Merge branch 'SengokuCola:debug' into debug

docs/manual_deploy.md

@@ -26,7 +26,25 @@
 
 ## 如果准备好了，就可以开始部署了
 
-### 1️⃣ **我们需要创建一个Python环境来运行程序**
+### 1️⃣ **首先，我们需要安装正确版本的Python**
+
+在创建虚拟环境之前，请确保你的电脑上安装了Python 3.9及以上版本。如果没有，可以按以下步骤安装：
+
+1. 访问Python官网下载页面：https://www.python.org/downloads/release/python-3913/
+2. 下载Windows安装程序 (64-bit): `python-3.9.13-amd64.exe`
+3. 运行安装程序，并确保勾选"Add Python 3.9 to PATH"选项
+4. 点击"Install Now"开始安装
+
+或者使用PowerShell自动下载安装（需要管理员权限）：
+```powershell
+# 下载并安装Python 3.9.13
+$pythonUrl = "https://www.python.org/ftp/python/3.9.13/python-3.9.13-amd64.exe"
+$pythonInstaller = "$env:TEMP\python-3.9.13-amd64.exe"
+Invoke-WebRequest -Uri $pythonUrl -OutFile $pythonInstaller
+Start-Process -Wait -FilePath $pythonInstaller -ArgumentList "/quiet", "InstallAllUsers=0", "PrependPath=1" -Verb RunAs
+```
+
+### 2️⃣ **创建Python虚拟环境来运行程序**
 
     你可以选择使用以下两种方法之一来创建Python环境：
 

run_maimai.bat

@@ -1,5 +1,5 @@
 chcp 65001
-call conda activate niuniu
+call conda activate maimbot
 cd .
 
 REM 执行nb run命令

src/plugins/chat/config.py

@@ -148,15 +148,6 @@ class BotConfig:
                 if "rerank" in model_config:
                     config.rerank = model_config["rerank"]
                 
-            if 'topic' in toml_dict:
-                topic_config=toml_dict['topic']
-                if 'topic_extract' in topic_config:
-                    config.topic_extract=topic_config.get('topic_extract',config.topic_extract)
-                    logger.info(f"载入自定义主题提取为{config.topic_extract}")
-                if config.topic_extract=='llm' and 'llm_topic' in topic_config:
-                    config.llm_topic_extract=topic_config['llm_topic']
-                    logger.info(f"载入自定义主题提取模型为{config.llm_topic_extract['name']}")
-                
             # 消息配置
             if "message" in toml_dict:
                 msg_config = toml_dict["message"]
@@ -190,13 +181,13 @@ class BotConfig:
 
 bot_config_floder_path = BotConfig.get_config_dir()
 print(f"正在品鉴配置文件目录: {bot_config_floder_path}")
-bot_config_path = os.path.join(bot_config_floder_path, "bot_config_dev.toml")
-if not os.path.exists(bot_config_path):
-    # 如果开发环境配置文件不存在，则使用默认配置文件
 bot_config_path = os.path.join(bot_config_floder_path, "bot_config.toml")
+if os.path.exists(bot_config_path):
+    # 如果开发环境配置文件不存在，则使用默认配置文件
+    print(f"异常的新鲜，异常的美味: {bot_config_path}")
     logger.info("使用bot配置文件")
 else:
-    logger.info("已找到开发bot配置文件")
+    logger.info("没有找到美味")
 
 global_config = BotConfig.load_config(config_path=bot_config_path)
 

src/plugins/chat/topic_identifier.py

@@ -12,8 +12,6 @@ config = driver.config
 class TopicIdentifier:
     def __init__(self):
         self.llm_client = LLM_request(model=global_config.llm_topic_extract)
-        self.select=global_config.topic_extract
-
 
     async def identify_topic_llm(self, text: str) -> Optional[List[str]]:
         """识别消息主题，返回主题列表"""

src/plugins/memory_system/memory.py

@@ -578,14 +578,11 @@ class Hippocampus:
 
     async def memory_activate_value(self, text: str, max_topics: int = 5, similarity_threshold: float = 0.3) -> int:
         """计算输入文本对记忆的激活程度"""
-        print(f"\033[1;32m[记忆激活]\033[0m 开始计算文本的记忆激活度: {text}")
+        print(f"\033[1;32m[记忆激活]\033[0m 识别主题: {await self._identify_topics(text)}")
         
         # 识别主题
         identified_topics = await self._identify_topics(text)
-        print(f"\033[1;32m[记忆激活]\033[0m 识别出的主题: {identified_topics}")
-        
         if not identified_topics:
-            # print(f"\033[1;32m[记忆激活]\033[0m 未识别出主题,返回0")
             return 0
             
         # 查找相似主题
@@ -596,7 +593,6 @@ class Hippocampus:
         )
         
         if not all_similar_topics:
-            print(f"\033[1;32m[记忆激活]\033[0m 未找到相似主题,返回0")
             return 0
             
         # 获取最相关的主题
@@ -605,19 +601,29 @@ class Hippocampus:
         # 如果只找到一个主题，进行惩罚
         if len(top_topics) == 1:
             topic, score = top_topics[0]
-            activation = int(score * 50)  # 单主题情况下，直接用相似度*50作为激活值
+            # 获取主题内容数量并计算惩罚系数
-            print(f"\033[1;32m[记忆激活]\033[0m 只找到一个主题，进行惩罚:")
+            memory_items = self.memory_graph.G.nodes[topic].get('memory_items', [])
-            print(f"\033[1;32m[记忆激活]\033[0m - 主题: {topic}")
+            if not isinstance(memory_items, list):
-            print(f"\033[1;32m[记忆激活]\033[0m - 相似度: {score:.3f}")
+                memory_items = [memory_items] if memory_items else []
-            print(f"\033[1;32m[记忆激活]\033[0m - 最终激活值: {activation}")
+            content_count = len(memory_items)
+            penalty = 1.0 / (1 + math.log(content_count + 1))
+            
+            activation = int(score * 50 * penalty)
+            print(f"\033[1;32m[记忆激活]\033[0m 单主题「{topic}」- 相似度: {score:.3f}, 内容数: {content_count}, 激活值: {activation}")
             return activation
             
-        # 计算关键词匹配率
+        # 计算关键词匹配率，同时考虑内容数量
         matched_topics = set()
         topic_similarities = {}
         
-        print(f"\033[1;32m[记忆激活]\033[0m 计算关键词匹配情况:")
         for memory_topic, similarity in top_topics:
+            # 计算内容数量惩罚
+            memory_items = self.memory_graph.G.nodes[memory_topic].get('memory_items', [])
+            if not isinstance(memory_items, list):
+                memory_items = [memory_items] if memory_items else []
+            content_count = len(memory_items)
+            penalty = 1.0 / (1 + math.log(content_count + 1))
+            
             # 对每个记忆主题，检查它与哪些输入主题相似
             for input_topic in identified_topics:
                 topic_vector = text_to_vector(input_topic)
@@ -628,27 +634,19 @@ class Hippocampus:
                 sim = cosine_similarity(v1, v2)
                 if sim >= similarity_threshold:
                     matched_topics.add(input_topic)
-                    topic_similarities[input_topic] = max(topic_similarities.get(input_topic, 0), sim)
+                    adjusted_sim = sim * penalty
-                    print(f"\033[1;32m[记忆激活]\033[0m - 输入主题「{input_topic}」匹配到记忆「{memory_topic}」, 相似度: {sim:.3f}")
+                    topic_similarities[input_topic] = max(topic_similarities.get(input_topic, 0), adjusted_sim)
+                    print(f"\033[1;32m[记忆激活]\033[0m 主题「{input_topic}」-> 「{memory_topic}」(内容数: {content_count}, 相似度: {adjusted_sim:.3f})")
         
-        # 计算主题匹配率
+        # 计算主题匹配率和平均相似度
         topic_match = len(matched_topics) / len(identified_topics)
-        print(f"\033[1;32m[记忆激活]\033[0m 主题匹配率:")
-        print(f"\033[1;32m[记忆激活]\033[0m - 匹配主题数: {len(matched_topics)}")
-        print(f"\033[1;32m[记忆激活]\033[0m - 总主题数: {len(identified_topics)}")
-        print(f"\033[1;32m[记忆激活]\033[0m - 匹配率: {topic_match:.3f}")
-        
-        # 计算匹配主题的平均相似度
         average_similarities = sum(topic_similarities.values()) / len(topic_similarities) if topic_similarities else 0
-        print(f"\033[1;32m[记忆激活]\033[0m 平均相似度:")
-        print(f"\033[1;32m[记忆激活]\033[0m - 各主题相似度: {[f'{k}:{v:.3f}' for k,v in topic_similarities.items()]}")
-        print(f"\033[1;32m[记忆激活]\033[0m - 平均相似度: {average_similarities:.3f}")
         
         # 计算最终激活值
-        activation = (topic_match + average_similarities) / 2 * 100
+        activation = int((topic_match + average_similarities) / 2 * 100)
-        print(f"\033[1;32m[记忆激活]\033[0m 最终激活值: {int(activation)}")
+        print(f"\033[1;32m[记忆激活]\033[0m 匹配率: {topic_match:.3f}, 平均相似度: {average_similarities:.3f}, 激活值: {activation}")
         
-        return int(activation)
+        return activation
 
     async def get_relevant_memories(self, text: str, max_topics: int = 5, similarity_threshold: float = 0.4, max_memory_num: int = 5) -> list:
         """根据输入文本获取相关的记忆内容"""