v0.3.1 实装了记忆系统和自动发言

哈哈哈

src/plugins/memory_system/draw_memory.py

@@ -0,0 +1,264 @@
+# -*- coding: utf-8 -*-
+import sys
+import jieba
+from llm_module import LLMModel
+import networkx as nx
+import matplotlib.pyplot as plt
+import math
+from collections import Counter
+import datetime
+import random
+import time
+# from chat.config import global_config
+import sys
+sys.path.append("C:/GitHub/MaiMBot")  # 添加项目根目录到 Python 路径
+from src.common.database import Database  # 使用正确的导入语法
+   
+class Memory_graph:
+    def __init__(self):
+        self.G = nx.Graph()  # 使用 networkx 的图结构
+        self.db = Database.get_instance()
+        
+    def connect_dot(self, concept1, concept2):
+        self.G.add_edge(concept1, concept2)
+    
+    def add_dot(self, concept, memory):
+        if concept in self.G:
+            # 如果节点已存在，将新记忆添加到现有列表中
+            if 'memory_items' in self.G.nodes[concept]:
+                if not isinstance(self.G.nodes[concept]['memory_items'], list):
+                    # 如果当前不是列表，将其转换为列表
+                    self.G.nodes[concept]['memory_items'] = [self.G.nodes[concept]['memory_items']]
+                self.G.nodes[concept]['memory_items'].append(memory)
+            else:
+                self.G.nodes[concept]['memory_items'] = [memory]
+        else:
+            # 如果是新节点，创建新的记忆列表
+            self.G.add_node(concept, memory_items=[memory])
+        
+    def get_dot(self, concept):
+        # 检查节点是否存在于图中
+        if concept in self.G:
+            # 从图中获取节点数据
+            node_data = self.G.nodes[concept]
+            # print(node_data)
+            # 创建新的Memory_dot对象
+            return concept,node_data
+        return None
+
+    def get_related_item(self, topic, depth=1):
+        if topic not in self.G:
+            return [], []
+            
+        first_layer_items = []
+        second_layer_items = []
+        
+        # 获取相邻节点
+        neighbors = list(self.G.neighbors(topic))
+        # print(f"第一层: {topic}")
+        
+        # 获取当前节点的记忆项
+        node_data = self.get_dot(topic)
+        if node_data:
+            concept, data = node_data
+            if 'memory_items' in data:
+                memory_items = data['memory_items']
+                if isinstance(memory_items, list):
+                    first_layer_items.extend(memory_items)
+                else:
+                    first_layer_items.append(memory_items)
+        
+        # 只在depth=2时获取第二层记忆
+        if depth >= 2:
+            # 获取相邻节点的记忆项
+            for neighbor in neighbors:
+                # print(f"第二层: {neighbor}")
+                node_data = self.get_dot(neighbor)
+                if node_data:
+                    concept, data = node_data
+                    if 'memory_items' in data:
+                        memory_items = data['memory_items']
+                        if isinstance(memory_items, list):
+                            second_layer_items.extend(memory_items)
+                        else:
+                            second_layer_items.append(memory_items)
+        
+        return first_layer_items, second_layer_items
+    
+    def store_memory(self):
+        for node in self.G.nodes():
+            dot_data = {
+                "concept": node
+            }
+            self.db.db.store_memory_dots.insert_one(dot_data)
+    
+    @property
+    def dots(self):
+        # 返回所有节点对应的 Memory_dot 对象
+        return [self.get_dot(node) for node in self.G.nodes()]
+    
+    
+    def get_random_chat_from_db(self, length: int, timestamp: str):
+        # 从数据库中根据时间戳获取离其最近的聊天记录
+        chat_text = ''
+        closest_record = self.db.db.messages.find_one({"time": {"$lte": timestamp}}, sort=[('time', -1)])  # 调试输出
+        print(f"距离time最近的消息时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(int(closest_record['time'])))}")
+        
+        if closest_record:
+            closest_time = closest_record['time']
+            group_id = closest_record['group_id']  # 获取groupid
+            # 获取该时间戳之后的length条消息，且groupid相同
+            chat_record = list(self.db.db.messages.find({"time": {"$gt": closest_time}, "group_id": group_id}).sort('time', 1).limit(length))
+            for record in chat_record:
+                time_str = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(int(record['time'])))
+                chat_text += f'[{time_str}] {record["user_nickname"] or "用户" + str(record["user_id"])}: {record["processed_plain_text"]}\n'  # 添加发送者和时间信息
+            return chat_text
+        
+        return []  # 如果没有找到记录，返回空列表
+
+    def save_graph_to_db(self):
