From 30f2eac278b34e02e40eb304ba4d269aded086fd Mon Sep 17 00:00:00 2001 From: SengokuCola <1026294844@qq.com> Date: Sat, 14 Jun 2025 21:55:16 +0800 Subject: [PATCH] =?UTF-8?q?fix=EF=BC=9B=E8=B0=83=E6=95=B4=E6=A6=82?= =?UTF-8?q?=E7=8E=87=E5=92=8CLog=E3=80=81?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- src/chat/knowledge/embedding_store.py | 2 +- src/chat/knowledge/qa_manager.py | 2 +- src/chat/memory_system/Hippocampus.py | 4 +- src/chat/normal_chat/normal_chat.py | 154 +++----------------- src/chat/normal_chat/normal_chat_planner.py | 8 +- 5 files changed, 27 insertions(+), 143 deletions(-) diff --git a/src/chat/knowledge/embedding_store.py b/src/chat/knowledge/embedding_store.py index 90987576b..1609158aa 100644 --- a/src/chat/knowledge/embedding_store.py +++ b/src/chat/knowledge/embedding_store.py @@ -267,7 +267,7 @@ class EmbeddingStore: result: 最相似的k个项的(hash, 余弦相似度)列表 """ if self.faiss_index is None: - logger.warning("FaissIndex尚未构建,返回None") + logger.debug("FaissIndex尚未构建,返回None") return None if self.idx2hash is None: logger.warning("idx2hash尚未构建,返回None") diff --git a/src/chat/knowledge/qa_manager.py b/src/chat/knowledge/qa_manager.py index b6bbd1207..01a3e82d3 100644 --- a/src/chat/knowledge/qa_manager.py +++ b/src/chat/knowledge/qa_manager.py @@ -121,5 +121,5 @@ class QAManager: found_knowledge = found_knowledge[:MAX_KNOWLEDGE_LENGTH] + "\n" return found_knowledge else: - logger.info("LPMM知识库并未初始化，可能是从未导入过知识...") + logger.debug("LPMM知识库并未初始化，可能是从未导入过知识...") return None diff --git a/src/chat/memory_system/Hippocampus.py b/src/chat/memory_system/Hippocampus.py index 457b00de8..bd8a171f4 100644 --- a/src/chat/memory_system/Hippocampus.py +++ b/src/chat/memory_system/Hippocampus.py @@ -366,7 +366,7 @@ class Hippocampus: # 过滤掉不存在于记忆图中的关键词 valid_keywords = [keyword for keyword in keywords if keyword in self.memory_graph.G] if not valid_keywords: - logger.info("没有找到有效的关键词节点") + logger.debug("没有找到有效的关键词节点") return [] logger.debug(f"有效的关键词: {', '.join(valid_keywords)}") @@ -537,7 +537,7 @@ class Hippocampus: # 过滤掉不存在于记忆图中的关键词 valid_keywords = [keyword for keyword in keywords if keyword in self.memory_graph.G] if not valid_keywords: - logger.info("没有找到有效的关键词节点") + logger.debug("没有找到有效的关键词节点") return [] logger.debug(f"有效的关键词: {', '.join(valid_keywords)}") diff --git a/src/chat/normal_chat/normal_chat.py b/src/chat/normal_chat/normal_chat.py index c616d3cc6..c8cff24a1 100644 --- a/src/chat/normal_chat/normal_chat.py +++ b/src/chat/normal_chat/normal_chat.py @@ -587,14 +587,14 @@ class NormalChat: if differ > 0.1: mapped = 1 + (differ - 0.1) * 4 / 0.9 mapped = max(1, min(5, mapped)) - logger.info( + logger.debug( f"[{self.stream_name}] 回复频率低于{global_config.normal_chat.talk_frequency}，增加回复概率，differ={differ:.3f}，映射值={mapped:.2f}" ) self.willing_amplifier += mapped * 0.1 # 你可以根据实际需要调整系数 elif differ < -0.1: mapped = 1 - (differ + 0.1) * 4 / 0.9 mapped = max(1, min(5, mapped)) - logger.info( + logger.debug( f"[{self.stream_name}] 回复频率高于{global_config.normal_chat.talk_frequency}，减少回复概率，differ={differ:.3f}，映射值={mapped:.2f}" ) self.willing_amplifier -= mapped * 0.1 @@ -689,143 +689,20 @@ class NormalChat: self.engaging_persons[person_id]["last_time"] = current_time logger.debug(f"[{self.stream_name}] 用户 {person_id} 消息次数更新: {self.engaging_persons[person_id]['receive_count']}") - def get_engaging_persons(self) -> dict: - """获取所有engaging_persons统计信息 - - Returns: - dict: person_id -> {first_time, last_time, receive_count, reply_count} - """ - return self.engaging_persons.copy() - - def get_engaging_person_stats(self, person_id: str) -> dict: - """获取特定用户的统计信息 - - Args: - person_id: 用户ID - - Returns: - dict: 用户统计信息，如果用户不存在则返回None - """ - return self.engaging_persons.get(person_id) - - def get_top_engaging_persons(self, limit: int = 10, sort_by: str = "receive_count") -> list: - """获取最活跃的用户列表 - - Args: - limit: 返回的用户数量限制 - sort_by: 排序依据，可选值: "receive_count", "reply_count", "last_time" - - Returns: - list: 按指定条件排序的用户列表 - """ - if sort_by not in ["receive_count", "reply_count", "last_time"]: - sort_by = "receive_count" - - sorted_persons = sorted( - self.engaging_persons.items(), - key=lambda x: x[1][sort_by], - reverse=True - ) - - return sorted_persons[:limit] - - def clear_engaging_persons_stats(self): - """清空engaging_persons统计信息""" - self.engaging_persons.clear() - logger.info(f"[{self.stream_name}] 已清空engaging_persons统计信息") - - def get_relation_building_stats(self) -> dict: - """获取关系构建相关统计信息 - - Returns: - dict: 关系构建统计信息 - """ - total_persons = len(self.engaging_persons) - relation_built_count = sum(1 for stats in self.engaging_persons.values() - if stats.get("relation_built", False)) - pending_persons = [] - - current_time = time.time() - for person_id, stats in self.engaging_persons.items(): - if not stats.get("relation_built", False): - time_elapsed = current_time - stats["first_time"] - total_messages = self._get_total_messages_in_timerange( - stats["first_time"], stats["last_time"] - ) - - # 检查是否接近满足条件 - progress_info = { - "person_id": person_id, - "time_elapsed": time_elapsed, - "total_messages": total_messages, - "receive_count": stats["receive_count"], - "reply_count": stats["reply_count"], - "progress": { - "50_messages": f"{total_messages}/50 ({total_messages/50*100:.1f}%)", - "35_msg_10min": f"{total_messages}/35 + {time_elapsed}/600s", - "25_msg_30min": f"{total_messages}/25 + {time_elapsed}/1800s", - "10_msg_1hour": f"{total_messages}/10 + {time_elapsed}/3600s" - } - } - pending_persons.append(progress_info) - - return { - "total_persons": total_persons, - "relation_built_count": relation_built_count, - "pending_count": len(pending_persons), - "pending_persons": pending_persons - } - - def get_engaging_persons_summary(self) -> dict: - """获取engaging_persons统计摘要 - - Returns: - dict: 包含总用户数、总消息数、总回复数等统计信息 - """ - if not self.engaging_persons: - return { - "total_persons": 0, - "total_messages": 0, - "total_replies": 0, - "most_active_person": None, - "most_replied_person": None - } - - total_messages = sum(stats["receive_count"] for stats in self.engaging_persons.values()) - total_replies = sum(stats["reply_count"] for stats in self.engaging_persons.values()) - - most_active = max(self.engaging_persons.items(), key=lambda x: x[1]["receive_count"]) - most_replied = max(self.engaging_persons.items(), key=lambda x: x[1]["reply_count"]) - - return { - "total_persons": len(self.engaging_persons), - "total_messages": total_messages, - "total_replies": total_replies, - "most_active_person": { - "person_id": most_active[0], - "message_count": most_active[1]["receive_count"] - }, - "most_replied_person": { - "person_id": most_replied[0], - "reply_count": most_replied[1]["reply_count"] - } - } async def _check_relation_building_conditions(self): """检查engaging_persons中是否有满足关系构建条件的用户""" current_time = time.time() for person_id, stats in list(self.engaging_persons.items()): - # 跳过已经进行过关系构建的用户 - if stats.get("relation_built", False): - continue - # 计算时间差和消息数量 time_elapsed = current_time - stats["first_time"] total_messages = self._get_total_messages_in_timerange( stats["first_time"], stats["last_time"] ) + print(f"person_id: {person_id}, total_messages: {total_messages}, time_elapsed: {time_elapsed}") + # 检查是否满足关系构建条件 should_build_relation = ( total_messages >= 50 # 50条消息必定满足 @@ -843,6 +720,10 @@ class NormalChat: # 计算构建概率并决定是否构建 await self._evaluate_and_build_relation(person_id, stats, total_messages) + + # 评估完成后移除该用户，重新开始统计 + del self.engaging_persons[person_id] + logger.info(f"[{self.stream_name}] 用户 {person_id} 评估完成，已移除记录，将重新开始统计") def _get_total_messages_in_timerange(self, start_time: float, end_time: float) -> int: @@ -856,24 +737,31 @@ class NormalChat: async def _evaluate_and_build_relation(self, person_id: str, stats: dict, total_messages: int): """评估并执行关系构建""" + import math + receive_count = stats["receive_count"] reply_count = stats["reply_count"] # 计算回复概率（reply_count在总消息中的比值） reply_ratio = reply_count / total_messages if total_messages > 0 else 0 - reply_build_probability = reply_ratio # 100%回复则100%构建 + # 使用对数函数让低比率时概率上升更快：log(1 + ratio * k) / log(1 + k) + # k=10时，0.1比率对应约0.67概率，0.5比率对应约0.95概率 + k_reply = 10 + reply_build_probability = math.log(1 + reply_ratio * k_reply) / math.log(1 + k_reply) if reply_ratio > 0 else 0 # 计算接收概率（receive_count的影响） receive_ratio = receive_count / total_messages if total_messages > 0 else 0 - receive_build_probability = receive_ratio * 0.25 # 100%接收则25%构建 + # 接收概率使用更温和的对数曲线，最大0.4 + k_receive = 8 + receive_build_probability = (math.log(1 + receive_ratio * k_receive) / math.log(1 + k_receive)) * 0.4 if receive_ratio > 0 else 0 # 取最高概率 final_probability = max(reply_build_probability, receive_build_probability) logger.info( f"[{self.stream_name}] 用户 {person_id} 关系构建概率评估：" - f"回复比例：{reply_ratio:.2f}({reply_build_probability:.2f})" - f"，接收比例：{receive_ratio:.2f}({receive_build_probability:.2f})" + f"回复比例：{reply_ratio:.2f}(对数概率:{reply_build_probability:.2f})" + f"，接收比例：{receive_ratio:.2f}(对数概率:{receive_build_probability:.2f})" f"，最终概率：{final_probability:.2f}" ) @@ -881,12 +769,8 @@ class NormalChat: if random() < final_probability: logger.info(f"[{self.stream_name}] 决定为用户 {person_id} 构建关系") await self._build_relation_for_person(person_id, stats) - # 标记已构建 - stats["relation_built"] = True else: logger.info(f"[{self.stream_name}] 用户 {person_id} 未通过关系构建概率判定") - # 即使未构建，也标记为已处理，避免重复判定 - stats["relation_built"] = True async def _build_relation_for_person(self, person_id: str, stats: dict): """为特定用户构建关系""" diff --git a/src/chat/normal_chat/normal_chat_planner.py b/src/chat/normal_chat/normal_chat_planner.py index 34517d955..581dcd82f 100644 --- a/src/chat/normal_chat/normal_chat_planner.py +++ b/src/chat/normal_chat/normal_chat_planner.py @@ -158,10 +158,10 @@ class NormalChatPlanner: try: content, (reasoning_content, model_name) = await self.planner_llm.generate_response_async(prompt) - logger.info(f"{self.log_prefix}规划器原始提示词: {prompt}") - logger.info(f"{self.log_prefix}规划器原始响应: {content}") - logger.info(f"{self.log_prefix}规划器推理: {reasoning_content}") - logger.info(f"{self.log_prefix}规划器模型: {model_name}") + logger.debug(f"{self.log_prefix}规划器原始提示词: {prompt}") + logger.debug(f"{self.log_prefix}规划器原始响应: {content}") + logger.debug(f"{self.log_prefix}规划器推理: {reasoning_content}") + logger.debug(f"{self.log_prefix}规划器模型: {model_name}") # 解析JSON响应 try: