317 lines
12 KiB
Python
317 lines
12 KiB
Python
|
"""人参果1号飞书长连接 — 固定路由到 AI学习助手 Agent（行为约束版，禁止主动消息）"""
|
|||
|
|
from __future__ import annotations
|
|||
|
|
|
|||
|
|
import asyncio
|
|||
|
|
import json
|
|||
|
|
import logging
|
|||
|
|
from collections import deque
|
|||
|
|
from typing import Optional
|
|||
|
|
|
|||
|
|
from app.core.config import settings
|
|||
|
|
|
|||
|
|
logger = logging.getLogger(__name__)
|
|||
|
|
|
|||
|
|
_processed_msg_ids: deque[str] = deque(maxlen=20)
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _get_message_id(data) -> Optional[str]:
|
|||
|
|
try:
|
|||
|
|
ev = data.event
|
|||
|
|
msg = getattr(ev, "message", None)
|
|||
|
|
if msg:
|
|||
|
|
return getattr(msg, "message_id", None)
|
|||
|
|
except Exception:
|
|||
|
|
return None
|
|||
|
|
return None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _get_message_text(data) -> Optional[str]:
|
|||
|
|
try:
|
|||
|
|
ev = data.event
|
|||
|
|
msg = getattr(ev, "message", None)
|
|||
|
|
if not msg:
|
|||
|
|
return None
|
|||
|
|
content_str = getattr(msg, "content", None)
|
|||
|
|
msg_type = getattr(msg, "message_type", "")
|
|||
|
|
if not content_str:
|
|||
|
|
return None
|
|||
|
|
if msg_type == "text":
|
|||
|
|
parsed = json.loads(content_str)
|
|||
|
|
return parsed.get("text", "")
|
|||
|
|
return None
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.warning("解析人参果1号消息内容失败: %s", e)
|
|||
|
|
return None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _get_sender_open_id(data) -> Optional[str]:
|
|||
|
|
try:
|
|||
|
|
ev = data.event
|
|||
|
|
sender = getattr(ev, "sender", None)
|
|||
|
|
if not sender:
|
|||
|
|
return None
|
|||
|
|
sender_id = getattr(sender, "sender_id", None)
|
|||
|
|
if not sender_id:
|
|||
|
|
return None
|
|||
|
|
return getattr(sender_id, "open_id", None)
|
|||
|
|
except Exception:
|
|||
|
|
return None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _get_sender_union_id(data) -> Optional[str]:
|
|||
|
|
try:
|
|||
|
|
ev = data.event
|
|||
|
|
sender = getattr(ev, "sender", None)
|
|||
|
|
if not sender:
|
|||
|
|
return None
|
|||
|
|
sender_id = getattr(sender, "sender_id", None)
|
|||
|
|
if not sender_id:
|
|||
|
|
return None
|
|||
|
|
return getattr(sender_id, "union_id", None)
|
|||
|
|
except Exception:
|
|||
|
|
return None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _get_chat_type(data) -> str:
|
|||
|
|
try:
|
|||
|
|
ev = data.event
|
|||
|
|
msg = getattr(ev, "message", None)
|
|||
|
|
return getattr(msg, "chat_type", "") if msg else ""
|
|||
|
|
except Exception:
|
|||
|
|
return ""
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _reply_to_feishu(open_id: str, text: str):
|
|||
|
|
try:
|
|||
|
|
from app.services.renshenguo2_app_service import send_plain_text
|
|||
|
|
send_plain_text(open_id, text)
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.warning("人参果1号回复消息失败: %s", e)
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _reply_card(open_id: str, title: str, content: str, status: str = "info"):
|
|||
|
|
try:
|
|||
|
|
from app.services.renshenguo2_app_service import send_message_to_user
|
|||
|
|
send_message_to_user(open_id, title, content, status=status)
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.warning("人参果1号回复卡片失败: %s", e)
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _make_llm_logger(db, agent_id: Optional[str] = None, user_id: Optional[str] = None):
|
|||
|
|
def _log(metrics: dict):
|
|||
|
|
try:
|
|||
|
|
from app.models.agent_llm_log import AgentLLMLog
|
|||
|
|
log = AgentLLMLog(
|
|||
|
|
agent_id=agent_id, session_id=metrics.get("session_id"),
|
|||
|
|
user_id=user_id, model=metrics.get("model", ""),
|
|||
|
|
provider=metrics.get("provider"),
|
|||
|
|
prompt_tokens=metrics.get("prompt_tokens", 0),
|
|||
|
|
completion_tokens=metrics.get("completion_tokens", 0),
|
|||
|
|
total_tokens=metrics.get("total_tokens", 0),
|
|||
|
|
latency_ms=metrics.get("latency_ms", 0),
|
|||
|
|
iteration_number=metrics.get("iteration_number", 0),
|
|||
|
|
step_type=metrics.get("step_type"),
|
|||
|
|
tool_name=metrics.get("tool_name"),
|
|||
|
|
status=metrics.get("status", "success"),
|
|||
|
|
error_message=metrics.get("error_message"),
|
|||
|
|
)
|
|||
|
|
db.add(log)
|
|||
|
|
db.commit()
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.warning("写入 AgentLLMLog 失败: %s", e)
|
|||
|
|
return _log
|
|||
|
|
|
|||
|
|
|
|||
|
|
async def _handle_message_async(data):
|
|||
|
|
open_id = _get_sender_open_id(data)
|
|||
|
|
union_id = _get_sender_union_id(data)
|
|||
|
|
chat_type = _get_chat_type(data)
|
|||
|
|
text = _get_message_text(data)
|
|||
|
|
|
|||
|
|
if not open_id or chat_type != "p2p":
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
logger.info("人参果1号收到消息: open_id=%s text=%s", open_id[:20], text[:50] if text else "(空)")
|
|||
|
|
|
|||
|
|
if not text:
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
from sqlalchemy.orm import Session
|
|||
|
|
from app.core.database import SessionLocal
|
|||
|
|
from app.models.agent import Agent
|
|||
|
|
from app.services.feishu_open_id_service import resolve_user_and_save
|
|||
|
|
|
|||
|
|
db: Optional[Session] = None
|
|||
|
|
try:
|
|||
|
|
db = SessionLocal()
|
|||
|
|
|
|||
|
|
# 自动保存/关联此应用的 open_id（跨应用识别）
|
|||
|
|
resolved_uid = resolve_user_and_save(
|
|||
|
|
db, app_id=settings.RENSHENGUO2_APP_ID or "",
|
|||
|
|
open_id=open_id, union_id=union_id,
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
agent_id = settings.RENSHENGUO2_AGENT_ID
|
|||
|
|
if not agent_id:
|
|||
|
|
_reply_to_feishu(open_id, "人参果1号尚未配置，请联系管理员。")
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
agent = db.query(Agent).filter(Agent.id == agent_id).first()
|
|||
|
|
if not agent:
|
|||
|
|
_reply_to_feishu(open_id, "人参果1号 Agent 已不存在，请联系管理员。")
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
_reply_to_feishu(open_id, "正在思考，请稍候...")
|
|||
|
|
|
|||
|
|
from app.agent_runtime import AgentRuntime, AgentConfig, AgentLLMConfig, AgentToolConfig, AgentMemoryConfig
|
|||
|
|
|
|||
|
|
wc = agent.workflow_config or {}
|
|||
|
|
nodes = wc.get("nodes", [])
|
|||
|
|
system_prompt = agent.description or ""
|
|||
|
|
model = "deepseek-v4-flash"
|
|||
|
|
provider = "deepseek"
|
|||
|
|
temperature = 0.7
|
|||
|
|
max_iterations = 15
|
|||
|
|
tools_whitelist = []
|
|||
|
|
|
|||
|
|
for n in nodes:
|
|||
|
|
if n.get("type") not in ("agent", "llm", "template"):
|
|||
|
|
continue
|
|||
|
|
cfg = n.get("data", {}) if isinstance(n, dict) else getattr(n, "data", {})
|
|||
|
|
system_prompt = cfg.get("system_prompt", "") or system_prompt
|
|||
|
|
model = cfg.get("model", model)
|
|||
|
|
provider = cfg.get("provider", provider)
|
|||
|
|
temperature = float(cfg.get("temperature", temperature))
|
|||
|
|
max_iterations = int(cfg.get("max_iterations", max_iterations))
|
|||
|
|
tools_whitelist = cfg.get("tools", tools_whitelist)
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 人参果1号：在 system_prompt 末尾强制注入行为约束
|
|||
|
|
# 无论 DB 中的 system_prompt 如何配置，这些规则始终生效
|
|||
|
|
behavior_rules = (
|
|||
|
|
f"\n\n## 系统信息\n"
|
|||
|
|
f"你的 Agent ID 是: {agent.id}\n"
|
|||
|
|
f"在调用 schedule_list、schedule_delete 等工具时，使用此 ID 作为 agent_id 参数。"
|
|||
|
|
f"\n\n## 行为约束（必须严格遵守，违反将导致严重问题）\n"
|
|||
|
|
f"1. **被动响应原则**：仅在用户主动发送消息时回复。不得根据时间流逝、用户在线状态或任何隐式信号主动发起对话。用户不说话，你就别说话。\n"
|
|||
|
|
f"2. **单轮回复限制**：对于用户的每条消息，只回复一次。回复完毕后必须停止，不得继续追问、猜测用户状态或发送额外消息。\n"
|
|||
|
|
f"3. **禁止重复关心**：如果上一轮已经表达过关心、提醒或催促（如催睡觉、催休息），不得在用户没有新输入的情况下重复类似内容。\n"
|
|||
|
|
f"4. **禁止猜测用户状态**：不得主动猜测用户的行为、情绪或状态（如\"你是不是睡不着\"、\"你是不是又回来了\"、\"让我猜猜你现在\"）。只回应用户明确提出的问题或话题。\n"
|
|||
|
|
f"5. **简洁直接**：回答应简洁、直接、切题。避免过度的寒暄、emoji堆砌、表格罗列猜测和冗余内容。\n"
|
|||
|
|
f"6. **上下文中立**：记忆系统提供的历史信息仅供理解上下文使用，不得用于主动发起话题、回顾\"上次你如何如何\"或过度个性化回应。\n"
|
|||
|
|
f"7. **禁止自言自语**：一次回复结束后，除非用户再次发送消息，否则绝对不要产生任何新输出。"
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
config = AgentConfig(
|
|||
|
|
name=agent.name or "人参果1号",
|
|||
|
|
system_prompt=system_prompt + behavior_rules,
|
|||
|
|
llm=AgentLLMConfig(
|
|||
|
|
model=model, provider=provider,
|
|||
|
|
temperature=temperature, max_iterations=max_iterations,
|
|||
|
|
),
|
|||
|
|
tools=AgentToolConfig(include_tools=tools_whitelist),
|
|||
|
|
memory=AgentMemoryConfig(
|
|||
|
|
max_history_messages=int(cfg.get("memory_max_history", 40)),
|
|||
|
|
vector_memory_top_k=int(cfg.get("memory_vector_top_k", 10)),
|
|||
|
|
persist_to_db=bool(cfg.get("memory_persist", True)),
|
|||
|
|
vector_memory_enabled=bool(cfg.get("memory_vector_enabled", True)),
|
|||
|
|
learning_enabled=bool(cfg.get("memory_learning", True)),
|
|||
|
|
),
|
|||
|
|
user_id=resolved_uid,
|
|||
|
|
memory_scope_id=str(agent.id),
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
on_llm_call = _make_llm_logger(db, agent_id=str(agent.id))
|
|||
|
|
runtime = AgentRuntime(config=config, on_llm_call=on_llm_call)
|
|||
|
|
result = await runtime.run(text)
|
|||
|
|
|
|||
|
|
if result.content:
|
|||
|
|
_reply_card(open_id, f"{agent.name}", result.content.strip(), status="success")
|
|||
|
|
else:
|
|||
|
|
_reply_to_feishu(open_id, "Agent 未返回有效回复，请重试。")
|
|||
|
|
|
|||
|
|
logger.info(
|
|||
|
|
"人参果1号 Agent 回复完成: open_id=%s agent=%s iterations=%d tools=%d",
|
|||
|
|
open_id[:20], agent.name, result.iterations_used, result.tool_calls_made,
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.error("人参果1号消息处理失败: %s", e)
|
|||
|
|
try:
|
|||
|
|
_reply_to_feishu(open_id, f"处理失败: {e!s}")
|
|||
|
|
except Exception:
|
|||
|
|
pass
|
|||
|
|
finally:
|
|||
|
|
if db:
|
|||
|
|
db.close()
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _handle_message_internal(data):
|
|||
|
|
msg_id = _get_message_id(data)
|
|||
|
|
if msg_id:
|
|||
|
|
if msg_id in _processed_msg_ids:
|
|||
|
|
return
|
|||
|
|
_processed_msg_ids.append(msg_id)
|
|||
|
|
|
|||
|
|
open_id = _get_sender_open_id(data)
|
|||
|
|
chat_type = _get_chat_type(data)
|
|||
|
|
text = _get_message_text(data)
|
|||
|
|
|
|||
|
|
if not open_id or chat_type != "p2p" or not text:
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
loop = asyncio.get_event_loop()
|
|||
|
|
if loop.is_running():
|
|||
|
|
asyncio.ensure_future(_handle_message_async(data))
|
|||
|
|
else:
|
|||
|
|
loop.run_until_complete(_handle_message_async(data))
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.error("人参果1号创建消息处理任务失败: %s", e)
|
|||
|
|
try:
|
|||
|
|
_reply_to_feishu(open_id, f"处理失败: {e!s}")
|
|||
|
|
except Exception:
|
|||
|
|
pass
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _build_event_handler():
|
|||
|
|
from lark_oapi.event.dispatcher_handler import EventDispatcherHandler
|
|||
|
|
|
|||
|
|
def on_message_receive(data):
|
|||
|
|
_handle_message_internal(data)
|
|||
|
|
|
|||
|
|
builder = EventDispatcherHandler.builder(encrypt_key="", verification_token="")
|
|||
|
|
builder.register_p2_im_message_receive_v1(on_message_receive)
|
|||
|
|
return builder.build()
|
|||
|
|
|
|||
|
|
|
|||
|
|
async def start_ws_client():
|
|||
|
|
if not settings.RENSHENGUO2_APP_ID or not settings.RENSHENGUO2_APP_SECRET:
|
|||
|
|
logger.warning("人参果1号应用未配置，跳过人参果1号长连接启动")
|
|||
|
|
return
|
|||
|
|
|
|||
|
|
from lark_oapi.ws import Client as WSClient
|
|||
|
|
|
|||
|
|
handler = _build_event_handler()
|
|||
|
|
client = WSClient(
|
|||
|
|
app_id=settings.RENSHENGUO2_APP_ID,
|
|||
|
|
app_secret=settings.RENSHENGUO2_APP_SECRET,
|
|||
|
|
event_handler=handler,
|
|||
|
|
auto_reconnect=True,
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
logger.info("人参果1号长连接客户端启动中...")
|
|||
|
|
|
|||
|
|
while True:
|
|||
|
|
try:
|
|||
|
|
await client._connect()
|
|||
|
|
logger.info("人参果1号长连接已建立")
|
|||
|
|
asyncio.ensure_future(client._ping_loop())
|
|||
|
|
while True:
|
|||
|
|
await asyncio.sleep(3600)
|
|||
|
|
except asyncio.CancelledError:
|
|||
|
|
break
|
|||
|
|
except Exception as e:
|
|||
|
|
logger.warning("人参果1号长连接断开，3秒后重连: %s", e)
|
|||
|
|
await asyncio.sleep(3)
|