+        # 清空现有的图数据
+        self.db.db.graph_data.delete_many({})
+        # 保存节点
+        for node in self.G.nodes(data=True):
+            node_data = {
+                'concept': node[0],
+                'memory_items': node[1].get('memory_items', [])  # 默认为空列表
+            }
+            self.db.db.graph_data.nodes.insert_one(node_data)
+        # 保存边
+        for edge in self.G.edges():
+            edge_data = {
+                'source': edge[0],
+                'target': edge[1]
+            }
+            self.db.db.graph_data.edges.insert_one(edge_data)
+
+    def load_graph_from_db(self):
+        # 清空当前图
+        self.G.clear()
+        # 加载节点
+        nodes = self.db.db.graph_data.nodes.find()
+        for node in nodes:
+            memory_items = node.get('memory_items', [])
+            if not isinstance(memory_items, list):
+                memory_items = [memory_items] if memory_items else []
+            self.G.add_node(node['concept'], memory_items=memory_items)
+        # 加载边
+        edges = self.db.db.graph_data.edges.find()
+        for edge in edges:
+            self.G.add_edge(edge['source'], edge['target'])
+
+
+def main():
+    # 初始化数据库
+    Database.initialize(
+        "127.0.0.1",
+        27017,
+        "MegBot"
+    )
+    
+    memory_graph = Memory_graph()
+    # 创建LLM模型实例
+
+    memory_graph.load_graph_from_db()
+    # 展示两种不同的可视化方式
+    print("\n按连接数量着色的图谱：")
+    visualize_graph(memory_graph, color_by_memory=False)
+    
+    print("\n按记忆数量着色的图谱：")
+    visualize_graph(memory_graph, color_by_memory=True)
+    
+    # memory_graph.save_graph_to_db()
+    
+    while True:
+        query = input("请输入新的查询概念（输入'退出'以结束）：")
+        if query.lower() == '退出':
+            break
+        items_list = memory_graph.get_related_item(query)
+        if items_list:
+            # print(items_list)
+            for memory_item in items_list:
+                print(memory_item)
+        else:
+            print("未找到相关记忆。")
+            
+
+def segment_text(text):
+    seg_text = list(jieba.cut(text))
+    return seg_text    
+
+def find_topic(text, topic_num):
+    prompt = f'这是一段文字：{text}。请你从这段话中总结出{topic_num}个话题，帮我列出来，用逗号隔开，尽可能精简。只需要列举{topic_num}个话题就好，不要告诉我其他内容。'
+    return prompt
+
+def topic_what(text, topic):
+    prompt = f'这是一段文字：{text}。我想知道这记忆里有什么关于{topic}的话题，帮我总结成一句自然的话，可以包含时间和人物。只输出这句话就好'
+    return prompt
+
+def visualize_graph(memory_graph: Memory_graph, color_by_memory: bool = False):
+    # 设置中文字体
+    plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
+    plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
+    
+    G = memory_graph.G
+    
+    # 保存图到本地
+    nx.write_gml(G, "memory_graph.gml")  # 保存为 GML 格式
+
+    # 根据连接条数或记忆数量设置节点颜色
+    node_colors = []
+    nodes = list(G.nodes())  # 获取图中实际的节点列表
+    
+    if color_by_memory:
+        # 计算每个节点的记忆数量
+        memory_counts = []
+        for node in nodes:
+            memory_items = G.nodes[node].get('memory_items', [])
+            if isinstance(memory_items, list):
+                count = len(memory_items)
+            else:
+                count = 1 if memory_items else 0
+            memory_counts.append(count)
+        max_memories = max(memory_counts) if memory_counts else 1
+        
+        for count in memory_counts:
+            # 使用不同的颜色方案：红色表示记忆多，蓝色表示记忆少
+            if max_memories > 0:
+                intensity = min(1.0, count / max_memories)
+                color = (intensity, 0, 1.0 - intensity)  # 从蓝色渐变到红色
+            else:
+                color = (0, 0, 1)  # 如果没有记忆，则为蓝色
+            node_colors.append(color)
+    else:
+        # 使用原来的连接数量着色方案
+        max_degree = max(G.degree(), key=lambda x: x[1])[1] if G.degree() else 1
+        for node in nodes:
+            degree = G.degree(node)
+            if max_degree > 0:
+                red = min(1.0, degree / max_degree)
+                blue = 1.0 - red
+                color = (red, 0, blue)
+            else:
+                color = (0, 0, 1)
+            node_colors.append(color)
+    
+    # 绘制图形
+    plt.figure(figsize=(12, 8))
+    pos = nx.spring_layout(G, k=1, iterations=50)
+    nx.draw(G, pos, 
+           with_labels=True, 
+           node_color=node_colors,
+           node_size=2000,
+           font_size=10,
+           font_family='SimHei',
+           font_weight='bold')
+    
+    title = '记忆图谱可视化 - ' + ('按记忆数量着色' if color_by_memory else '按连接数量着色')
+    plt.title(title, fontsize=16, fontfamily='SimHei')
+    plt.show()
+
+if __name__ == "__main__":
+    main()
+
+    

src/plugins/memory_system/llm_module_memory_make.py

@@ -0,0 +1,82 @@
+import os
+import requests
+from dotenv import load_dotenv
+from typing import Tuple, Union
+import time
+from ..chat.config import BotConfig
+
+# 获取当前文件的绝对路径
+current_dir = os.path.dirname(os.path.abspath(__file__))
+root_dir = os.path.abspath(os.path.join(current_dir, '..', '..', '..'))
+env_path = os.path.join(root_dir, 'config', '.env')
+
+# 加载环境变量
+print(f"尝试从 {env_path} 加载环境变量配置")
+if os.path.exists(env_path):
+    load_dotenv(env_path)
+    print("成功加载环境变量配置")
+else:
+    print(f"环境变量配置文件不存在: {env_path}")
+
+class LLMModel:
+    # def __init__(self, model_name="deepseek-ai/DeepSeek-R1-Distill-Qwen-32B", **kwargs):
+    def __init__(self, model_name="Pro/deepseek-ai/DeepSeek-V3", **kwargs):
+        self.model_name = model_name
+        self.params = kwargs
+        self.api_key = os.getenv("SILICONFLOW_KEY")
+        self.base_url = os.getenv("SILICONFLOW_BASE_URL")
+        
+        if not self.api_key or not self.base_url:
+            raise ValueError("环境变量未正确加载：SILICONFLOW_KEY 或 SILICONFLOW_BASE_URL 未设置")
+            
+        print(f"API URL: {self.base_url}")  # 打印 base_url 用于调试
+
+    def generate_response(self, prompt: str) -> Tuple[str, str]:
+        """根据输入的提示生成模型的响应"""
+        headers = {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json"
+        }
+        
+        # 构建请求体
+        data = {
+            "model": self.model_name,
+            "messages": [{"role": "user", "content": prompt}],
+            "temperature": 0.5,
+            **self.params
+        }
+        
+        # 发送请求到完整的chat/completions端点
+        api_url = f"{self.base_url.rstrip('/')}/chat/completions"
+        
+        max_retries = 3
+        base_wait_time = 15  # 基础等待时间（秒）
+        
+        for retry in range(max_retries):
+            try:
+                response = requests.post(api_url, headers=headers, json=data)
+                
+                if response.status_code == 429:
+                    wait_time = base_wait_time * (2 ** retry)  # 指数退避
+                    print(f"遇到请求限制(429)，等待{wait_time}秒后重试...")
+                    time.sleep(wait_time)
+                    continue
+                    
+                response.raise_for_status()  # 检查其他响应状态
+                
+                result = response.json()
+                if "choices" in result and len(result["choices"]) > 0:
+                    content = result["choices"][0]["message"]["content"]
+                    reasoning_content = result["choices"][0]["message"].get("reasoning_content", "")
+                    return content, reasoning_content
+                return "没有返回结果", ""
+                
+            except requests.exceptions.RequestException as e:
+                if retry < max_retries - 1:  # 如果还有重试机会
+                    wait_time = base_wait_time * (2 ** retry)
+                    print(f"请求失败，等待{wait_time}秒后重试... 错误: {str(e)}")
+                    time.sleep(wait_time)
+                else:
+                    return f"请求失败: {str(e)}", ""
+        
+        return "达到最大重试次数，请求仍然失败", ""

