feat(affinity-flow): 重构兴趣度评分系统为智能embedding匹配

- 移除传统关键词匹配方式，改用embedding计算智能兴趣匹配度
- 添加异步方法支持机器人兴趣管理器的智能匹配计算
- 增加详细的日志记录和错误处理机制
- 添加数据库关键词提取和降级处理逻辑
- 集成智能兴趣系统初始化到人设构建流程
- 防止回复自身消息的死循环保护机制

BREAKING CHANGE: 兴趣匹配评分机制完全重构，从基于关键词的硬编码匹配改为基于embedding的智能匹配，需要重新初始化兴趣系统

src/chat/planner_actions/plan_executor.py

@@ -6,6 +6,7 @@ import asyncio
 import time
 from typing import Dict, List
 
+from src.config.config import global_config
 from src.chat.planner_actions.action_manager import ActionManager
 from src.common.data_models.info_data_model import Plan, ActionPlannerInfo
 from src.common.logger import get_logger
@@ -122,6 +123,16 @@ class PlanExecutor:
         try:
             logger.info(f"执行回复动作: {action_info.action_type}, 原因: {action_info.reasoning}")
 
+            if action_info.action_message.get("user_id","") == str(global_config.bot.qq_account):
+                logger.warning("尝试回复自己，跳过此动作以防止死循环。")
+                return {
+                    "action_type": action_info.action_type,
+                    "success": False,
+                    "error_message": "尝试回复自己，跳过此动作以防止死循环。",
+                    "execution_time": 0,
+                    "reasoning": action_info.reasoning,
+                    "reply_content": "",
+                }
             # 构建回复动作参数
             action_params = {
                 "chat_id": plan.chat_id,

src/chat/planner_actions/planner.py

@@ -97,7 +97,7 @@ class ActionPlanner:
             # 2. 兴趣度评分 - 只对未读消息进行评分
             if unread_messages:
                 bot_nickname = global_config.bot.nickname
-                interest_scores = self.interest_scoring.calculate_interest_scores(
+                interest_scores = await self.interest_scoring.calculate_interest_scores(
                     unread_messages, bot_nickname
                 )
 
@@ -175,33 +175,14 @@ class ActionPlanner:
 
         return final_actions_dict, final_target_message_dict
 
-    def _build_return_result(self, plan: Plan) -> Tuple[List[Dict], Optional[Dict]]:
-        """构建返回结果"""
-        final_actions = plan.decided_actions or []
-        final_target_message = next(
-            (act.action_message for act in final_actions if act.action_message), None
-        )
-
-        final_actions_dict = [asdict(act) for act in final_actions]
-
-        if final_target_message:
-            if hasattr(final_target_message, '__dataclass_fields__'):
-                final_target_message_dict = asdict(final_target_message)
-            else:
-                final_target_message_dict = final_target_message
-        else:
-            final_target_message_dict = None
-
-        return final_actions_dict, final_target_message_dict
-
     def get_user_relationship(self, user_id: str) -> float:
         """获取用户关系分"""
         return self.interest_scoring.get_user_relationship(user_id)
 
     def update_interest_keywords(self, new_keywords: Dict[str, List[str]]):
-        """更新兴趣关键词"""
-        self.interest_scoring.interest_keywords.update(new_keywords)
-        logger.info(f"已更新兴趣关键词: {list(new_keywords.keys())}")
+        """更新兴趣关键词（已弃用，仅保留用于兼容性）"""
+        logger.info("传统关键词匹配已移除，此方法仅保留用于兼容性")
+        # 此方法已弃用，因为现在完全使用embedding匹配
 
     def get_planner_stats(self) -> Dict[str, any]:
         """获取规划器统计"""
@@ -226,5 +207,4 @@ class ActionPlanner:
         }
 
 
-# 全局兴趣度评分系统实例
-interest_scoring_system = InterestScoringSystem()
+# 全局兴趣度评分系统实例 - 在 individuality 模块中创建