()
        return agent

    def create_queue_consumer(self) -> MessageQueueBackend:
        # 只有在MQ模式下才需要创建多消费者
        if not self.config.get('debug_flags', {}).get('use_aliyun_mq', False):
            logger.warning("Do not create queue consumer in local mode")
            return self.receive_queue
        mq_config = self.config['mq']
        consumer = AliyunRocketMQQueueBackend(
            endpoints=mq_config['endpoints'],
            instance_id=mq_config['instance_id'],
            topic=mq_config['receive_topic'],
            has_consumer=True,
            has_producer=False,
            group_id=mq_config['receive_group'],
            topic_type='FIFO',
            await_duration=10
        )
        return consumer

    def process_messages(self):
        """持续处理接收队列消息，通过顺序消息的消息组保证同一<用户, 客服>的消费保序，可并发处理"""
        receive_queue = self.create_queue_consumer()
        # 消费者创建后等一会儿再开始消费，否则可能远端没准备好会报错
        time.sleep(1)
        while self.running:
            message = receive_queue.consume()
            if message:
                try:
                    self.process_single_message(message)
                    receive_queue.ack(message)
                except NoRetryException as e:
                    logger.error("Error processing message and skip retry: {}".format(e))
                    receive_queue.ack(message)
                except Exception as e:
                    error_stack = traceback.format_exc()
                    logger.error("Error processing message: {}, {}".format(e, error_stack))
            time.sleep(0.1)
        receive_queue.shutdown()
        logger.info("MqMessage processing thread exit")

    def start(self, blocking=False):
        self.running = True
        max_reply_workers = self.config.get('system', {}).get('max_reply_workers', 1)
        self.process_threads = []
        for i in range(max_reply_workers):
            thread = threading.Thread(target=self.process_messages)
            thread.start()
            self.process_threads.append(thread)
        self.setup_scheduler()
        # 只有企微场景需要主动发起
        if not self.config['debug_flags'].get('disable_active_conversation', False):
            schedule_param = self.config['agent_behavior'].get('active_conversation_schedule_param', None)
            self.setup_initiative_conversations(schedule_param)
        signal.signal(signal.SIGINT, self._handle_sigint)
        if blocking:
            for thread in self.process_threads:
                thread.join()
            logger.debug("process threads finished")

    def shutdown(self, sync=True):
        if not self.running:
            raise Exception("Service is not running")
        self.running = False
        self.scheduler.shutdown()
        logger.debug("scheduler shutdown")
        if sync:
            for thread in self.process_threads:
                thread.join()
            logger.debug("message processing threads finished")
            if self.msg_scheduler_thread:
                self.msg_scheduler_thread.join()
                self.scheduler_queue.shutdown()
                logger.debug("scheduler message processing thread finished")
            self.receive_queue.shutdown()
            self.send_queue.shutdown()
            logger.debug("receive and send queues shutdown")

    def _handle_sigint(self, signum, frame):
        self._sigint_cnt += 1
        if self._sigint_cnt == 1:
            logger.warning("Try to shutdown gracefully...")
            self.shutdown(sync=True)
        else:
            logger.warning("Forcing exit")
            sys.exit(0)

    def _update_user_profile(self, user_id, user_profile, recent_dialogue: List[Dict]):
        agent_info = get_agent_abtest_config('profile_extractor', user_id, self.service_module_manager, self.agent_config_manager)
        if agent_info:
            prompt_template = agent_info.task_prompt
        else:
            prompt_template = None
        profile_to_update = self.user_profile_extractor.extract_profile_info_v2(user_profile, recent_dialogue, prompt_template)
        if not profile_to_update:
            logger.debug("user_id: {}, no profile info extracted".format(user_id))
            return
        logger.warning("update user profile: {}".format(profile_to_update))
        if profile_to_update.get('interaction_frequency', None) == 'stopped':
            # 和企微日常push联动，减少对用户的干扰
            if self.user_relation_manager.stop_user_daily_push(user_id):
                logger.warning(f"user[{user_id}]: daily push set to be stopped")
        merged_profile = self.user_profile_extractor.merge_profile_info(user_profile, profile_to_update)
        self.user_manager.save_user_profile(user_id, merged_profile)
        return merged_profile

    def _schedule_aggregation_trigger(self, staff_id: str, user_id: str, delay_sec: int):
        logger.debug("user: {}, schedule trigger message after {} seconds".format(user_id, delay_sec))
        message_ts = int((time.time() + delay_sec) * 1000)
        msg = MqMessage.build(MessageType.AGGREGATION_TRIGGER, MessageChannel.SYSTEM, user_id, staff_id, None, message_ts)
        # 系统消息使用特定的msgId，无实际意义
        msg.msgId = -MessageType.AGGREGATION_TRIGGER.value
        if self.scheduler_mode == 'mq':
            self.scheduler_queue.produce(msg, msg_group='agent_system')
        else:
            self.scheduler.add_job(lambda: self.receive_queue.produce(msg),
                                   'date',
                                   run_date=datetime.now() + timedelta(seconds=delay_sec))

    def process_single_message(self, message: MqMessage):
        user_id = message.sender
        staff_id = message.receiver

        # 获取用户信息和Agent实例
        user_profile = self.user_manager.get_user_profile(user_id)
        agent = self.get_agent_instance(staff_id, user_id)
        if not agent.is_valid():
            logger.error(f"staff[{staff_id}] user[{user_id}]: agent is invalid")
            raise Exception('agent is invalid')

        # 更新对话状态
        logger.debug("process message: {}".format(message))
        need_response, message_text = agent.update_state(message)
        logger.debug("user: {}, next state: {}".format(user_id, agent.current_state))

        # 根据状态路由消息
        try:
            if agent.is_in_human_intervention():
                self._route_to_human_intervention(user_id, message)
            elif agent.current_state == DialogueState.MESSAGE_AGGREGATING:
                if message.type != MessageType.AGGREGATION_TRIGGER:
                    # 产生一个触发器，但是不能由触发器递归产生
                    logger.debug("user: {}, waiting next message for aggregation".format(user_id))
                    self._schedule_aggregation_trigger(staff_id, user_id, agent.message_aggregation_sec)
            elif need_response:
                # 先更新用户画像再处理回复
                self._update_user_profile(user_id, user_profile, agent.dialogue_history[-10:])
                resp = self.get_chat_response(agent, message_text)
                self.send_responses(agent, resp)
            else:
                logger.debug(f"staff[{staff_id}], user[{user_id}]: do not need response")
            # 当前消息处理成功，commit并持久化agent状态
            agent.persist_state()
        except Exception as e:
            agent.rollback_state()
            raise e

    def send_responses(self, agent: DialogueManager, contents: List[Dict]):
        staff_id = agent.staff_id
        user_id = agent.user_id
        recent_dialogue = agent.dialogue_history[-10:]
        agent_voice_whitelist = set(apollo_config.get_json_value("agent_voice_whitelist", []))
        current_ts = int(time.time())
        for item in contents:
            item["timestamp"] = current_ts * 1000
            if item["type"] == MessageType.TEXT:
                if staff_id in agent_voice_whitelist or True:
                    message_type = self.response_type_detector.detect_type(
                        recent_dialogue, item, enable_random=True)
                    item["type"] = message_type
        if contents:
            for response in contents:
                self.send_multimodal_response(staff_id, user_id, response)
            agent.update_last_active_interaction_time(current_ts)
        else:
            logger.debug(f"staff[{staff_id}], user[{user_id}]: no messages to send")

    def can_send_to_user(self, staff_id, user_id) -> bool:
        user_tags = self.user_manager.get_user_tags([user_id]).get(user_id, [])
        white_list_tags = set(apollo_config.get_json_value("agent_response_whitelist_tags", []))
        hit_white_list_tags = len(set(user_tags).intersection(white_list_tags)) > 0
        staff_white_lists = set(apollo_config.get_json_value("agent_response_whitelist_staffs", []))
        if not (staff_id in staff_white_lists or hit_white_list_tags):
            logger.warning(f"staff[{staff_id}] user[{user_id}]: skip reply")
            return False
        return True

    def send_multimodal_response(self, staff_id, user_id, response: Dict, skip_check=False):
        message_type = response["type"]
        logger.warning(f"staff[{staff_id}] user[{user_id}]: response[{message_type}] {response}")
        if message_type not in (MessageType.TEXT, MessageType.IMAGE_QW, MessageType.VOICE):
            logger.error(f"staff[{staff_id}] user[{user_id}]: unsupported message type {message_type}")
            return
        if not skip_check and not self.can_send_to_user(staff_id, user_id):
            return
        current_ts = int(time.time() * 1000)
        self.send_rate_limiter.wait_for_sending(staff_id, response)
        self.send_queue.produce(
            MqMessage.build(message_type, MessageChannel.CORP_WECHAT,
                            staff_id, user_id, response["content"], current_ts)
        )

    def _route_to_human_intervention(self, user_id: str, origin_message: MqMessage):
        """路由到人工干预"""
        self.human_queue.produce(MqMessage.build(
            MessageType.TEXT,
            origin_message.channel,
            origin_message.sender,
            origin_message.receiver,
            "用户对话需人工介入，用户名：{}".format(user_id),
            int(time.time() * 1000)
        ))

    def _init_push_task_queue(self):
        credentials = rocketmq.Credentials()
        mq_conf = configs.get()['mq']
        rmq_client_conf = rocketmq.ClientConfiguration(mq_conf['endpoints'], credentials, mq_conf['instance_id'])
        rmq_topic = mq_conf['push_tasks_topic']
        rmq_group_generate = mq_conf['push_generate_task_group']
        rmq_group_send = mq_conf['push_send_task_group']
        self.push_task_rmq_topic = rmq_topic
        self.push_task_producer = rocketmq.Producer(rmq_client_conf, (rmq_topic,))
        self.push_task_producer.startup()
        # FIXME: 不应该暴露到agent service中
        self.push_generate_task_consumer = rocketmq.SimpleConsumer(rmq_client_conf, rmq_group_generate, await_duration=5)
        self.push_generate_task_consumer.startup()
        self.push_generate_task_consumer.subscribe(
            rmq_topic, filter_expression=FilterExpression(push_service.TaskType.GENERATE.value)
        )
        self.push_send_task_consumer = rocketmq.SimpleConsumer(rmq_client_conf, rmq_group_send, await_duration=5)
        self.push_send_task_consumer.startup()
        self.push_send_task_consumer.subscribe(
            rmq_topic, filter_expression=FilterExpression(push_service.TaskType.SEND.value)
        )


    def _resume_unfinished_push_task(self):
        def run_unfinished_push_task():
            logger.info("start to resume unfinished push task")
            push_task_worker_pool = PushTaskWorkerPool(
                self, self.push_task_rmq_topic, self.push_generate_task_consumer,
                self.push_send_task_consumer, self.push_task_producer
            )
            push_task_worker_pool.start()
            push_task_worker_pool.wait_to_finish()
            self.next_push_disabled = False
            logger.info("unfinished push tasks should be finished")
        thread = threading.Thread(target=run_unfinished_push_task)
        thread.start()

    def _check_initiative_conversations(self):
        logger.info("start to check initiative conversations")
        if self.next_push_disabled:
            logger.info("previous push tasks in processing, next push is disabled")
            return
        if not DialogueManager.is_time_suitable_for_active_conversation():
            logger.info("time is not suitable for active conversation")
            return

        push_scan_threads = []
        whitelist_staffs = apollo_config.get_json_value("agent_initiate_whitelist_staffs", [])
        for staff in self.user_relation_manager.list_staffs():
            staff_id = staff['third_party_user_id']
            if staff_id not in whitelist_staffs:
                logger.info(f"staff[{staff_id}] is not in whitelist, skip")
                continue
            scan_thread = threading.Thread(target=PushScanThread(
                staff_id, self, self.push_task_rmq_topic, self.push_task_producer).run)
            scan_thread.start()
            push_scan_threads.append(scan_thread)

        push_task_worker_pool = PushTaskWorkerPool(
            self, self.push_task_rmq_topic,
            self.push_generate_task_consumer, self.push_send_task_consumer, self.push_task_producer)
        push_task_worker_pool.start()
        for thread in push_scan_threads:
            thread.join()
        # 由于扫描和生成异步，两次扫描之间可能消息并未处理完，会有重复生成任务的情况，因此需等待上一次任务结束
        # 问题在于，如果每次创建出新的PushTaskWorkerPool，在上次任务有未处理完的消息即退出时，会有未处理的消息堆积
        push_task_worker_pool.wait_to_finish()

    def get_chat_response(self, agent: DialogueManager, user_message: Optional[str]) -> List[Dict]:
        chat_agent_ver = self.config.get('system', {}).get('chat_agent_version', 1)
        if chat_agent_ver == 2:
            return self._get_chat_response_v2(agent)
        else:
            text_resp = self._get_chat_response_v1(agent, user_message)
            return [{"type": MessageType.TEXT, "content": text_resp}] if text_resp else []

    def _get_chat_response_v1(self, agent: DialogueManager, user_message: Optional[str]) -> Optional[str]:
        chat_config = agent.build_chat_configuration(user_message, self.chat_service_type)
        config_for_logging = chat_config.copy()
        config_for_logging['messages'] = config_for_logging['messages'][-20:]
        logger.debug(config_for_logging)
        chat_response = self._call_chat_api(chat_config, self.chat_service_type)
        chat_response = self.sanitize_response(chat_response)

        if response := agent.generate_response(chat_response):
            return response
        else:
            logger.warning(f"staff[{agent.staff_id}] user[{agent.user_id}]: no response generated")
            return None

    def _get_chat_response_v2(self, main_agent: DialogueManager) -> List[Dict]:
        agent_config = get_agent_abtest_config('chat', main_agent.user_id,
                                               self.service_module_manager, self.agent_config_manager)
        if agent_config:
            chat_agent = MessageReplyAgent(model=agent_config.execution_model,
                                           system_prompt=agent_config.system_prompt,
                                           tools=get_tools(agent_config.tools))
        else:
            chat_agent = MessageReplyAgent()
        chat_responses = chat_agent.generate_message(
            context=main_agent.get_prompt_context(None),
            dialogue_history=main_agent.dialogue_history[-100:]
        )
        if not chat_responses:
            logger.warning(f"staff[{main_agent.staff_id}] user[{main_agent.user_id}]: no response generated")
            return []
        final_responses = []
        for chat_response in chat_responses:
            if response := main_agent.generate_multimodal_response(chat_response):
                final_responses.append(response)
            else:
                # 存在非法/结束消息，清空待发消息
                final_responses.clear()
        return final_responses

    def _call_chat_api(self, chat_config: Dict, chat_service_type: ChatServiceType) -> str:
        if configs.get().get('debug_flags', {}).get('disable_llm_api_call', False):
            return 'LLM模拟回复 {}'.format(datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
        if chat_service_type == ChatServiceType.OPENAI_COMPATIBLE:
            # 指定了LLM模型则优先使用指定模型
            if chat_config.get('model_name', None):
                llm_client = chat_service.OpenAICompatible.create_client(chat_config['model_name'])
                chat_completion = llm_client.chat.completions.create(
                    messages=chat_config['messages'],
                    model=chat_config['model_name'],
                )
            elif chat_config.get('use_multimodal_model', False):
                chat_completion = self.multimodal_model_client.chat.completions.create(
                    messages=chat_config['messages'],
                    model=self.multimodal_model_name,
                )
            else:
                chat_completion = self.text_model_client.chat.completions.create(
                    messages=chat_config['messages'],
                    model=self.text_model_name,
                )
            response = chat_completion.choices[0].message.content
        elif chat_service_type == ChatServiceType.COZE_CHAT:
            bot_user_id = 'qywx_{}'.format(chat_config['user_id'])
            response = self.coze_client.create(
                chat_config['bot_id'], bot_user_id, chat_config['messages'],
                chat_config['custom_variables']
            )
        else:
            raise Exception('Unsupported chat service type: {}'.format(self.chat_service_type))
        return response

    @staticmethod
    def sanitize_response(response: str):
        pattern = r'\[?\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}\]?'
        response = re.sub(pattern, '', response)
        response = response.strip()
        return response