src/plugins/memory_system/memory.py

@@ -1,5 +1,4 @@
 # -*- coding: utf-8 -*-
-import sys
 import jieba
 from .llm_module import LLMModel
 import networkx as nx
@@ -11,8 +10,8 @@ import random
 import time
 from ..chat.config import global_config
 import sys
-sys.path.append("C:/GitHub/MaiMBot")  # 添加项目根目录到 Python 路径
-from src.common.database import Database  # 使用正确的导入语法
+from ...common.database import Database # 使用正确的导入语法
+from ..chat.utils import calculate_information_content, get_cloest_chat_from_db
    
 class Memory_graph:
     def __init__(self):
@@ -85,54 +84,66 @@ class Memory_graph:
         
         return first_layer_items, second_layer_items
     
-    def store_memory(self):
-        for node in self.G.nodes():
-            dot_data = {
-                "concept": node
-            }
-            self.db.db.store_memory_dots.insert_one(dot_data)
-    
     @property
     def dots(self):
         # 返回所有节点对应的 Memory_dot 对象
         return [self.get_dot(node) for node in self.G.nodes()]
-    
-    
-    def get_random_chat_from_db(self, length: int, timestamp: str):
-        # 从数据库中根据时间戳获取离其最近的聊天记录
-        chat_text = ''
-        closest_record = self.db.db.messages.find_one({"time": {"$lte": timestamp}}, sort=[('time', -1)])  # 调试输出
-        print(f"距离time最近的消息时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(int(closest_record['time'])))}")
-        
-        if closest_record:
-            closest_time = closest_record['time']
-            group_id = closest_record['group_id']  # 获取groupid
-            # 获取该时间戳之后的length条消息，且groupid相同
-            chat_record = list(self.db.db.messages.find({"time": {"$gt": closest_time}, "group_id": group_id}).sort('time', 1).limit(length))
-            for record in chat_record:
-                time_str = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(int(record['time'])))
-                chat_text += f'[{time_str}] {record["user_nickname"] or "用户" + str(record["user_id"])}: {record["processed_plain_text"]}\n'  # 添加发送者和时间信息
-            return chat_text
-        
-        return []  # 如果没有找到记录，返回空列表
 
     def save_graph_to_db(self):
-        # 清空现有的图数据
-        self.db.db.graph_data.delete_many({})
         # 保存节点
         for node in self.G.nodes(data=True):
-            node_data = {
-                'concept': node[0],
-                'memory_items': node[1].get('memory_items', [])  # 默认为空列表
-            }
-            self.db.db.graph_data.nodes.insert_one(node_data)
+            concept = node[0]
+            memory_items = node[1].get('memory_items', [])
+            
+            # 查找是否存在同名节点
+            existing_node = self.db.db.graph_data.nodes.find_one({'concept': concept})
+            if existing_node:
+                # 如果存在,合并memory_items并去重
+                existing_items = existing_node.get('memory_items', [])
+                if not isinstance(existing_items, list):
+                    existing_items = [existing_items] if existing_items else []
+                
+                # 合并并去重
+                all_items = list(set(existing_items + memory_items))
+                
+                # 更新节点
+                self.db.db.graph_data.nodes.update_one(
+                    {'concept': concept},
+                    {'$set': {'memory_items': all_items}}
+                )
+            else:
+                # 如果不存在,创建新节点
+                node_data = {
+                    'concept': concept,
+                    'memory_items': memory_items
+                }
+                self.db.db.graph_data.nodes.insert_one(node_data)
+        
         # 保存边
         for edge in self.G.edges():
-            edge_data = {
-                'source': edge[0],
-                'target': edge[1]
-            }
-            self.db.db.graph_data.edges.insert_one(edge_data)
+            source, target = edge
+            
+            # 查找是否存在同样的边
+            existing_edge = self.db.db.graph_data.edges.find_one({
+                'source': source,
+                'target': target
+            })
+            
+            if existing_edge:
+                # 如果存在,增加num属性
+                num = existing_edge.get('num', 1) + 1
+                self.db.db.graph_data.edges.update_one(
+                    {'source': source, 'target': target},
+                    {'$set': {'num': num}}
+                )
+            else:
+                # 如果不存在,创建新边
+                edge_data = {
+                    'source': source,
+                    'target': target,
+                    'num': 1
+                }
+                self.db.db.graph_data.edges.insert_one(edge_data)
 
     def load_graph_from_db(self):
         # 清空当前图
@@ -147,150 +158,92 @@ class Memory_graph:
         # 加载边
         edges = self.db.db.graph_data.edges.find()
         for edge in edges:
-            self.G.add_edge(edge['source'], edge['target'])
-
-def calculate_information_content(text):
-    
-    """计算文本的信息量（熵）"""
-    # 统计字符频率
-    char_count = Counter(text)
-    total_chars = len(text)
-    
-    # 计算熵
-    entropy = 0
-    for count in char_count.values():
-        probability = count / total_chars
-        entropy -= probability * math.log2(probability)
-    
-    return entropy
-
-
-start_time = time.time()
-
-Database.initialize(
-    global_config.MONGODB_HOST,
-    global_config.MONGODB_PORT,
-    global_config.DATABASE_NAME
-)
-memory_graph = Memory_graph()
-
-llm_model = LLMModel()
-llm_model_small = LLMModel(model_name="deepseek-ai/DeepSeek-V2.5")
-
-memory_graph.load_graph_from_db()
-
-end_time = time.time()
-print(f"加载海马体耗时: {end_time - start_time:.2f} 秒")
+            self.G.add_edge(edge['source'], edge['target'], num=edge.get('num', 1))
 
 
 
-def main():
-    # 初始化数据库
-    Database.initialize(
-        "127.0.0.1",
-        27017,
-        "MegBot"
-    )
-    
-    memory_graph = Memory_graph()
-    # 创建LLM模型实例
-    llm_model = LLMModel()
-    llm_model_small = LLMModel(model_name="deepseek-ai/DeepSeek-V2.5")
-    
-    # 使用当前时间戳进行测试
-    current_timestamp = datetime.datetime.now().timestamp()
-    chat_text = []
-    
-    chat_size =40
-    
-    for _ in range(100):  # 循环10次
-        random_time = current_timestamp - random.randint(1, 3600*39)  # 随机时间
-        print(f"随机时间戳对应的时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(random_time))}")
-        chat_ = memory_graph.get_random_chat_from_db(chat_size, random_time)
-        chat_text.append(chat_)  # 拼接所有text
-        time.sleep(5)
 
 
-    
-    for input_text in chat_text:
-        print(input_text)
-        first_memory = set()
-        first_memory = memory_compress(input_text, llm_model_small, llm_model_small, rate=2.5)
+# 海马体 
+class Hippocampus:
+    def __init__(self,memory_graph:Memory_graph):
+        self.memory_graph = memory_graph
+        self.llm_model = LLMModel()
+        self.llm_model_small = LLMModel(model_name="deepseek-ai/DeepSeek-V2.5")
         
-        #将记忆加入到图谱中
-        for topic, memory in first_memory:
-            topics = segment_text(topic)
-            print(f"\033[1;34m话题\033[0m: {topic},节点: {topics}, 记忆: {memory}")
-            for split_topic in topics:
-                memory_graph.add_dot(split_topic,memory)
-            for split_topic in topics:
-                for other_split_topic in topics:
-                    if split_topic != other_split_topic:
-                        memory_graph.connect_dot(split_topic, other_split_topic)
+    def get_memory_sample(self,chat_size=20,time_frequency:dict={'near':2,'mid':4,'far':3}):
+        current_timestamp = datetime.datetime.now().timestamp()
+        chat_text = []
+        #短期：1h   中期：4h   长期：24h
+        for _ in range(time_frequency.get('near')):  # 循环10次
+            random_time = current_timestamp - random.randint(1, 3600)  # 随机时间
+            # print(f"获得 最近 随机时间戳对应的时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(random_time))}")
+            chat_ = get_cloest_chat_from_db(db=self.memory_graph.db, length=chat_size, timestamp=random_time)
+            chat_text.append(chat_)  
+        for _ in range(time_frequency.get('mid')):  # 循环10次
+            random_time = current_timestamp - random.randint(3600, 3600*4)  # 随机时间
+            # print(f"获得 最近 随机时间戳对应的时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(random_time))}")
+            chat_ = get_cloest_chat_from_db(db=self.memory_graph.db, length=chat_size, timestamp=random_time)
+            chat_text.append(chat_)  
+        for _ in range(time_frequency.get('far')):  # 循环10次
+            random_time = current_timestamp - random.randint(3600*4, 3600*24)  # 随机时间
+            # print(f"获得 最近 随机时间戳对应的时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(random_time))}")
+            chat_ = get_cloest_chat_from_db(db=self.memory_graph.db, length=chat_size, timestamp=random_time)
+            chat_text.append(chat_)
+        return chat_text
     
-    # memory_graph.store_memory()
-    
-    # 展示两种不同的可视化方式
-    print("\n按连接数量着色的图谱：")
-    visualize_graph(memory_graph, color_by_memory=False)
-    
-    print("\n按记忆数量着色的图谱：")
-    visualize_graph(memory_graph, color_by_memory=True)
-    
-    memory_graph.save_graph_to_db()
-    # memory_graph.load_graph_from_db()
-    
-    while True:
-        query = input("请输入新的查询概念（输入'退出'以结束）：")
-        if query.lower() == '退出':
-            break
-        items_list = memory_graph.get_related_item(query)
-        if items_list:
-            # print(items_list)
-            for memory_item in items_list:
-                print(memory_item)
-        else:
-            print("未找到相关记忆。")
+    def build_memory(self,chat_size=12):
+        #最近消息获取频率
+        time_frequency = {'near':1,'mid':2,'far':2}
+        memory_sample = self.get_memory_sample(chat_size,time_frequency)
+        # print(f"\033[1;32m[记忆构建]\033[0m 获取记忆样本: {memory_sample}")   
+
+
+        for i, input_text in enumerate(memory_sample, 1):
+            #加载进度可视化
+            progress = (i / len(memory_sample)) * 100
+            bar_length = 30
+            filled_length = int(bar_length * i // len(memory_sample))
+            bar = '█' * filled_length + '-' * (bar_length - filled_length)
+            print(f"\n进度: [{bar}] {progress:.1f}% ({i}/{len(memory_sample)})")
             
-    while True:
-        query = input("请输入问题：")
-        
-        if query.lower() == '退出':
-            break
-        
-        topic_prompt = find_topic(query, 3)
-        topic_response = llm_model.generate_response(topic_prompt)
+            # 生成压缩后记忆
+            first_memory = set()
+            first_memory = self.memory_compress(input_text, 2.5)
+            # 延时防止访问超频
+            # time.sleep(5)
+            #将记忆加入到图谱中
+            for topic, memory in first_memory:
+                topics = segment_text(topic)
+                print(f"\033[1;34m话题\033[0m: {topic},节点: {topics}, 记忆: {memory}")
+                for split_topic in topics:
+                    self.memory_graph.add_dot(split_topic,memory)
+                for split_topic in topics:
+                    for other_split_topic in topics:
+                        if split_topic != other_split_topic:
+                            self.memory_graph.connect_dot(split_topic, other_split_topic)
+            
+            self.memory_graph.save_graph_to_db()
+    
+    def memory_compress(self, input_text, rate=1):
+        information_content = calculate_information_content(input_text)
+        print(f"文本的信息量（熵）: {information_content:.4f} bits")
+        topic_num = max(1, min(5, int(information_content * rate / 4)))
+        # print(topic_num)
+        topic_prompt = find_topic(input_text, topic_num)
+        topic_response = self.llm_model.generate_response(topic_prompt)
         # 检查 topic_response 是否为元组
         if isinstance(topic_response, tuple):
             topics = topic_response[0].split(",")  # 假设第一个元素是我们需要的字符串
         else:
             topics = topic_response.split(",")
-        print(topics)
-        
-        for keyword in topics:
-            items_list = memory_graph.get_related_item(keyword)
-            if items_list:
-                print(items_list)
-    
-def memory_compress(input_text, llm_model, llm_model_small, rate=1):
-    information_content = calculate_information_content(input_text)
-    print(f"文本的信息量（熵）: {information_content:.4f} bits")
-    topic_num = max(1, min(5, int(information_content * rate / 4)))
-    print(topic_num)
-    topic_prompt = find_topic(input_text, topic_num)
-    topic_response = llm_model.generate_response(topic_prompt)
-    # 检查 topic_response 是否为元组
-    if isinstance(topic_response, tuple):
-        topics = topic_response[0].split(",")  # 假设第一个元素是我们需要的字符串
-    else:
-        topics = topic_response.split(",")
-    print(topics)
-    compressed_memory = set()
-    for topic in topics:
-        topic_what_prompt = topic_what(input_text,topic)
-        topic_what_response = llm_model_small.generate_response(topic_what_prompt)
-        compressed_memory.add((topic.strip(), topic_what_response[0]))  # 将话题和记忆作为元组存储
-    return compressed_memory
+        # print(topics)
+        compressed_memory = set()
+        for topic in topics:
+            topic_what_prompt = topic_what(input_text,topic)
+            topic_what_response = self.llm_model_small.generate_response(topic_what_prompt)
+            compressed_memory.add((topic.strip(), topic_what_response[0]))  # 将话题和记忆作为元组存储
+        return compressed_memory
 
 
 def segment_text(text):
@@ -305,69 +258,21 @@ def topic_what(text, topic):
     prompt = f'这是一段文字：{text}。我想知道这记忆里有什么关于{topic}的话题，帮我总结成一句自然的话，可以包含时间和人物。只输出这句话就好'
     return prompt
 
-def visualize_graph(memory_graph: Memory_graph, color_by_memory: bool = False):
-    # 设置中文字体
-    plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
-    plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
-    
-    G = memory_graph.G
-    
-    # 保存图到本地
-    nx.write_gml(G, "memory_graph.gml")  # 保存为 GML 格式
-
-    # 根据连接条数或记忆数量设置节点颜色
-    node_colors = []
-    nodes = list(G.nodes())  # 获取图中实际的节点列表
-    
-    if color_by_memory:
-        # 计算每个节点的记忆数量
-        memory_counts = []
-        for node in nodes:
-            memory_items = G.nodes[node].get('memory_items', [])
-            if isinstance(memory_items, list):
-                count = len(memory_items)
-            else:
-                count = 1 if memory_items else 0
-            memory_counts.append(count)
-        max_memories = max(memory_counts) if memory_counts else 1
-        
-        for count in memory_counts:
-            # 使用不同的颜色方案：红色表示记忆多，蓝色表示记忆少
-            if max_memories > 0:
-                intensity = min(1.0, count / max_memories)
-                color = (intensity, 0, 1.0 - intensity)  # 从蓝色渐变到红色
-            else:
-                color = (0, 0, 1)  # 如果没有记忆，则为蓝色
-            node_colors.append(color)
-    else:
-        # 使用原来的连接数量着色方案
-        max_degree = max(G.degree(), key=lambda x: x[1])[1] if G.degree() else 1
-        for node in nodes:
-            degree = G.degree(node)
-            if max_degree > 0:
-                red = min(1.0, degree / max_degree)
-                blue = 1.0 - red
-                color = (red, 0, blue)
-            else:
-                color = (0, 0, 1)
-            node_colors.append(color)
-    
-    # 绘制图形
-    plt.figure(figsize=(12, 8))
-    pos = nx.spring_layout(G, k=1, iterations=50)
-    nx.draw(G, pos, 
-           with_labels=True, 
-           node_color=node_colors,
-           node_size=2000,
-           font_size=10,
-           font_family='SimHei',
-           font_weight='bold')
-    
-    title = '记忆图谱可视化 - ' + ('按记忆数量着色' if color_by_memory else '按连接数量着色')
-    plt.title(title, fontsize=16, fontfamily='SimHei')
-    plt.show()
-
-if __name__ == "__main__":
-    main()
 
     
+start_time = time.time()
+
+Database.initialize(
+    global_config.MONGODB_HOST,
+    global_config.MONGODB_PORT,
+    global_config.DATABASE_NAME
+)
+#创建记忆图
+memory_graph = Memory_graph()
+#加载数据库中存储的记忆图
+memory_graph.load_graph_from_db()
+#创建海马体
+hippocampus = Hippocampus(memory_graph)
+
+end_time = time.time()
+print(f"\033[32m[加载海马体耗时: {end_time - start_time:.2f} 秒]\033[0m")

src/plugins/memory_system/memory_make.py

@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 import sys
 import jieba
-from llm_module import LLMModel
 import networkx as nx
 import matplotlib.pyplot as plt
 import math
@@ -9,10 +8,12 @@ from collections import Counter
 import datetime
 import random
 import time
+import os
+from dotenv import load_dotenv
 # from chat.config import global_config
-import sys
 sys.path.append("C:/GitHub/MaiMBot")  # 添加项目根目录到 Python 路径
 from src.common.database import Database  # 使用正确的导入语法
+from src.plugins.memory_system.llm_module import LLMModel
    
 class Memory_graph:
     def __init__(self):
@@ -117,22 +118,60 @@ class Memory_graph:
         return []  # 如果没有找到记录，返回空列表
 
     def save_graph_to_db(self):
-        # 清空现有的图数据
-        self.db.db.graph_data.delete_many({})
         # 保存节点
         for node in self.G.nodes(data=True):
-            node_data = {
-                'concept': node[0],
-                'memory_items': node[1].get('memory_items', [])  # 默认为空列表
-            }
-            self.db.db.graph_data.nodes.insert_one(node_data)
+            concept = node[0]
+            memory_items = node[1].get('memory_items', [])
+            
+            # 查找是否存在同名节点
+            existing_node = self.db.db.graph_data.nodes.find_one({'concept': concept})
+            if existing_node:
+                # 如果存在,合并memory_items并去重
+                existing_items = existing_node.get('memory_items', [])
+                if not isinstance(existing_items, list):
+                    existing_items = [existing_items] if existing_items else []
+                
+                # 合并并去重
+                all_items = list(set(existing_items + memory_items))
+                
+                # 更新节点
+                self.db.db.graph_data.nodes.update_one(
+                    {'concept': concept},
+                    {'$set': {'memory_items': all_items}}
+                )
+            else:
+                # 如果不存在,创建新节点
+                node_data = {
+                    'concept': concept,
+                    'memory_items': memory_items
+                }
+                self.db.db.graph_data.nodes.insert_one(node_data)
+        
         # 保存边
         for edge in self.G.edges():
-            edge_data = {
-                'source': edge[0],
-                'target': edge[1]
-            }
-            self.db.db.graph_data.edges.insert_one(edge_data)
+            source, target = edge
+            
+            # 查找是否存在同样的边
+            existing_edge = self.db.db.graph_data.edges.find_one({
+                'source': source,
+                'target': target
+            })
+            
+            if existing_edge:
+                # 如果存在,增加num属性
+                num = existing_edge.get('num', 1) + 1
+                self.db.db.graph_data.edges.update_one(
+                    {'source': source, 'target': target},
+                    {'$set': {'num': num}}
+                )
+            else:
+                # 如果不存在,创建新边
+                edge_data = {
+                    'source': source,
+                    'target': target,
+                    'num': 1
+                }
+                self.db.db.graph_data.edges.insert_one(edge_data)
 
     def load_graph_from_db(self):
         # 清空当前图
@@ -147,7 +186,7 @@ class Memory_graph:
         # 加载边
         edges = self.db.db.graph_data.edges.find()
         for edge in edges:
-            self.G.add_edge(edge['source'], edge['target'])
+            self.G.add_edge(edge['source'], edge['target'], num=edge.get('num', 1))
 
 def calculate_information_content(text):
     
@@ -180,6 +219,19 @@ def calculate_information_content(text):
 
 
 def main():
+    # 获取当前文件的绝对路径
+    current_dir = os.path.dirname(os.path.abspath(__file__))
+    root_dir = os.path.abspath(os.path.join(current_dir, '..', '..', '..'))
+    env_path = os.path.join(root_dir, 'config', '.env')
+
+    # 加载环境变量
+    print(f"尝试从 {env_path} 加载环境变量配置")
+    if os.path.exists(env_path):
+        load_dotenv(env_path)
+        print("成功加载环境变量配置")
+    else:
+        print(f"环境变量配置文件不存在: {env_path}")
+
     # 初始化数据库
     Database.initialize(
         "127.0.0.1",
@@ -196,10 +248,10 @@ def main():
     current_timestamp = datetime.datetime.now().timestamp()
     chat_text = []
     
-    chat_size =20
+    chat_size =25
     
-    for _ in range(10):  # 循环10次
-        random_time = current_timestamp - random.randint(1, 3600*3)  # 随机时间
+    for _ in range(30):  # 循环10次
+        random_time = current_timestamp - random.randint(1, 3600*10)  # 随机时间
         print(f"随机时间戳对应的时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(random_time))}")
         chat_ = memory_graph.get_random_chat_from_db(chat_size, random_time)
         chat_text.append(chat_)  # 拼接所有text
@@ -218,7 +270,7 @@ def main():
         # print(input_text)
         first_memory = set()
         first_memory = memory_compress(input_text, llm_model_small, llm_model_small, rate=2.5)
-        time.sleep(5)
+        # time.sleep(5)
         
         #将记忆加入到图谱中
         for topic, memory in first_memory: