Merge branch 'main-fix' into main
This commit is contained in:
覅是
GitHub
@@ -101,7 +101,7 @@
|
||||
|
||||
<div align="center">
|
||||
<a href="https://www.bilibili.com/video/BV1amAneGE3P" target="_blank">
|
||||
<img src="docs/video.png" width="300" alt="麦麦演示视频">
|
||||
<img src="docs/pic/video.png" width="300" alt="麦麦演示视频">
|
||||
<br>
|
||||
👆 点击观看麦麦演示视频 👆
|
||||
|
||||
@@ -149,6 +149,8 @@ MaiMBot是一个开源项目，我们非常欢迎你的参与。你的贡献，
|
||||
|
||||
- [📦 Linux 手动部署指南 ](docs/manual_deploy_linux.md)
|
||||
|
||||
- [📦 macOS 手动部署指南 ](docs/manual_deploy_macos.md)
|
||||
|
||||
如果你不知道Docker是什么，建议寻找相关教程或使用手动部署 **（现在不建议使用docker，更新慢，可能不适配）**
|
||||
|
||||
- [🐳 Docker部署指南](docs/docker_deploy.md)
|
||||
|
||||
4
bot.py
4
bot.py
@@ -221,7 +221,9 @@ def check_eula():
|
||||
# 如果EULA或隐私条款有更新，提示用户重新确认
|
||||
if eula_updated or privacy_updated:
|
||||
print("EULA或隐私条款内容已更新，请在阅读后重新确认，继续运行视为同意更新后的以上两款协议")
|
||||
print(f'输入"同意"或"confirmed"或设置环境变量"EULA_AGREE={eula_new_hash}"和"PRIVACY_AGREE={privacy_new_hash}"继续运行')
|
||||
print(
|
||||
f'输入"同意"或"confirmed"或设置环境变量"EULA_AGREE={eula_new_hash}"和"PRIVACY_AGREE={privacy_new_hash}"继续运行'
|
||||
)
|
||||
while True:
|
||||
user_input = input().strip().lower()
|
||||
if user_input in ["同意", "confirmed"]:
|
||||
|
||||
@@ -10,7 +10,7 @@
|
||||
|
||||
- 为什么显示:"缺失必要的API KEY" ❓
|
||||
|
||||
<img src="API_KEY.png" width=650>
|
||||
<img src="./pic/API_KEY.png" width=650>
|
||||
|
||||
>你需要在 [Silicon Flow Api](https://cloud.siliconflow.cn/account/ak) 网站上注册一个账号，然后点击这个链接打开API KEY获取页面。
|
||||
>
|
||||
@@ -41,19 +41,19 @@
|
||||
|
||||
>打开你的MongoDB Compass软件，你会在左上角看到这样的一个界面：
|
||||
>
|
||||
><img src="MONGO_DB_0.png" width=250>
|
||||
><img src="./pic/MONGO_DB_0.png" width=250>
|
||||
>
|
||||
><br>
|
||||
>
|
||||
>点击 "CONNECT" 之后，点击展开 MegBot 标签栏
|
||||
>
|
||||
><img src="MONGO_DB_1.png" width=250>
|
||||
><img src="./pic/MONGO_DB_1.png" width=250>
|
||||
>
|
||||
><br>
|
||||
>
|
||||
>点进 "emoji" 再点击 "DELETE" 删掉所有条目，如图所示
|
||||
>
|
||||
><img src="MONGO_DB_2.png" width=450>
|
||||
><img src="./pic/MONGO_DB_2.png" width=450>
|
||||
>
|
||||
><br>
|
||||
>
|
||||
|
||||
@@ -147,9 +147,7 @@ enable_check = false # 是否要检查表情包是不是合适的喵
|
||||
check_prompt = "符合公序良俗" # 检查表情包的标准呢
|
||||
|
||||
[others]
|
||||
enable_advance_output = true # 是否要显示更多的运行信息呢
|
||||
enable_kuuki_read = true # 让机器人能够"察言观色"喵
|
||||
enable_debug_output = false # 是否启用调试输出喵
|
||||
enable_friend_chat = false # 是否启用好友聊天喵
|
||||
|
||||
[groups]
|
||||
|
||||
@@ -115,9 +115,7 @@ talk_frequency_down = [] # 降低回复频率的群号
|
||||
ban_user_id = [] # 禁止回复的用户QQ号
|
||||
|
||||
[others]
|
||||
enable_advance_output = true # 是否启用高级输出
|
||||
enable_kuuki_read = true # 是否启用读空气功能
|
||||
enable_debug_output = false # 是否启用调试输出
|
||||
enable_friend_chat = false # 是否启用好友聊天
|
||||
|
||||
# 模型配置
|
||||
|
||||
@@ -1,48 +1,51 @@
|
||||
# 面向纯新手的Linux服务器麦麦部署指南
|
||||
|
||||
## 你得先有一个服务器
|
||||
|
||||
为了能使麦麦在你的电脑关机之后还能运行，你需要一台不间断开机的主机，也就是我们常说的服务器。
|
||||
## 事前准备
|
||||
为了能使麦麦不间断的运行，你需要一台一直开着的主机。
|
||||
|
||||
### 如果你想购买服务器
|
||||
华为云、阿里云、腾讯云等等都是在国内可以选择的选择。
|
||||
|
||||
你可以去租一台最低配置的就足敷需要了，按月租大概十几块钱就能租到了。
|
||||
租一台最低配置的就足敷需要了，按月租大概十几块钱就能租到了。
|
||||
|
||||
我们假设你已经租好了一台Linux架构的云服务器。我用的是阿里云ubuntu24.04，其他的原理相似。
|
||||
### 如果你不想购买服务器
|
||||
你可以准备一台可以一直开着的电脑/主机，只需要保证能够正常访问互联网即可
|
||||
|
||||
我们假设你已经有了一台Linux架构的服务器。举例使用的是Ubuntu24.04，其他的原理相似。
|
||||
|
||||
## 0.我们就从零开始吧
|
||||
|
||||
### 网络问题
|
||||
|
||||
为访问github相关界面，推荐去下一款加速器，新手可以试试watttoolkit。
|
||||
为访问Github相关界面，推荐去下一款加速器，新手可以试试[Watt Toolkit](https://gitee.com/rmbgame/SteamTools/releases/latest)。
|
||||
|
||||
### 安装包下载
|
||||
|
||||
#### MongoDB
|
||||
进入[MongoDB下载页](https://www.mongodb.com/try/download/community-kubernetes-operator)，并选择版本
|
||||
|
||||
对于ubuntu24.04 x86来说是这个：
|
||||
以Ubuntu24.04 x86为例，保持如图所示选项，点击`Download`即可，如果是其他系统，请在`Platform`中自行选择：
|
||||
|
||||
https://repo.mongodb.org/apt/ubuntu/dists/noble/mongodb-org/8.0/multiverse/binary-amd64/mongodb-org-server_8.0.5_amd64.deb
|
||||
|
||||
|
||||
如果不是就在这里自行选择对应版本
|
||||
|
||||
https://www.mongodb.com/try/download/community-kubernetes-operator
|
||||
不想使用上述方式？你也可以参考[官方文档](https://www.mongodb.com/zh-cn/docs/manual/administration/install-on-linux/#std-label-install-mdb-community-edition-linux)进行安装，进入后选择自己的系统版本即可
|
||||
|
||||
#### Napcat
|
||||
|
||||
在这里选择对应版本。
|
||||
|
||||
https://github.com/NapNeko/NapCatQQ/releases/tag/v4.6.7
|
||||
|
||||
对于ubuntu24.04 x86来说是这个：
|
||||
|
||||
https://dldir1.qq.com/qqfile/qq/QQNT/ee4bd910/linuxqq_3.2.16-32793_amd64.deb
|
||||
#### QQ（可选）/Napcat
|
||||
*如果你使用Napcat的脚本安装，可以忽略此步*
|
||||
访问https://github.com/NapNeko/NapCatQQ/releases/latest
|
||||
在图中所示区域可以找到QQ的下载链接，选择对应版本下载即可
|
||||
从这里下载，可以保证你下载到的QQ版本兼容最新版Napcat
|
||||
|
||||
如果你不想使用Napcat的脚本安装，还需参考[Napcat-Linux手动安装](https://www.napcat.wiki/guide/boot/Shell-Linux-SemiAuto)
|
||||
|
||||
#### 麦麦
|
||||
|
||||
https://github.com/SengokuCola/MaiMBot/archive/refs/tags/0.5.8-alpha.zip
|
||||
|
||||
下载这个官方压缩包。
|
||||
先打开https://github.com/MaiM-with-u/MaiBot/releases
|
||||
往下滑找到这个
|
||||
|
||||
下载箭头所指这个压缩包。
|
||||
|
||||
### 路径
|
||||
|
||||
@@ -53,10 +56,10 @@ https://github.com/SengokuCola/MaiMBot/archive/refs/tags/0.5.8-alpha.zip
|
||||
```
|
||||
moi
|
||||
└─ mai
|
||||
├─ linuxqq_3.2.16-32793_amd64.deb
|
||||
├─ mongodb-org-server_8.0.5_amd64.deb
|
||||
├─ linuxqq_3.2.16-32793_amd64.deb # linuxqq安装包
|
||||
├─ mongodb-org-server_8.0.5_amd64.deb # MongoDB的安装包
|
||||
└─ bot
|
||||
└─ MaiMBot-0.5.8-alpha.zip
|
||||
└─ MaiMBot-0.5.8-alpha.zip # 麦麦的压缩包
|
||||
```
|
||||
|
||||
### 网络
|
||||
@@ -69,7 +72,7 @@ moi
|
||||
|
||||
## 2. Python的安装
|
||||
|
||||
- 导入 Python 的稳定版 PPA：
|
||||
- 导入 Python 的稳定版 PPA（Ubuntu需执行此步，Debian可忽略）：
|
||||
|
||||
```bash
|
||||
sudo add-apt-repository ppa:deadsnakes/ppa
|
||||
@@ -92,6 +95,11 @@ sudo apt install python3.12
|
||||
```bash
|
||||
python3.12 --version
|
||||
```
|
||||
- （可选）更新替代方案，设置 python3.12 为默认的 python3 版本:
|
||||
```bash
|
||||
sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.12 1
|
||||
sudo update-alternatives --config python3
|
||||
```
|
||||
|
||||
- 在「终端」中，执行以下命令安装 pip：
|
||||
|
||||
@@ -141,23 +149,17 @@ systemctl status mongod #通过这条指令检查运行状态
|
||||
sudo systemctl enable mongod
|
||||
```
|
||||
|
||||
## 5.napcat的安装
|
||||
## 5.Napcat的安装
|
||||
|
||||
``` bash
|
||||
# 该脚本适用于支持Ubuntu 20+/Debian 10+/Centos9
|
||||
curl -o napcat.sh https://nclatest.znin.net/NapNeko/NapCat-Installer/main/script/install.sh && sudo bash napcat.sh
|
||||
```
|
||||
|
||||
上面的不行试试下面的
|
||||
|
||||
``` bash
|
||||
dpkg -i linuxqq_3.2.16-32793_amd64.deb
|
||||
apt-get install -f
|
||||
dpkg -i linuxqq_3.2.16-32793_amd64.deb
|
||||
```
|
||||
执行后，脚本会自动帮你部署好QQ及Napcat
|
||||
|
||||
成功的标志是输入``` napcat ```出来炫酷的彩虹色界面
|
||||
|
||||
## 6.napcat的运行
|
||||
## 6.Napcat的运行
|
||||
|
||||
此时你就可以根据提示在```napcat```里面登录你的QQ号了。
|
||||
|
||||
@@ -170,6 +172,13 @@ napcat status #检查运行状态
|
||||
|
||||
```http://<你服务器的公网IP>:6099/webui?token=napcat```
|
||||
|
||||
如果你部署在自己的电脑上：
|
||||
```http://127.0.0.1:6099/webui?token=napcat```
|
||||
|
||||
> [!WARNING]
|
||||
> 如果你的麦麦部署在公网，请**务必**修改Napcat的默认密码
|
||||
|
||||
|
||||
第一次是这个，后续改了密码之后token就会对应修改。你也可以使用```napcat log <你的QQ号>```来查看webui地址。把里面的```127.0.0.1```改成<你服务器的公网IP>即可。
|
||||
|
||||
登录上之后在网络配置界面添加websocket客户端，名称随便输一个，url改成`ws://127.0.0.1:8080/onebot/v11/ws`保存之后点启用，就大功告成了。
|
||||
@@ -178,7 +187,7 @@ napcat status #检查运行状态
|
||||
|
||||
### step 1 安装解压软件
|
||||
|
||||
```
|
||||
```bash
|
||||
sudo apt-get install unzip
|
||||
```
|
||||
|
||||
@@ -229,138 +238,11 @@ bot
|
||||
|
||||
你可以注册一个硅基流动的账号，通过邀请码注册有14块钱的免费额度：https://cloud.siliconflow.cn/i/7Yld7cfg。
|
||||
|
||||
#### 在.env.prod中定义API凭证：
|
||||
#### 修改配置文件
|
||||
请参考
|
||||
- [🎀 新手配置指南](./installation_cute.md) - 通俗易懂的配置教程，适合初次使用的猫娘
|
||||
- [⚙️ 标准配置指南](./installation_standard.md) - 简明专业的配置说明，适合有经验的用户
|
||||
|
||||
```
|
||||
# API凭证配置
|
||||
SILICONFLOW_KEY=your_key # 硅基流动API密钥
|
||||
SILICONFLOW_BASE_URL=https://api.siliconflow.cn/v1/ # 硅基流动API地址
|
||||
|
||||
DEEP_SEEK_KEY=your_key # DeepSeek API密钥
|
||||
DEEP_SEEK_BASE_URL=https://api.deepseek.com/v1 # DeepSeek API地址
|
||||
|
||||
CHAT_ANY_WHERE_KEY=your_key # ChatAnyWhere API密钥
|
||||
CHAT_ANY_WHERE_BASE_URL=https://api.chatanywhere.tech/v1 # ChatAnyWhere API地址
|
||||
```
|
||||
|
||||
#### 在bot_config.toml中引用API凭证：
|
||||
|
||||
```
|
||||
[model.llm_reasoning]
|
||||
name = "Pro/deepseek-ai/DeepSeek-R1"
|
||||
base_url = "SILICONFLOW_BASE_URL" # 引用.env.prod中定义的地址
|
||||
key = "SILICONFLOW_KEY" # 引用.env.prod中定义的密钥
|
||||
```
|
||||
|
||||
如需切换到其他API服务，只需修改引用：
|
||||
|
||||
```
|
||||
[model.llm_reasoning]
|
||||
name = "Pro/deepseek-ai/DeepSeek-R1"
|
||||
base_url = "DEEP_SEEK_BASE_URL" # 切换为DeepSeek服务
|
||||
key = "DEEP_SEEK_KEY" # 使用DeepSeek密钥
|
||||
```
|
||||
|
||||
#### 配置文件详解
|
||||
|
||||
##### 环境配置文件 (.env.prod)
|
||||
|
||||
```
|
||||
# API配置
|
||||
SILICONFLOW_KEY=your_key
|
||||
SILICONFLOW_BASE_URL=https://api.siliconflow.cn/v1/
|
||||
DEEP_SEEK_KEY=your_key
|
||||
DEEP_SEEK_BASE_URL=https://api.deepseek.com/v1
|
||||
CHAT_ANY_WHERE_KEY=your_key
|
||||
CHAT_ANY_WHERE_BASE_URL=https://api.chatanywhere.tech/v1
|
||||
|
||||
# 服务配置
|
||||
HOST=127.0.0.1 # 如果使用Docker部署，需要改成0.0.0.0，否则QQ消息无法传入
|
||||
PORT=8080
|
||||
|
||||
# 数据库配置
|
||||
MONGODB_HOST=127.0.0.1 # 如果使用Docker部署，需要改成数据库容器的名字，默认是mongodb
|
||||
MONGODB_PORT=27017
|
||||
DATABASE_NAME=MegBot
|
||||
MONGODB_USERNAME = "" # 数据库用户名
|
||||
MONGODB_PASSWORD = "" # 数据库密码
|
||||
MONGODB_AUTH_SOURCE = "" # 认证数据库
|
||||
|
||||
# 插件配置
|
||||
PLUGINS=["src2.plugins.chat"]
|
||||
```
|
||||
|
||||
##### 机器人配置文件 (bot_config.toml)
|
||||
|
||||
```
|
||||
[bot]
|
||||
qq = "机器人QQ号" # 必填
|
||||
nickname = "麦麦" # 机器人昵称（你希望机器人怎么称呼它自己）
|
||||
|
||||
[personality]
|
||||
prompt_personality = [
|
||||
"曾经是一个学习地质的女大学生，现在学习心理学和脑科学，你会刷贴吧",
|
||||
"是一个女大学生，你有黑色头发，你会刷小红书"
|
||||
]
|
||||
prompt_schedule = "一个曾经学习地质,现在学习心理学和脑科学的女大学生，喜欢刷qq，贴吧，知乎和小红书"
|
||||
|
||||
[message]
|
||||
min_text_length = 2 # 最小回复长度
|
||||
max_context_size = 15 # 上下文记忆条数
|
||||
emoji_chance = 0.2 # 表情使用概率
|
||||
ban_words = [] # 禁用词列表
|
||||
|
||||
[emoji]
|
||||
auto_save = true # 自动保存表情
|
||||
enable_check = false # 启用表情审核
|
||||
check_prompt = "符合公序良俗"
|
||||
|
||||
[groups]
|
||||
talk_allowed = [] # 允许对话的群号
|
||||
talk_frequency_down = [] # 降低回复频率的群号
|
||||
ban_user_id = [] # 禁止回复的用户QQ号
|
||||
|
||||
[others]
|
||||
enable_advance_output = true # 启用详细日志
|
||||
enable_kuuki_read = true # 启用场景理解
|
||||
|
||||
# 模型配置
|
||||
[model.llm_reasoning] # 推理模型
|
||||
name = "Pro/deepseek-ai/DeepSeek-R1"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
[model.llm_reasoning_minor] # 轻量推理模型
|
||||
name = "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
[model.llm_normal] # 对话模型
|
||||
name = "Pro/deepseek-ai/DeepSeek-V3"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
[model.llm_normal_minor] # 备用对话模型
|
||||
name = "deepseek-ai/DeepSeek-V2.5"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
[model.vlm] # 图像识别模型
|
||||
name = "deepseek-ai/deepseek-vl2"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
[model.embedding] # 文本向量模型
|
||||
name = "BAAI/bge-m3"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
|
||||
|
||||
[topic.llm_topic]
|
||||
name = "Pro/deepseek-ai/DeepSeek-V3"
|
||||
base_url = "SILICONFLOW_BASE_URL"
|
||||
key = "SILICONFLOW_KEY"
|
||||
```
|
||||
|
||||
**step # 6** 运行
|
||||
|
||||
@@ -438,7 +320,7 @@ sudo systemctl enable bot.service # 启动bot服务
|
||||
sudo systemctl status bot.service # 检查bot服务状态
|
||||
```
|
||||
|
||||
```
|
||||
python bot.py
|
||||
```bash
|
||||
python bot.py # 运行麦麦
|
||||
```
|
||||
|
||||
|
||||
@@ -6,7 +6,7 @@
|
||||
- QQ小号（QQ框架的使用可能导致qq被风控，严重（小概率）可能会导致账号封禁，强烈不推荐使用大号）
|
||||
- 可用的大模型API
|
||||
- 一个AI助手，网上随便搜一家打开来用都行，可以帮你解决一些不懂的问题
|
||||
- 以下内容假设你对Linux系统有一定的了解，如果觉得难以理解，请直接用Windows系统部署[Windows系统部署指南](./manual_deploy_windows.md)
|
||||
- 以下内容假设你对Linux系统有一定的了解，如果觉得难以理解，请直接用Windows系统部署[Windows系统部署指南](./manual_deploy_windows.md)或[使用Windows一键包部署](https://github.com/MaiM-with-u/MaiBot/releases/tag/EasyInstall-windows)
|
||||
|
||||
## 你需要知道什么？
|
||||
|
||||
@@ -24,6 +24,9 @@
|
||||
|
||||
---
|
||||
|
||||
## 一键部署
|
||||
请下载并运行项目根目录中的run.sh并按照提示安装，部署完成后请参照后续配置指南进行配置
|
||||
|
||||
## 环境配置
|
||||
|
||||
### 1️⃣ **确认Python版本**
|
||||
@@ -36,17 +39,26 @@ python --version
|
||||
python3 --version
|
||||
```
|
||||
|
||||
如果版本低于3.9，请更新Python版本。
|
||||
如果版本低于3.9，请更新Python版本，目前建议使用python3.12
|
||||
|
||||
```bash
|
||||
# Ubuntu/Debian
|
||||
# Debian
|
||||
sudo apt update
|
||||
sudo apt install python3.9
|
||||
# 如执行了这一步，建议在执行时将python3指向python3.9
|
||||
# 更新替代方案，设置 python3.9 为默认的 python3 版本:
|
||||
sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1
|
||||
sudo apt install python3.12
|
||||
# Ubuntu
|
||||
sudo add-apt-repository ppa:deadsnakes/ppa
|
||||
sudo apt update
|
||||
sudo apt install python3.12
|
||||
|
||||
# 执行完以上命令后，建议在执行时将python3指向python3.12
|
||||
# 更新替代方案，设置 python3.12 为默认的 python3 版本:
|
||||
sudo update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.12 1
|
||||
sudo update-alternatives --config python3
|
||||
```
|
||||
建议再执行以下命令，使后续运行命令中的`python3`等同于`python`
|
||||
```bash
|
||||
sudo apt install python-is-python3
|
||||
```
|
||||
|
||||
### 2️⃣ **创建虚拟环境**
|
||||
|
||||
@@ -73,7 +85,7 @@ pip install -r requirements.txt
|
||||
|
||||
### 3️⃣ **安装并启动MongoDB**
|
||||
|
||||
- 安装与启动：Debian参考[官方文档](https://docs.mongodb.com/manual/tutorial/install-mongodb-on-debian/)，Ubuntu参考[官方文档](https://docs.mongodb.com/manual/tutorial/install-mongodb-on-ubuntu/)
|
||||
- 安装与启动：请参考[官方文档](https://www.mongodb.com/zh-cn/docs/manual/administration/install-on-linux/#std-label-install-mdb-community-edition-linux)，进入后选择自己的系统版本即可
|
||||
- 默认连接本地27017端口
|
||||
|
||||
---
|
||||
@@ -82,7 +94,11 @@ pip install -r requirements.txt
|
||||
|
||||
### 4️⃣ **安装NapCat框架**
|
||||
|
||||
- 参考[NapCat官方文档](https://www.napcat.wiki/guide/boot/Shell#napcat-installer-linux%E4%B8%80%E9%94%AE%E4%BD%BF%E7%94%A8%E8%84%9A%E6%9C%AC-%E6%94%AF%E6%8C%81ubuntu-20-debian-10-centos9)安装
|
||||
- 执行NapCat的Linux一键使用脚本(支持Ubuntu 20+/Debian 10+/Centos9)
|
||||
```bash
|
||||
curl -o napcat.sh https://nclatest.znin.net/NapNeko/NapCat-Installer/main/script/install.sh && sudo bash napcat.sh
|
||||
```
|
||||
- 如果你不想使用Napcat的脚本安装，可参考[Napcat-Linux手动安装](https://www.napcat.wiki/guide/boot/Shell-Linux-SemiAuto)
|
||||
|
||||
- 使用QQ小号登录，添加反向WS地址: `ws://127.0.0.1:8080/onebot/v11/ws`
|
||||
|
||||
@@ -91,9 +107,17 @@ pip install -r requirements.txt
|
||||
## 配置文件设置
|
||||
|
||||
### 5️⃣ **配置文件设置，让麦麦Bot正常工作**
|
||||
|
||||
- 修改环境配置文件：`.env.prod`
|
||||
- 修改机器人配置文件：`bot_config.toml`
|
||||
可先运行一次
|
||||
```bash
|
||||
# 在项目目录下操作
|
||||
nb run
|
||||
# 或
|
||||
python3 bot.py
|
||||
```
|
||||
之后你就可以找到`.env.prod`和`bot_config.toml`这两个文件了
|
||||
关于文件内容的配置请参考：
|
||||
- [🎀 新手配置指南](./installation_cute.md) - 通俗易懂的配置教程，适合初次使用的猫娘
|
||||
- [⚙️ 标准配置指南](./installation_standard.md) - 简明专业的配置说明，适合有经验的用户
|
||||
|
||||
---
|
||||
|
||||
|
||||
201
docs/manual_deploy_macos.md
Normal file
201
docs/manual_deploy_macos.md
Normal file
@@ -0,0 +1,201 @@
|
||||
# 📦 macOS系统手动部署MaiMbot麦麦指南
|
||||
|
||||
## 准备工作
|
||||
|
||||
- 一台搭载了macOS系统的设备（macOS 12.0 或以上）
|
||||
- QQ小号（QQ框架的使用可能导致qq被风控，严重（小概率）可能会导致账号封禁，强烈不推荐使用大号）
|
||||
- Homebrew包管理器
|
||||
- 如未安装，你可以在https://github.com/Homebrew/brew/releases/latest 找到.pkg格式的安装包
|
||||
- 可用的大模型API
|
||||
- 一个AI助手，网上随便搜一家打开来用都行，可以帮你解决一些不懂的问题
|
||||
- 以下内容假设你对macOS系统有一定的了解，如果觉得难以理解，请直接用Windows系统部署[Windows系统部署指南](./manual_deploy_windows.md)或[使用Windows一键包部署](https://github.com/MaiM-with-u/MaiBot/releases/tag/EasyInstall-windows)
|
||||
- 终端应用（iTerm2等）
|
||||
|
||||
---
|
||||
|
||||
## 环境配置
|
||||
|
||||
### 1️⃣ **Python环境配置**
|
||||
|
||||
```bash
|
||||
# 检查Python版本（macOS自带python可能为2.7）
|
||||
python3 --version
|
||||
|
||||
# 通过Homebrew安装Python
|
||||
brew install python@3.12
|
||||
|
||||
# 设置环境变量（如使用zsh）
|
||||
echo 'export PATH="/usr/local/opt/python@3.12/bin:$PATH"' >> ~/.zshrc
|
||||
source ~/.zshrc
|
||||
|
||||
# 验证安装
|
||||
python3 --version # 应显示3.12.x
|
||||
pip3 --version # 应关联3.12版本
|
||||
```
|
||||
|
||||
### 2️⃣ **创建虚拟环境**
|
||||
|
||||
```bash
|
||||
# 方法1：使用venv（推荐）
|
||||
python3 -m venv maimbot-venv
|
||||
source maimbot-venv/bin/activate # 激活虚拟环境
|
||||
|
||||
# 方法2：使用conda
|
||||
brew install --cask miniconda
|
||||
conda create -n maimbot python=3.9
|
||||
conda activate maimbot # 激活虚拟环境
|
||||
|
||||
# 安装项目依赖
|
||||
# 请确保已经进入虚拟环境再执行
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 数据库配置
|
||||
|
||||
### 3️⃣ **安装MongoDB**
|
||||
|
||||
请参考[官方文档](https://www.mongodb.com/zh-cn/docs/manual/tutorial/install-mongodb-on-os-x/#install-mongodb-community-edition)
|
||||
|
||||
---
|
||||
|
||||
## NapCat
|
||||
|
||||
### 4️⃣ **安装与配置Napcat**
|
||||
- 安装
|
||||
可以使用Napcat官方提供的[macOS安装工具](https://github.com/NapNeko/NapCat-Mac-Installer/releases/)
|
||||
由于权限问题，补丁过程需要手动替换 package.json，请注意备份原文件～
|
||||
- 配置
|
||||
使用QQ小号登录，添加反向WS地址: `ws://127.0.0.1:8080/onebot/v11/ws`
|
||||
|
||||
---
|
||||
|
||||
## 配置文件设置
|
||||
|
||||
### 5️⃣ **生成配置文件**
|
||||
可先运行一次
|
||||
```bash
|
||||
# 在项目目录下操作
|
||||
nb run
|
||||
# 或
|
||||
python3 bot.py
|
||||
```
|
||||
|
||||
之后你就可以找到`.env.prod`和`bot_config.toml`这两个文件了
|
||||
|
||||
关于文件内容的配置请参考：
|
||||
- [🎀 新手配置指南](./installation_cute.md) - 通俗易懂的配置教程，适合初次使用的猫娘
|
||||
- [⚙️ 标准配置指南](./installation_standard.md) - 简明专业的配置说明，适合有经验的用户
|
||||
|
||||
|
||||
---
|
||||
|
||||
## 启动机器人
|
||||
|
||||
### 6️⃣ **启动麦麦机器人**
|
||||
|
||||
```bash
|
||||
# 在项目目录下操作
|
||||
nb run
|
||||
# 或
|
||||
python3 bot.py
|
||||
```
|
||||
|
||||
## 启动管理
|
||||
|
||||
### 7️⃣ **通过launchd管理服务**
|
||||
|
||||
创建plist文件：
|
||||
|
||||
```bash
|
||||
nano ~/Library/LaunchAgents/com.maimbot.plist
|
||||
```
|
||||
|
||||
内容示例（需替换实际路径）：
|
||||
|
||||
```xml
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
|
||||
<plist version="1.0">
|
||||
<dict>
|
||||
<key>Label</key>
|
||||
<string>com.maimbot</string>
|
||||
|
||||
<key>ProgramArguments</key>
|
||||
<array>
|
||||
<string>/path/to/maimbot-venv/bin/python</string>
|
||||
<string>/path/to/MaiMbot/bot.py</string>
|
||||
</array>
|
||||
|
||||
<key>WorkingDirectory</key>
|
||||
<string>/path/to/MaiMbot</string>
|
||||
|
||||
<key>StandardOutPath</key>
|
||||
<string>/tmp/maimbot.log</string>
|
||||
<key>StandardErrorPath</key>
|
||||
<string>/tmp/maimbot.err</string>
|
||||
|
||||
<key>RunAtLoad</key>
|
||||
<true/>
|
||||
<key>KeepAlive</key>
|
||||
<true/>
|
||||
</dict>
|
||||
</plist>
|
||||
```
|
||||
|
||||
加载服务：
|
||||
|
||||
```bash
|
||||
launchctl load ~/Library/LaunchAgents/com.maimbot.plist
|
||||
launchctl start com.maimbot
|
||||
```
|
||||
|
||||
查看日志：
|
||||
|
||||
```bash
|
||||
tail -f /tmp/maimbot.log
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 常见问题处理
|
||||
|
||||
1. **权限问题**
|
||||
```bash
|
||||
# 遇到文件权限错误时
|
||||
chmod -R 755 ~/Documents/MaiMbot
|
||||
```
|
||||
|
||||
2. **Python模块缺失**
|
||||
```bash
|
||||
# 确保在虚拟环境中
|
||||
source maimbot-venv/bin/activate # 或 conda 激活
|
||||
pip install --force-reinstall -r requirements.txt
|
||||
```
|
||||
|
||||
3. **MongoDB连接失败**
|
||||
```bash
|
||||
# 检查服务状态
|
||||
brew services list
|
||||
# 重置数据库权限
|
||||
mongosh --eval "db.adminCommand({setFeatureCompatibilityVersion: '5.0'})"
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 系统优化建议
|
||||
|
||||
1. **关闭App Nap**
|
||||
```bash
|
||||
# 防止系统休眠NapCat进程
|
||||
defaults write NSGlobalDomain NSAppSleepDisabled -bool YES
|
||||
```
|
||||
|
||||
2. **电源管理设置**
|
||||
```bash
|
||||
# 防止睡眠影响机器人运行
|
||||
sudo systemsetup -setcomputersleep Never
|
||||
```
|
||||
|
||||
---
|
||||
|
Before Width: | Height: | Size: 47 KiB After Width: | Height: | Size: 47 KiB |
|
Before Width: | Height: | Size: 13 KiB After Width: | Height: | Size: 13 KiB |
|
Before Width: | Height: | Size: 27 KiB After Width: | Height: | Size: 27 KiB |
|
Before Width: | Height: | Size: 31 KiB After Width: | Height: | Size: 31 KiB |
BIN
docs/pic/MongoDB_Ubuntu_guide.png
Normal file
BIN
docs/pic/MongoDB_Ubuntu_guide.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 14 KiB |
BIN
docs/pic/QQ_Download_guide_Linux.png
Normal file
BIN
docs/pic/QQ_Download_guide_Linux.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 37 KiB |
BIN
docs/pic/linux_beginner_downloadguide.png
Normal file
BIN
docs/pic/linux_beginner_downloadguide.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 10 KiB |
|
Before Width: | Height: | Size: 107 KiB After Width: | Height: | Size: 107 KiB |
|
Before Width: | Height: | Size: 208 KiB After Width: | Height: | Size: 208 KiB |
|
Before Width: | Height: | Size: 170 KiB After Width: | Height: | Size: 170 KiB |
|
Before Width: | Height: | Size: 133 KiB After Width: | Height: | Size: 133 KiB |
|
Before Width: | Height: | Size: 27 KiB After Width: | Height: | Size: 27 KiB |
@@ -16,7 +16,7 @@
|
||||
|
||||
docker-compose.yml: https://github.com/SengokuCola/MaiMBot/blob/main/docker-compose.yml
|
||||
下载后打开，将 `services-mongodb-image` 修改为 `mongo:4.4.24`。这是因为最新的 MongoDB 强制要求 AVX 指令集，而群晖似乎不支持这个指令集
|
||||
|
||||
|
||||
|
||||
bot_config.toml: https://github.com/SengokuCola/MaiMBot/blob/main/template/bot_config_template.toml
|
||||
下载后，重命名为 `bot_config.toml`
|
||||
@@ -26,13 +26,13 @@ bot_config.toml: https://github.com/SengokuCola/MaiMBot/blob/main/template/bot_c
|
||||
下载后，重命名为 `.env.prod`
|
||||
将 `HOST` 修改为 `0.0.0.0`，确保 maimbot 能被 napcat 访问
|
||||
按下图修改 mongodb 设置，使用 `MONGODB_URI`
|
||||
|
||||
|
||||
|
||||
把 `bot_config.toml` 和 `.env.prod` 放入之前创建的 `MaiMBot`文件夹
|
||||
|
||||
#### 如何下载？
|
||||
|
||||
点这里！
|
||||
点这里！
|
||||
|
||||
### 创建项目
|
||||
|
||||
@@ -45,7 +45,7 @@ bot_config.toml: https://github.com/SengokuCola/MaiMBot/blob/main/template/bot_c
|
||||
|
||||
图例：
|
||||
|
||||
|
||||
|
||||
|
||||
一路点下一步，等待项目创建完成
|
||||
|
||||
|
||||
@@ -31,9 +31,10 @@ _handler_registry: Dict[str, List[int]] = {}
|
||||
current_file_path = Path(__file__).resolve()
|
||||
LOG_ROOT = "logs"
|
||||
|
||||
ENABLE_ADVANCE_OUTPUT = False
|
||||
SIMPLE_OUTPUT = os.getenv("SIMPLE_OUTPUT", "false")
|
||||
print(f"SIMPLE_OUTPUT: {SIMPLE_OUTPUT}")
|
||||
|
||||
if ENABLE_ADVANCE_OUTPUT:
|
||||
if not SIMPLE_OUTPUT:
|
||||
# 默认全局配置
|
||||
DEFAULT_CONFIG = {
|
||||
# 日志级别配置
|
||||
@@ -85,7 +86,6 @@ MEMORY_STYLE_CONFIG = {
|
||||
},
|
||||
}
|
||||
|
||||
# 海马体日志样式配置
|
||||
SENDER_STYLE_CONFIG = {
|
||||
"advanced": {
|
||||
"console_format": (
|
||||
@@ -152,17 +152,17 @@ CHAT_STYLE_CONFIG = {
|
||||
"file_format": ("{time:YYYY-MM-DD HH:mm:ss} | {level: <8} | {extra[module]: <15} | 见闻 | {message}"),
|
||||
},
|
||||
"simple": {
|
||||
"console_format": ("<green>{time:MM-DD HH:mm}</green> | <light-blue>见闻</light-blue> | {message}"),
|
||||
"console_format": ("<green>{time:MM-DD HH:mm}</green> | <light-blue>见闻</light-blue> | <green>{message}</green>"), # noqa: E501
|
||||
"file_format": ("{time:YYYY-MM-DD HH:mm:ss} | {level: <8} | {extra[module]: <15} | 见闻 | {message}"),
|
||||
},
|
||||
}
|
||||
|
||||
# 根据ENABLE_ADVANCE_OUTPUT选择配置
|
||||
MEMORY_STYLE_CONFIG = MEMORY_STYLE_CONFIG["advanced"] if ENABLE_ADVANCE_OUTPUT else MEMORY_STYLE_CONFIG["simple"]
|
||||
TOPIC_STYLE_CONFIG = TOPIC_STYLE_CONFIG["advanced"] if ENABLE_ADVANCE_OUTPUT else TOPIC_STYLE_CONFIG["simple"]
|
||||
SENDER_STYLE_CONFIG = SENDER_STYLE_CONFIG["advanced"] if ENABLE_ADVANCE_OUTPUT else SENDER_STYLE_CONFIG["simple"]
|
||||
LLM_STYLE_CONFIG = LLM_STYLE_CONFIG["advanced"] if ENABLE_ADVANCE_OUTPUT else LLM_STYLE_CONFIG["simple"]
|
||||
CHAT_STYLE_CONFIG = CHAT_STYLE_CONFIG["advanced"] if ENABLE_ADVANCE_OUTPUT else CHAT_STYLE_CONFIG["simple"]
|
||||
# 根据SIMPLE_OUTPUT选择配置
|
||||
MEMORY_STYLE_CONFIG = MEMORY_STYLE_CONFIG["simple"] if SIMPLE_OUTPUT else MEMORY_STYLE_CONFIG["advanced"]
|
||||
TOPIC_STYLE_CONFIG = TOPIC_STYLE_CONFIG["simple"] if SIMPLE_OUTPUT else TOPIC_STYLE_CONFIG["advanced"]
|
||||
SENDER_STYLE_CONFIG = SENDER_STYLE_CONFIG["simple"] if SIMPLE_OUTPUT else SENDER_STYLE_CONFIG["advanced"]
|
||||
LLM_STYLE_CONFIG = LLM_STYLE_CONFIG["simple"] if SIMPLE_OUTPUT else LLM_STYLE_CONFIG["advanced"]
|
||||
CHAT_STYLE_CONFIG = CHAT_STYLE_CONFIG["simple"] if SIMPLE_OUTPUT else CHAT_STYLE_CONFIG["advanced"]
|
||||
|
||||
|
||||
def is_registered_module(record: dict) -> bool:
|
||||
|
||||
@@ -98,6 +98,7 @@ async def _(bot: Bot, event: MessageEvent, state: T_State):
|
||||
else:
|
||||
await chat_bot.handle_message(event, bot)
|
||||
|
||||
|
||||
@notice_matcher.handle()
|
||||
async def _(bot: Bot, event: NoticeEvent, state: T_State):
|
||||
logger.debug(f"收到通知：{event}")
|
||||
@@ -110,7 +111,7 @@ async def build_memory_task():
|
||||
"""每build_memory_interval秒执行一次记忆构建"""
|
||||
logger.debug("[记忆构建]------------------------------------开始构建记忆--------------------------------------")
|
||||
start_time = time.time()
|
||||
await hippocampus.operation_build_memory(chat_size=20)
|
||||
await hippocampus.operation_build_memory()
|
||||
end_time = time.time()
|
||||
logger.success(
|
||||
f"[记忆构建]--------------------------记忆构建完成：耗时: {end_time - start_time:.2f} "
|
||||
|
||||
@@ -154,7 +154,7 @@ class ChatBot:
|
||||
)
|
||||
# 开始思考的时间点
|
||||
thinking_time_point = round(time.time(), 2)
|
||||
logger.info(f"开始思考的时间点: {thinking_time_point}")
|
||||
# logger.debug(f"开始思考的时间点: {thinking_time_point}")
|
||||
think_id = "mt" + str(thinking_time_point)
|
||||
thinking_message = MessageThinking(
|
||||
message_id=think_id,
|
||||
@@ -424,7 +424,6 @@ class ChatBot:
|
||||
if event.group_id not in global_config.talk_allowed_groups:
|
||||
return
|
||||
|
||||
|
||||
# 获取合并转发消息的详细信息
|
||||
forward_info = await bot.get_forward_msg(message_id=event.message_id)
|
||||
messages = forward_info["messages"]
|
||||
@@ -456,11 +455,7 @@ class ChatBot:
|
||||
# 构建群聊信息（如果是群聊）
|
||||
group_info = None
|
||||
if isinstance(event, GroupMessageEvent):
|
||||
group_info = GroupInfo(
|
||||
group_id=event.group_id,
|
||||
group_name=None,
|
||||
platform="qq"
|
||||
)
|
||||
group_info = GroupInfo(group_id=event.group_id, group_name=None, platform="qq")
|
||||
|
||||
# 创建消息对象
|
||||
message_cq = MessageRecvCQ(
|
||||
@@ -512,5 +507,6 @@ class ChatBot:
|
||||
else:
|
||||
return f"[{seg_type}]"
|
||||
|
||||
|
||||
# 创建全局ChatBot实例
|
||||
chat_bot = ChatBot()
|
||||
|
||||
@@ -56,7 +56,6 @@ class BotConfig:
|
||||
llm_reasoning: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_reasoning_minor: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_normal: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_normal_minor: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_topic_judge: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_summary_by_topic: Dict[str, str] = field(default_factory=lambda: {})
|
||||
llm_emotion_judge: Dict[str, str] = field(default_factory=lambda: {})
|
||||
@@ -68,9 +67,9 @@ class BotConfig:
|
||||
MODEL_V3_PROBABILITY: float = 0.1 # V3模型概率
|
||||
MODEL_R1_DISTILL_PROBABILITY: float = 0.1 # R1蒸馏模型概率
|
||||
|
||||
enable_advance_output: bool = False # 是否启用高级输出
|
||||
# enable_advance_output: bool = False # 是否启用高级输出
|
||||
enable_kuuki_read: bool = True # 是否启用读空气功能
|
||||
enable_debug_output: bool = False # 是否启用调试输出
|
||||
# enable_debug_output: bool = False # 是否启用调试输出
|
||||
enable_friend_chat: bool = False # 是否启用好友聊天
|
||||
|
||||
mood_update_interval: float = 1.0 # 情绪更新间隔 单位秒
|
||||
@@ -106,6 +105,11 @@ class BotConfig:
|
||||
memory_forget_time: int = 24 # 记忆遗忘时间（小时）
|
||||
memory_forget_percentage: float = 0.01 # 记忆遗忘比例
|
||||
memory_compress_rate: float = 0.1 # 记忆压缩率
|
||||
build_memory_sample_num: int = 10 # 记忆构建采样数量
|
||||
build_memory_sample_length: int = 20 # 记忆构建采样长度
|
||||
memory_build_distribution: list = field(
|
||||
default_factory=lambda: [4,2,0.6,24,8,0.4]
|
||||
) # 记忆构建分布，参数：分布1均值，标准差，权重，分布2均值，标准差，权重
|
||||
memory_ban_words: list = field(
|
||||
default_factory=lambda: ["表情包", "图片", "回复", "聊天记录"]
|
||||
) # 添加新的配置项默认值
|
||||
@@ -230,7 +234,6 @@ class BotConfig:
|
||||
"llm_reasoning",
|
||||
"llm_reasoning_minor",
|
||||
"llm_normal",
|
||||
"llm_normal_minor",
|
||||
"llm_topic_judge",
|
||||
"llm_summary_by_topic",
|
||||
"llm_emotion_judge",
|
||||
@@ -315,6 +318,20 @@ class BotConfig:
|
||||
"memory_forget_percentage", config.memory_forget_percentage
|
||||
)
|
||||
config.memory_compress_rate = memory_config.get("memory_compress_rate", config.memory_compress_rate)
|
||||
if config.INNER_VERSION in SpecifierSet(">=0.0.11"):
|
||||
config.memory_build_distribution = memory_config.get(
|
||||
"memory_build_distribution",
|
||||
config.memory_build_distribution
|
||||
)
|
||||
config.build_memory_sample_num = memory_config.get(
|
||||
"build_memory_sample_num",
|
||||
config.build_memory_sample_num
|
||||
)
|
||||
config.build_memory_sample_length = memory_config.get(
|
||||
"build_memory_sample_length",
|
||||
config.build_memory_sample_length
|
||||
)
|
||||
|
||||
|
||||
def remote(parent: dict):
|
||||
remote_config = parent["remote"]
|
||||
@@ -351,10 +368,10 @@ class BotConfig:
|
||||
|
||||
def others(parent: dict):
|
||||
others_config = parent["others"]
|
||||
config.enable_advance_output = others_config.get("enable_advance_output", config.enable_advance_output)
|
||||
# config.enable_advance_output = others_config.get("enable_advance_output", config.enable_advance_output)
|
||||
config.enable_kuuki_read = others_config.get("enable_kuuki_read", config.enable_kuuki_read)
|
||||
if config.INNER_VERSION in SpecifierSet(">=0.0.7"):
|
||||
config.enable_debug_output = others_config.get("enable_debug_output", config.enable_debug_output)
|
||||
# config.enable_debug_output = others_config.get("enable_debug_output", config.enable_debug_output)
|
||||
config.enable_friend_chat = others_config.get("enable_friend_chat", config.enable_friend_chat)
|
||||
|
||||
# 版本表达式：>=1.0.0,<2.0.0
|
||||
|
||||
@@ -38,9 +38,9 @@ class EmojiManager:
|
||||
|
||||
def __init__(self):
|
||||
self._scan_task = None
|
||||
self.vlm = LLM_request(model=global_config.vlm, temperature=0.3, max_tokens=1000, request_type="image")
|
||||
self.vlm = LLM_request(model=global_config.vlm, temperature=0.3, max_tokens=1000, request_type="emoji")
|
||||
self.llm_emotion_judge = LLM_request(
|
||||
model=global_config.llm_emotion_judge, max_tokens=600, temperature=0.8, request_type="image"
|
||||
model=global_config.llm_emotion_judge, max_tokens=600, temperature=0.8, request_type="emoji"
|
||||
) # 更高的温度，更少的token（后续可以根据情绪来调整温度）
|
||||
|
||||
def _ensure_emoji_dir(self):
|
||||
@@ -111,7 +111,7 @@ class EmojiManager:
|
||||
if not text_for_search:
|
||||
logger.error("无法获取文本的情绪")
|
||||
return None
|
||||
text_embedding = await get_embedding(text_for_search)
|
||||
text_embedding = await get_embedding(text_for_search, request_type="emoji")
|
||||
if not text_embedding:
|
||||
logger.error("无法获取文本的embedding")
|
||||
return None
|
||||
@@ -242,7 +242,33 @@ class EmojiManager:
|
||||
image_hash = hashlib.md5(image_bytes).hexdigest()
|
||||
image_format = Image.open(io.BytesIO(image_bytes)).format.lower()
|
||||
# 检查是否已经注册过
|
||||
existing_emoji = db["emoji"].find_one({"hash": image_hash})
|
||||
existing_emoji_by_path = db["emoji"].find_one({"filename": filename})
|
||||
existing_emoji_by_hash = db["emoji"].find_one({"hash": image_hash})
|
||||
if existing_emoji_by_path and existing_emoji_by_hash:
|
||||
if existing_emoji_by_path["_id"] != existing_emoji_by_hash["_id"]:
|
||||
logger.error(f"[错误] 表情包已存在但记录不一致: {filename}")
|
||||
db.emoji.delete_one({"_id": existing_emoji_by_path["_id"]})
|
||||
db.emoji.update_one(
|
||||
{"_id": existing_emoji_by_hash["_id"]}, {"$set": {"path": image_path, "filename": filename}}
|
||||
)
|
||||
existing_emoji_by_hash["path"] = image_path
|
||||
existing_emoji_by_hash["filename"] = filename
|
||||
existing_emoji = existing_emoji_by_hash
|
||||
elif existing_emoji_by_hash:
|
||||
logger.error(f"[错误] 表情包hash已存在但path不存在: {filename}")
|
||||
db.emoji.update_one(
|
||||
{"_id": existing_emoji_by_hash["_id"]}, {"$set": {"path": image_path, "filename": filename}}
|
||||
)
|
||||
existing_emoji_by_hash["path"] = image_path
|
||||
existing_emoji_by_hash["filename"] = filename
|
||||
existing_emoji = existing_emoji_by_hash
|
||||
elif existing_emoji_by_path:
|
||||
logger.error(f"[错误] 表情包path已存在但hash不存在: {filename}")
|
||||
db.emoji.delete_one({"_id": existing_emoji_by_path["_id"]})
|
||||
existing_emoji = None
|
||||
else:
|
||||
existing_emoji = None
|
||||
|
||||
description = None
|
||||
|
||||
if existing_emoji:
|
||||
@@ -284,7 +310,7 @@ class EmojiManager:
|
||||
logger.info(f"[检查] 表情包检查通过: {check}")
|
||||
|
||||
if description is not None:
|
||||
embedding = await get_embedding(description)
|
||||
embedding = await get_embedding(description, request_type="emoji")
|
||||
# 准备数据库记录
|
||||
emoji_record = {
|
||||
"filename": filename,
|
||||
@@ -366,6 +392,12 @@ class EmojiManager:
|
||||
logger.warning(f"[检查] 发现缺失记录（缺少hash字段），ID: {emoji.get('_id', 'unknown')}")
|
||||
hash = hashlib.md5(open(emoji["path"], "rb").read()).hexdigest()
|
||||
db.emoji.update_one({"_id": emoji["_id"]}, {"$set": {"hash": hash}})
|
||||
else:
|
||||
file_hash = hashlib.md5(open(emoji["path"], "rb").read()).hexdigest()
|
||||
if emoji["hash"] != file_hash:
|
||||
logger.warning(f"[检查] 表情包文件hash不匹配，ID: {emoji.get('_id', 'unknown')}")
|
||||
db.emoji.delete_one({"_id": emoji["_id"]})
|
||||
removed_count += 1
|
||||
|
||||
except Exception as item_error:
|
||||
logger.error(f"[错误] 处理表情包记录时出错: {str(item_error)}")
|
||||
|
||||
@@ -32,11 +32,19 @@ class ResponseGenerator:
|
||||
temperature=0.7,
|
||||
max_tokens=1000,
|
||||
stream=True,
|
||||
request_type="response",
|
||||
)
|
||||
self.model_v3 = LLM_request(
|
||||
model=global_config.llm_normal, temperature=0.7, max_tokens=3000, request_type="response"
|
||||
)
|
||||
self.model_r1_distill = LLM_request(
|
||||
model=global_config.llm_reasoning_minor, temperature=0.7, max_tokens=3000, request_type="response"
|
||||
)
|
||||
self.model_sum = LLM_request(
|
||||
model=global_config.llm_summary_by_topic, temperature=0.7, max_tokens=3000, request_type="relation"
|
||||
)
|
||||
self.model_v3 = LLM_request(model=global_config.llm_normal, temperature=0.7, max_tokens=3000)
|
||||
self.model_r1_distill = LLM_request(model=global_config.llm_reasoning_minor, temperature=0.7, max_tokens=3000)
|
||||
self.model_v25 = LLM_request(model=global_config.llm_normal_minor, temperature=0.7, max_tokens=3000)
|
||||
self.current_model_type = "r1" # 默认使用 R1
|
||||
self.current_model_name = "unknown model"
|
||||
|
||||
async def generate_response(self, message: MessageThinking) -> Optional[Union[str, List[str]]]:
|
||||
"""根据当前模型类型选择对应的生成函数"""
|
||||
@@ -107,7 +115,7 @@ class ResponseGenerator:
|
||||
|
||||
# 生成回复
|
||||
try:
|
||||
content, reasoning_content = await model.generate_response(prompt)
|
||||
content, reasoning_content, self.current_model_name = await model.generate_response(prompt)
|
||||
except Exception:
|
||||
logger.exception("生成回复时出错")
|
||||
return None
|
||||
@@ -144,7 +152,7 @@ class ResponseGenerator:
|
||||
"chat_id": message.chat_stream.stream_id,
|
||||
"user": sender_name,
|
||||
"message": message.processed_plain_text,
|
||||
"model": self.current_model_type,
|
||||
"model": self.current_model_name,
|
||||
# 'reasoning_check': reasoning_content_check,
|
||||
# 'response_check': content_check,
|
||||
"reasoning": reasoning_content,
|
||||
@@ -174,7 +182,7 @@ class ResponseGenerator:
|
||||
"""
|
||||
|
||||
# 调用模型生成结果
|
||||
result, _ = await self.model_v25.generate_response(prompt)
|
||||
result, _, _ = await self.model_sum.generate_response(prompt)
|
||||
result = result.strip()
|
||||
|
||||
# 解析模型输出的结果
|
||||
@@ -215,7 +223,7 @@ class InitiativeMessageGenerate:
|
||||
topic_select_prompt, dots_for_select, prompt_template = prompt_builder._build_initiative_prompt_select(
|
||||
message.group_id
|
||||
)
|
||||
content_select, reasoning = self.model_v3.generate_response(topic_select_prompt)
|
||||
content_select, reasoning, _ = self.model_v3.generate_response(topic_select_prompt)
|
||||
logger.debug(f"{content_select} {reasoning}")
|
||||
topics_list = [dot[0] for dot in dots_for_select]
|
||||
if content_select:
|
||||
@@ -226,7 +234,7 @@ class InitiativeMessageGenerate:
|
||||
else:
|
||||
return None
|
||||
prompt_check, memory = prompt_builder._build_initiative_prompt_check(select_dot[1], prompt_template)
|
||||
content_check, reasoning_check = self.model_v3.generate_response(prompt_check)
|
||||
content_check, reasoning_check, _ = self.model_v3.generate_response(prompt_check)
|
||||
logger.info(f"{content_check} {reasoning_check}")
|
||||
if "yes" not in content_check.lower():
|
||||
return None
|
||||
|
||||
@@ -220,7 +220,7 @@ class MessageManager:
|
||||
|
||||
message_timeout = container.get_timeout_messages()
|
||||
if message_timeout:
|
||||
logger.warning(f"发现{len(message_timeout)}条超时消息")
|
||||
logger.debug(f"发现{len(message_timeout)}条超时消息")
|
||||
for msg in message_timeout:
|
||||
if msg == message_earliest:
|
||||
continue
|
||||
|
||||
@@ -141,21 +141,21 @@ class PromptBuilder:
|
||||
logger.debug(f"知识检索耗时: {(end_time - start_time):.3f}秒")
|
||||
|
||||
prompt = f"""
|
||||
今天是{current_date}，现在是{current_time}，你今天的日程是：\
|
||||
`<schedule>`\n
|
||||
{bot_schedule.today_schedule}\n
|
||||
`</schedule>`\n
|
||||
{prompt_info}\n
|
||||
{memory_prompt}\n
|
||||
{chat_target}\n
|
||||
{chat_talking_prompt}\n
|
||||
现在"{sender_name}"说的:\n
|
||||
`<UserMessage>`\n
|
||||
{message_txt}\n
|
||||
`</UserMessage>`\n
|
||||
今天是{current_date}，现在是{current_time}，你今天的日程是：
|
||||
`<schedule>`
|
||||
{bot_schedule.today_schedule}
|
||||
`</schedule>`
|
||||
{prompt_info}
|
||||
{memory_prompt}
|
||||
{chat_target}
|
||||
{chat_talking_prompt}
|
||||
现在"{sender_name}"说的:
|
||||
`<UserMessage>`
|
||||
{message_txt}
|
||||
`</UserMessage>`
|
||||
引起了你的注意,{relation_prompt_all}{mood_prompt}\n
|
||||
`<MainRule>`
|
||||
你的网名叫{global_config.BOT_NICKNAME}，{prompt_personality}。
|
||||
你的网名叫{global_config.BOT_NICKNAME}，有人也叫你{"/".join(global_config.BOT_ALIAS_NAMES)}，{prompt_personality},{prompt_personality}。
|
||||
正在{bot_schedule_now_activity}的你同时也在一边{chat_target_2},现在请你读读之前的聊天记录，然后给出日常且口语化的回复，平淡一些，
|
||||
尽量简短一些。{keywords_reaction_prompt}请注意把握聊天内容，不要刻意突出自身学科背景，不要回复的太有条理，可以有个性。
|
||||
{prompt_ger}
|
||||
|
||||
@@ -33,7 +33,7 @@ class TopicIdentifier:
|
||||
消息内容：{text}"""
|
||||
|
||||
# 使用 LLM_request 类进行请求
|
||||
topic, _ = await self.llm_topic_judge.generate_response(prompt)
|
||||
topic, _, _ = await self.llm_topic_judge.generate_response(prompt)
|
||||
|
||||
if not topic:
|
||||
logger.error("LLM API 返回为空")
|
||||
|
||||
@@ -55,9 +55,9 @@ def is_mentioned_bot_in_message(message: MessageRecv) -> bool:
|
||||
return False
|
||||
|
||||
|
||||
async def get_embedding(text):
|
||||
async def get_embedding(text, request_type="embedding"):
|
||||
"""获取文本的embedding向量"""
|
||||
llm = LLM_request(model=global_config.embedding, request_type="embedding")
|
||||
llm = LLM_request(model=global_config.embedding, request_type=request_type)
|
||||
# return llm.get_embedding_sync(text)
|
||||
return await llm.get_embedding(text)
|
||||
|
||||
@@ -76,18 +76,11 @@ def calculate_information_content(text):
|
||||
|
||||
|
||||
def get_closest_chat_from_db(length: int, timestamp: str):
|
||||
"""从数据库中获取最接近指定时间戳的聊天记录
|
||||
|
||||
Args:
|
||||
length: 要获取的消息数量
|
||||
timestamp: 时间戳
|
||||
|
||||
Returns:
|
||||
list: 消息记录列表，每个记录包含时间和文本信息
|
||||
"""
|
||||
# print(f"获取最接近指定时间戳的聊天记录，长度: {length}, 时间戳: {timestamp}")
|
||||
# print(f"当前时间: {timestamp},转换后时间: {time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(timestamp))}")
|
||||
chat_records = []
|
||||
closest_record = db.messages.find_one({"time": {"$lte": timestamp}}, sort=[("time", -1)])
|
||||
|
||||
# print(f"最接近的记录: {closest_record}")
|
||||
if closest_record:
|
||||
closest_time = closest_record["time"]
|
||||
chat_id = closest_record["chat_id"] # 获取chat_id
|
||||
@@ -102,7 +95,9 @@ def get_closest_chat_from_db(length: int, timestamp: str):
|
||||
.sort("time", 1)
|
||||
.limit(length)
|
||||
)
|
||||
|
||||
# print(f"获取到的记录: {chat_records}")
|
||||
length = len(chat_records)
|
||||
# print(f"获取到的记录长度: {length}")
|
||||
# 转换记录格式
|
||||
formatted_records = []
|
||||
for record in chat_records:
|
||||
@@ -319,7 +314,7 @@ def split_into_sentences_w_remove_punctuation(text: str) -> List[str]:
|
||||
sentence = sentence.replace("，", " ").replace(",", " ")
|
||||
sentences_done.append(sentence)
|
||||
|
||||
logger.info(f"处理后的句子: {sentences_done}")
|
||||
logger.debug(f"处理后的句子: {sentences_done}")
|
||||
return sentences_done
|
||||
|
||||
|
||||
|
||||
@@ -112,7 +112,7 @@ class ImageManager:
|
||||
# 查询缓存的描述
|
||||
cached_description = self._get_description_from_db(image_hash, "emoji")
|
||||
if cached_description:
|
||||
logger.info(f"缓存表情包描述: {cached_description}")
|
||||
logger.debug(f"缓存表情包描述: {cached_description}")
|
||||
return f"[表情包：{cached_description}]"
|
||||
|
||||
# 调用AI获取描述
|
||||
@@ -184,7 +184,7 @@ class ImageManager:
|
||||
logger.warning(f"虽然生成了描述，但是找到缓存图片描述 {cached_description}")
|
||||
return f"[图片：{cached_description}]"
|
||||
|
||||
logger.info(f"描述是{description}")
|
||||
logger.debug(f"描述是{description}")
|
||||
|
||||
if description is None:
|
||||
logger.warning("AI未能生成图片描述")
|
||||
|
||||
@@ -18,6 +18,7 @@ from ..chat.utils import (
|
||||
)
|
||||
from ..models.utils_model import LLM_request
|
||||
from src.common.logger import get_module_logger, LogConfig, MEMORY_STYLE_CONFIG
|
||||
from src.plugins.memory_system.sample_distribution import MemoryBuildScheduler
|
||||
|
||||
# 定义日志配置
|
||||
memory_config = LogConfig(
|
||||
@@ -25,6 +26,11 @@ memory_config = LogConfig(
|
||||
console_format=MEMORY_STYLE_CONFIG["console_format"],
|
||||
file_format=MEMORY_STYLE_CONFIG["file_format"],
|
||||
)
|
||||
# print(f"memory_config: {memory_config}")
|
||||
# print(f"MEMORY_STYLE_CONFIG: {MEMORY_STYLE_CONFIG}")
|
||||
# print(f"MEMORY_STYLE_CONFIG['console_format']: {MEMORY_STYLE_CONFIG['console_format']}")
|
||||
# print(f"MEMORY_STYLE_CONFIG['file_format']: {MEMORY_STYLE_CONFIG['file_format']}")
|
||||
|
||||
|
||||
logger = get_module_logger("memory_system", config=memory_config)
|
||||
|
||||
@@ -168,9 +174,9 @@ class Memory_graph:
|
||||
class Hippocampus:
|
||||
def __init__(self, memory_graph: Memory_graph):
|
||||
self.memory_graph = memory_graph
|
||||
self.llm_topic_judge = LLM_request(model=global_config.llm_topic_judge, temperature=0.5, request_type="topic")
|
||||
self.llm_topic_judge = LLM_request(model=global_config.llm_topic_judge, temperature=0.5, request_type="memory")
|
||||
self.llm_summary_by_topic = LLM_request(
|
||||
model=global_config.llm_summary_by_topic, temperature=0.5, request_type="topic"
|
||||
model=global_config.llm_summary_by_topic, temperature=0.5, request_type="memory"
|
||||
)
|
||||
|
||||
def get_all_node_names(self) -> list:
|
||||
@@ -195,25 +201,17 @@ class Hippocampus:
|
||||
return hash(f"{nodes[0]}:{nodes[1]}")
|
||||
|
||||
def random_get_msg_snippet(self, target_timestamp: float, chat_size: int, max_memorized_time_per_msg: int) -> list:
|
||||
"""随机抽取一段时间内的消息片段
|
||||
Args:
|
||||
- target_timestamp: 目标时间戳
|
||||
- chat_size: 抽取的消息数量
|
||||
- max_memorized_time_per_msg: 每条消息的最大记忆次数
|
||||
|
||||
Returns:
|
||||
- list: 抽取出的消息记录列表
|
||||
|
||||
"""
|
||||
try_count = 0
|
||||
# 最多尝试三次抽取
|
||||
# 最多尝试2次抽取
|
||||
while try_count < 3:
|
||||
messages = get_closest_chat_from_db(length=chat_size, timestamp=target_timestamp)
|
||||
if messages:
|
||||
# print(f"抽取到的消息: {messages}")
|
||||
# 检查messages是否均没有达到记忆次数限制
|
||||
for message in messages:
|
||||
if message["memorized_times"] >= max_memorized_time_per_msg:
|
||||
messages = None
|
||||
# print(f"抽取到的消息提取次数达到限制，跳过")
|
||||
break
|
||||
if messages:
|
||||
# 成功抽取短期消息样本
|
||||
@@ -224,63 +222,48 @@ class Hippocampus:
|
||||
)
|
||||
return messages
|
||||
try_count += 1
|
||||
# 三次尝试均失败
|
||||
return None
|
||||
|
||||
def get_memory_sample(self, chat_size=20, time_frequency=None):
|
||||
"""获取记忆样本
|
||||
|
||||
Returns:
|
||||
list: 消息记录列表，每个元素是一个消息记录字典列表
|
||||
"""
|
||||
def get_memory_sample(self):
|
||||
# 硬编码：每条消息最大记忆次数
|
||||
# 如有需求可写入global_config
|
||||
if time_frequency is None:
|
||||
time_frequency = {"near": 2, "mid": 4, "far": 3}
|
||||
max_memorized_time_per_msg = 3
|
||||
|
||||
current_timestamp = datetime.datetime.now().timestamp()
|
||||
# 创建双峰分布的记忆调度器
|
||||
scheduler = MemoryBuildScheduler(
|
||||
n_hours1=global_config.memory_build_distribution[0], # 第一个分布均值（4小时前）
|
||||
std_hours1=global_config.memory_build_distribution[1], # 第一个分布标准差
|
||||
weight1=global_config.memory_build_distribution[2], # 第一个分布权重 60%
|
||||
n_hours2=global_config.memory_build_distribution[3], # 第二个分布均值（24小时前）
|
||||
std_hours2=global_config.memory_build_distribution[4], # 第二个分布标准差
|
||||
weight2=global_config.memory_build_distribution[5], # 第二个分布权重 40%
|
||||
total_samples=global_config.build_memory_sample_num # 总共生成10个时间点
|
||||
)
|
||||
|
||||
# 生成时间戳数组
|
||||
timestamps = scheduler.get_timestamp_array()
|
||||
# logger.debug(f"生成的时间戳数组: {timestamps}")
|
||||
# print(f"生成的时间戳数组: {timestamps}")
|
||||
# print(f"时间戳的实际时间: {[time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(ts)) for ts in timestamps]}")
|
||||
logger.info(f"回忆往事: {[time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(ts)) for ts in timestamps]}")
|
||||
chat_samples = []
|
||||
|
||||
# 短期：1h 中期：4h 长期：24h
|
||||
logger.debug("正在抽取短期消息样本")
|
||||
for i in range(time_frequency.get("near")):
|
||||
random_time = current_timestamp - random.randint(1, 3600)
|
||||
messages = self.random_get_msg_snippet(random_time, chat_size, max_memorized_time_per_msg)
|
||||
for timestamp in timestamps:
|
||||
messages = self.random_get_msg_snippet(
|
||||
timestamp,
|
||||
global_config.build_memory_sample_length,
|
||||
max_memorized_time_per_msg
|
||||
)
|
||||
if messages:
|
||||
logger.debug(f"成功抽取短期消息样本{len(messages)}条")
|
||||
time_diff = (datetime.datetime.now().timestamp() - timestamp) / 3600
|
||||
logger.debug(f"成功抽取 {time_diff:.1f} 小时前的消息样本，共{len(messages)}条")
|
||||
# print(f"成功抽取 {time_diff:.1f} 小时前的消息样本，共{len(messages)}条")
|
||||
chat_samples.append(messages)
|
||||
else:
|
||||
logger.warning(f"第{i}次短期消息样本抽取失败")
|
||||
|
||||
logger.debug("正在抽取中期消息样本")
|
||||
for i in range(time_frequency.get("mid")):
|
||||
random_time = current_timestamp - random.randint(3600, 3600 * 4)
|
||||
messages = self.random_get_msg_snippet(random_time, chat_size, max_memorized_time_per_msg)
|
||||
if messages:
|
||||
logger.debug(f"成功抽取中期消息样本{len(messages)}条")
|
||||
chat_samples.append(messages)
|
||||
else:
|
||||
logger.warning(f"第{i}次中期消息样本抽取失败")
|
||||
|
||||
logger.debug("正在抽取长期消息样本")
|
||||
for i in range(time_frequency.get("far")):
|
||||
random_time = current_timestamp - random.randint(3600 * 4, 3600 * 24)
|
||||
messages = self.random_get_msg_snippet(random_time, chat_size, max_memorized_time_per_msg)
|
||||
if messages:
|
||||
logger.debug(f"成功抽取长期消息样本{len(messages)}条")
|
||||
chat_samples.append(messages)
|
||||
else:
|
||||
logger.warning(f"第{i}次长期消息样本抽取失败")
|
||||
logger.debug(f"时间戳 {timestamp} 的消息样本抽取失败")
|
||||
|
||||
return chat_samples
|
||||
|
||||
async def memory_compress(self, messages: list, compress_rate=0.1):
|
||||
"""压缩消息记录为记忆
|
||||
|
||||
Returns:
|
||||
tuple: (压缩记忆集合, 相似主题字典)
|
||||
"""
|
||||
if not messages:
|
||||
return set(), {}
|
||||
|
||||
@@ -313,15 +296,23 @@ class Hippocampus:
|
||||
topics_response = await self.llm_topic_judge.generate_response(self.find_topic_llm(input_text, topic_num))
|
||||
|
||||
# 过滤topics
|
||||
# 从配置文件获取需要过滤的关键词列表
|
||||
filter_keywords = global_config.memory_ban_words
|
||||
|
||||
# 将topics_response[0]中的中文逗号、顿号、空格都替换成英文逗号
|
||||
# 然后按逗号分割成列表,并去除每个topic前后的空白字符
|
||||
topics = [
|
||||
topic.strip()
|
||||
for topic in topics_response[0].replace("，", ",").replace("、", ",").replace(" ", ",").split(",")
|
||||
if topic.strip()
|
||||
]
|
||||
|
||||
# 过滤掉包含禁用关键词的topic
|
||||
# any()检查topic中是否包含任何一个filter_keywords中的关键词
|
||||
# 只保留不包含禁用关键词的topic
|
||||
filtered_topics = [topic for topic in topics if not any(keyword in topic for keyword in filter_keywords)]
|
||||
|
||||
logger.info(f"过滤后话题: {filtered_topics}")
|
||||
logger.debug(f"过滤后话题: {filtered_topics}")
|
||||
|
||||
# 创建所有话题的请求任务
|
||||
tasks = []
|
||||
@@ -331,31 +322,42 @@ class Hippocampus:
|
||||
tasks.append((topic.strip(), task))
|
||||
|
||||
# 等待所有任务完成
|
||||
compressed_memory = set()
|
||||
# 初始化压缩后的记忆集合和相似主题字典
|
||||
compressed_memory = set() # 存储压缩后的(主题,内容)元组
|
||||
similar_topics_dict = {} # 存储每个话题的相似主题列表
|
||||
|
||||
# 遍历每个主题及其对应的LLM任务
|
||||
for topic, task in tasks:
|
||||
response = await task
|
||||
if response:
|
||||
# 将主题和LLM生成的内容添加到压缩记忆中
|
||||
compressed_memory.add((topic, response[0]))
|
||||
# 为每个话题查找相似的已存在主题
|
||||
|
||||
# 为当前主题寻找相似的已存在主题
|
||||
existing_topics = list(self.memory_graph.G.nodes())
|
||||
similar_topics = []
|
||||
|
||||
# 计算当前主题与每个已存在主题的相似度
|
||||
for existing_topic in existing_topics:
|
||||
# 使用jieba分词,将主题转换为词集合
|
||||
topic_words = set(jieba.cut(topic))
|
||||
existing_words = set(jieba.cut(existing_topic))
|
||||
|
||||
all_words = topic_words | existing_words
|
||||
v1 = [1 if word in topic_words else 0 for word in all_words]
|
||||
v2 = [1 if word in existing_words else 0 for word in all_words]
|
||||
# 构建词向量用于计算余弦相似度
|
||||
all_words = topic_words | existing_words # 所有不重复的词
|
||||
v1 = [1 if word in topic_words else 0 for word in all_words] # 当前主题的词向量
|
||||
v2 = [1 if word in existing_words else 0 for word in all_words] # 已存在主题的词向量
|
||||
|
||||
# 计算余弦相似度
|
||||
similarity = cosine_similarity(v1, v2)
|
||||
|
||||
if similarity >= 0.6:
|
||||
# 如果相似度超过阈值,添加到相似主题列表
|
||||
if similarity >= 0.7:
|
||||
similar_topics.append((existing_topic, similarity))
|
||||
|
||||
# 按相似度降序排序,只保留前3个最相似的主题
|
||||
similar_topics.sort(key=lambda x: x[1], reverse=True)
|
||||
similar_topics = similar_topics[:5]
|
||||
similar_topics = similar_topics[:3]
|
||||
similar_topics_dict[topic] = similar_topics
|
||||
|
||||
return compressed_memory, similar_topics_dict
|
||||
@@ -372,10 +374,13 @@ class Hippocampus:
|
||||
)
|
||||
return topic_num
|
||||
|
||||
async def operation_build_memory(self, chat_size=20):
|
||||
time_frequency = {"near": 1, "mid": 4, "far": 4}
|
||||
memory_samples = self.get_memory_sample(chat_size, time_frequency)
|
||||
|
||||
async def operation_build_memory(self):
|
||||
logger.debug("------------------------------------开始构建记忆--------------------------------------")
|
||||
start_time = time.time()
|
||||
memory_samples = self.get_memory_sample()
|
||||
all_added_nodes = []
|
||||
all_connected_nodes = []
|
||||
all_added_edges = []
|
||||
for i, messages in enumerate(memory_samples, 1):
|
||||
all_topics = []
|
||||
# 加载进度可视化
|
||||
@@ -387,12 +392,14 @@ class Hippocampus:
|
||||
|
||||
compress_rate = global_config.memory_compress_rate
|
||||
compressed_memory, similar_topics_dict = await self.memory_compress(messages, compress_rate)
|
||||
logger.info(f"压缩后记忆数量: {len(compressed_memory)}，似曾相识的话题: {len(similar_topics_dict)}")
|
||||
logger.debug(f"压缩后记忆数量: {compressed_memory}，似曾相识的话题: {similar_topics_dict}")
|
||||
|
||||
current_time = datetime.datetime.now().timestamp()
|
||||
logger.debug(f"添加节点: {', '.join(topic for topic, _ in compressed_memory)}")
|
||||
all_added_nodes.extend(topic for topic, _ in compressed_memory)
|
||||
# all_connected_nodes.extend(topic for topic, _ in similar_topics_dict)
|
||||
|
||||
for topic, memory in compressed_memory:
|
||||
logger.info(f"添加节点: {topic}")
|
||||
self.memory_graph.add_dot(topic, memory)
|
||||
all_topics.append(topic)
|
||||
|
||||
@@ -402,7 +409,13 @@ class Hippocampus:
|
||||
for similar_topic, similarity in similar_topics:
|
||||
if topic != similar_topic:
|
||||
strength = int(similarity * 10)
|
||||
logger.info(f"连接相似节点: {topic} 和 {similar_topic} (强度: {strength})")
|
||||
|
||||
logger.debug(f"连接相似节点: {topic} 和 {similar_topic} (强度: {strength})")
|
||||
all_added_edges.append(f"{topic}-{similar_topic}")
|
||||
|
||||
all_connected_nodes.append(topic)
|
||||
all_connected_nodes.append(similar_topic)
|
||||
|
||||
self.memory_graph.G.add_edge(
|
||||
topic,
|
||||
similar_topic,
|
||||
@@ -414,11 +427,22 @@ class Hippocampus:
|
||||
# 连接同批次的相关话题
|
||||
for i in range(len(all_topics)):
|
||||
for j in range(i + 1, len(all_topics)):
|
||||
logger.info(f"连接同批次节点: {all_topics[i]} 和 {all_topics[j]}")
|
||||
logger.debug(f"连接同批次节点: {all_topics[i]} 和 {all_topics[j]}")
|
||||
all_added_edges.append(f"{all_topics[i]}-{all_topics[j]}")
|
||||
self.memory_graph.connect_dot(all_topics[i], all_topics[j])
|
||||
|
||||
logger.success(f"更新记忆: {', '.join(all_added_nodes)}")
|
||||
logger.debug(f"强化连接: {', '.join(all_added_edges)}")
|
||||
logger.info(f"强化连接节点: {', '.join(all_connected_nodes)}")
|
||||
# logger.success(f"强化连接: {', '.join(all_added_edges)}")
|
||||
self.sync_memory_to_db()
|
||||
|
||||
end_time = time.time()
|
||||
logger.success(
|
||||
f"--------------------------记忆构建完成：耗时: {end_time - start_time:.2f} "
|
||||
"秒--------------------------"
|
||||
)
|
||||
|
||||
def sync_memory_to_db(self):
|
||||
"""检查并同步内存中的图结构与数据库"""
|
||||
# 获取数据库中所有节点和内存中所有节点
|
||||
@@ -844,10 +868,9 @@ 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"""
|
||||
logger.info(f"识别主题: {await self._identify_topics(text)}")
|
||||
|
||||
# 识别主题
|
||||
identified_topics = await self._identify_topics(text)
|
||||
|
||||
if not identified_topics:
|
||||
return 0
|
||||
|
||||
@@ -908,7 +931,8 @@ class Hippocampus:
|
||||
|
||||
# 计算最终激活值
|
||||
activation = int((topic_match + average_similarities) / 2 * 100)
|
||||
logger.info(f"匹配率: {topic_match:.3f}, 平均相似度: {average_similarities:.3f}, 激活值: {activation}")
|
||||
|
||||
logger.info(f"识别主题: {identified_topics}, 匹配率: {topic_match:.3f}, 激活值: {activation}")
|
||||
|
||||
return activation
|
||||
|
||||
|
||||
@@ -7,7 +7,6 @@ import sys
|
||||
import time
|
||||
from collections import Counter
|
||||
from pathlib import Path
|
||||
|
||||
import matplotlib.pyplot as plt
|
||||
import networkx as nx
|
||||
from dotenv import load_dotenv
|
||||
@@ -16,7 +15,6 @@ sys.path.insert(0, sys.path[0]+"/../")
|
||||
sys.path.insert(0, sys.path[0]+"/../")
|
||||
sys.path.insert(0, sys.path[0]+"/../")
|
||||
sys.path.insert(0, sys.path[0]+"/../")
|
||||
print(sys.path)
|
||||
from src.common.logger import get_module_logger
|
||||
import jieba
|
||||
|
||||
@@ -25,6 +23,7 @@ import jieba
|
||||
root_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
|
||||
sys.path.append(root_path)
|
||||
|
||||
from src.common.logger import get_module_logger # noqa: E402
|
||||
from src.common.database import db # noqa E402
|
||||
from src.plugins.memory_system.offline_llm import LLMModel # noqa E402
|
||||
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
170
src/plugins/memory_system/sample_distribution.py
Normal file
170
src/plugins/memory_system/sample_distribution.py
Normal file
@@ -0,0 +1,170 @@
|
||||
import numpy as np
|
||||
from scipy import stats
|
||||
from datetime import datetime, timedelta
|
||||
|
||||
class DistributionVisualizer:
|
||||
def __init__(self, mean=0, std=1, skewness=0, sample_size=10):
|
||||
"""
|
||||
初始化分布可视化器
|
||||
|
||||
参数:
|
||||
mean (float): 期望均值
|
||||
std (float): 标准差
|
||||
skewness (float): 偏度
|
||||
sample_size (int): 样本大小
|
||||
"""
|
||||
self.mean = mean
|
||||
self.std = std
|
||||
self.skewness = skewness
|
||||
self.sample_size = sample_size
|
||||
self.samples = None
|
||||
|
||||
def generate_samples(self):
|
||||
"""生成具有指定参数的样本"""
|
||||
if self.skewness == 0:
|
||||
# 对于无偏度的情况，直接使用正态分布
|
||||
self.samples = np.random.normal(loc=self.mean, scale=self.std, size=self.sample_size)
|
||||
else:
|
||||
# 使用 scipy.stats 生成具有偏度的分布
|
||||
self.samples = stats.skewnorm.rvs(a=self.skewness,
|
||||
loc=self.mean,
|
||||
scale=self.std,
|
||||
size=self.sample_size)
|
||||
|
||||
def get_weighted_samples(self):
|
||||
"""获取加权后的样本数列"""
|
||||
if self.samples is None:
|
||||
self.generate_samples()
|
||||
# 将样本值乘以样本大小
|
||||
return self.samples * self.sample_size
|
||||
|
||||
def get_statistics(self):
|
||||
"""获取分布的统计信息"""
|
||||
if self.samples is None:
|
||||
self.generate_samples()
|
||||
|
||||
return {
|
||||
"均值": np.mean(self.samples),
|
||||
"标准差": np.std(self.samples),
|
||||
"实际偏度": stats.skew(self.samples)
|
||||
}
|
||||
|
||||
class MemoryBuildScheduler:
|
||||
def __init__(self,
|
||||
n_hours1, std_hours1, weight1,
|
||||
n_hours2, std_hours2, weight2,
|
||||
total_samples=50):
|
||||
"""
|
||||
初始化记忆构建调度器
|
||||
|
||||
参数:
|
||||
n_hours1 (float): 第一个分布的均值（距离现在的小时数）
|
||||
std_hours1 (float): 第一个分布的标准差（小时）
|
||||
weight1 (float): 第一个分布的权重
|
||||
n_hours2 (float): 第二个分布的均值（距离现在的小时数）
|
||||
std_hours2 (float): 第二个分布的标准差（小时）
|
||||
weight2 (float): 第二个分布的权重
|
||||
total_samples (int): 要生成的总时间点数量
|
||||
"""
|
||||
# 归一化权重
|
||||
total_weight = weight1 + weight2
|
||||
self.weight1 = weight1 / total_weight
|
||||
self.weight2 = weight2 / total_weight
|
||||
|
||||
self.n_hours1 = n_hours1
|
||||
self.std_hours1 = std_hours1
|
||||
self.n_hours2 = n_hours2
|
||||
self.std_hours2 = std_hours2
|
||||
self.total_samples = total_samples
|
||||
self.base_time = datetime.now()
|
||||
|
||||
def generate_time_samples(self):
|
||||
"""生成混合分布的时间采样点"""
|
||||
# 根据权重计算每个分布的样本数
|
||||
samples1 = int(self.total_samples * self.weight1)
|
||||
samples2 = self.total_samples - samples1
|
||||
|
||||
# 生成两个正态分布的小时偏移
|
||||
hours_offset1 = np.random.normal(
|
||||
loc=self.n_hours1,
|
||||
scale=self.std_hours1,
|
||||
size=samples1
|
||||
)
|
||||
|
||||
hours_offset2 = np.random.normal(
|
||||
loc=self.n_hours2,
|
||||
scale=self.std_hours2,
|
||||
size=samples2
|
||||
)
|
||||
|
||||
# 合并两个分布的偏移
|
||||
hours_offset = np.concatenate([hours_offset1, hours_offset2])
|
||||
|
||||
# 将偏移转换为实际时间戳（使用绝对值确保时间点在过去）
|
||||
timestamps = [self.base_time - timedelta(hours=abs(offset)) for offset in hours_offset]
|
||||
|
||||
# 按时间排序（从最早到最近）
|
||||
return sorted(timestamps)
|
||||
|
||||
def get_timestamp_array(self):
|
||||
"""返回时间戳数组"""
|
||||
timestamps = self.generate_time_samples()
|
||||
return [int(t.timestamp()) for t in timestamps]
|
||||
|
||||
def print_time_samples(timestamps, show_distribution=True):
|
||||
"""打印时间样本和分布信息"""
|
||||
print(f"\n生成的{len(timestamps)}个时间点分布：")
|
||||
print("序号".ljust(5), "时间戳".ljust(25), "距现在（小时）")
|
||||
print("-" * 50)
|
||||
|
||||
now = datetime.now()
|
||||
time_diffs = []
|
||||
|
||||
for i, timestamp in enumerate(timestamps, 1):
|
||||
hours_diff = (now - timestamp).total_seconds() / 3600
|
||||
time_diffs.append(hours_diff)
|
||||
print(f"{str(i).ljust(5)} {timestamp.strftime('%Y-%m-%d %H:%M:%S').ljust(25)} {hours_diff:.2f}")
|
||||
|
||||
# 打印统计信息
|
||||
print("\n统计信息：")
|
||||
print(f"平均时间偏移：{np.mean(time_diffs):.2f}小时")
|
||||
print(f"标准差：{np.std(time_diffs):.2f}小时")
|
||||
print(f"最早时间：{min(timestamps).strftime('%Y-%m-%d %H:%M:%S')} ({max(time_diffs):.2f}小时前)")
|
||||
print(f"最近时间：{max(timestamps).strftime('%Y-%m-%d %H:%M:%S')} ({min(time_diffs):.2f}小时前)")
|
||||
|
||||
if show_distribution:
|
||||
# 计算时间分布的直方图
|
||||
hist, bins = np.histogram(time_diffs, bins=40)
|
||||
print("\n时间分布（每个*代表一个时间点）：")
|
||||
for i in range(len(hist)):
|
||||
if hist[i] > 0:
|
||||
print(f"{bins[i]:6.1f}-{bins[i+1]:6.1f}小时: {'*' * int(hist[i])}")
|
||||
|
||||
# 使用示例
|
||||
if __name__ == "__main__":
|
||||
# 创建一个双峰分布的记忆调度器
|
||||
scheduler = MemoryBuildScheduler(
|
||||
n_hours1=12, # 第一个分布均值（12小时前）
|
||||
std_hours1=8, # 第一个分布标准差
|
||||
weight1=0.7, # 第一个分布权重 70%
|
||||
n_hours2=36, # 第二个分布均值（36小时前）
|
||||
std_hours2=24, # 第二个分布标准差
|
||||
weight2=0.3, # 第二个分布权重 30%
|
||||
total_samples=50 # 总共生成50个时间点
|
||||
)
|
||||
|
||||
# 生成时间分布
|
||||
timestamps = scheduler.generate_time_samples()
|
||||
|
||||
# 打印结果，包含分布可视化
|
||||
print_time_samples(timestamps, show_distribution=True)
|
||||
|
||||
# 打印时间戳数组
|
||||
timestamp_array = scheduler.get_timestamp_array()
|
||||
print("\n时间戳数组（Unix时间戳）：")
|
||||
print("[", end="")
|
||||
for i, ts in enumerate(timestamp_array):
|
||||
if i > 0:
|
||||
print(", ", end="")
|
||||
print(ts, end="")
|
||||
print("]")
|
||||
@@ -524,11 +524,11 @@ class LLM_request:
|
||||
return {"Authorization": f"Bearer {self.api_key}", "Content-Type": "application/json"}
|
||||
# 防止小朋友们截图自己的key
|
||||
|
||||
async def generate_response(self, prompt: str) -> Tuple[str, str]:
|
||||
async def generate_response(self, prompt: str) -> Tuple[str, str, str]:
|
||||
"""根据输入的提示生成模型的异步响应"""
|
||||
|
||||
content, reasoning_content = await self._execute_request(endpoint="/chat/completions", prompt=prompt)
|
||||
return content, reasoning_content
|
||||
return content, reasoning_content, self.model_name
|
||||
|
||||
async def generate_response_for_image(self, prompt: str, image_base64: str, image_format: str) -> Tuple[str, str]:
|
||||
"""根据输入的提示和图片生成模型的异步响应"""
|
||||
@@ -583,7 +583,8 @@ class LLM_request:
|
||||
completion_tokens=completion_tokens,
|
||||
total_tokens=total_tokens,
|
||||
user_id="system", # 可以根据需要修改 user_id
|
||||
request_type="embedding", # 请求类型为 embedding
|
||||
# request_type="embedding", # 请求类型为 embedding
|
||||
request_type=self.request_type, # 请求类型为 text
|
||||
endpoint="/embeddings", # API 端点
|
||||
)
|
||||
return result["data"][0].get("embedding", None)
|
||||
|
||||
@@ -15,10 +15,7 @@ env_path = project_root / ".env.prod"
|
||||
root_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
|
||||
sys.path.append(root_path)
|
||||
|
||||
from src.plugins.personality.scene import get_scene_by_factor,get_all_scenes,PERSONALITY_SCENES
|
||||
from src.plugins.personality.questionnaire import PERSONALITY_QUESTIONS,FACTOR_DESCRIPTIONS
|
||||
from src.plugins.personality.offline_llm import LLMModel
|
||||
|
||||
from src.plugins.personality.questionnaire import PERSONALITY_QUESTIONS, FACTOR_DESCRIPTIONS # noqa: E402
|
||||
|
||||
|
||||
class BigFiveTest:
|
||||
@@ -39,7 +36,7 @@ class BigFiveTest:
|
||||
print("\n请认真阅读每个描述，选择最符合您实际情况的选项。\n")
|
||||
|
||||
# 创建题目序号到题目的映射
|
||||
questions_map = {q['id']: q for q in self.questions}
|
||||
questions_map = {q["id"]: q for q in self.questions}
|
||||
|
||||
# 获取所有题目ID并随机打乱顺序
|
||||
question_ids = list(questions_map.keys())
|
||||
@@ -67,35 +64,25 @@ class BigFiveTest:
|
||||
def calculate_scores(self, answers):
|
||||
"""计算各维度得分"""
|
||||
results = {}
|
||||
factor_questions = {
|
||||
"外向性": [],
|
||||
"神经质": [],
|
||||
"严谨性": [],
|
||||
"开放性": [],
|
||||
"宜人性": []
|
||||
}
|
||||
factor_questions = {"外向性": [], "神经质": [], "严谨性": [], "开放性": [], "宜人性": []}
|
||||
|
||||
# 将题目按因子分类
|
||||
for q in self.questions:
|
||||
factor_questions[q['factor']].append(q)
|
||||
factor_questions[q["factor"]].append(q)
|
||||
|
||||
# 计算每个维度的得分
|
||||
for factor, questions in factor_questions.items():
|
||||
total_score = 0
|
||||
for q in questions:
|
||||
score = answers[q['id']]
|
||||
score = answers[q["id"]]
|
||||
# 处理反向计分题目
|
||||
if q['reverse_scoring']:
|
||||
if q["reverse_scoring"]:
|
||||
score = 7 - score # 6分量表反向计分为7减原始分
|
||||
total_score += score
|
||||
|
||||
# 计算平均分
|
||||
avg_score = round(total_score / len(questions), 2)
|
||||
results[factor] = {
|
||||
"得分": avg_score,
|
||||
"题目数": len(questions),
|
||||
"总分": total_score
|
||||
}
|
||||
results[factor] = {"得分": avg_score, "题目数": len(questions), "总分": total_score}
|
||||
|
||||
return results
|
||||
|
||||
@@ -103,6 +90,7 @@ class BigFiveTest:
|
||||
"""获取因子的详细描述"""
|
||||
return self.factors[factor]
|
||||
|
||||
|
||||
def main():
|
||||
test = BigFiveTest()
|
||||
results = test.run_test()
|
||||
@@ -114,9 +102,10 @@ def main():
|
||||
print(f"平均分: {data['得分']} (总分: {data['总分']}, 题目数: {data['题目数']})")
|
||||
print("-" * 30)
|
||||
description = test.get_factor_description(factor)
|
||||
print("维度说明:", description['description'][:100] + "...")
|
||||
print("\n特征词:", ", ".join(description['trait_words']))
|
||||
print("维度说明:", description["description"][:100] + "...")
|
||||
print("\n特征词:", ", ".join(description["trait_words"]))
|
||||
print("=" * 50)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
|
||||
@@ -1,4 +1,4 @@
|
||||
from typing import Dict, List
|
||||
from typing import Dict
|
||||
import json
|
||||
import os
|
||||
from pathlib import Path
|
||||
@@ -14,9 +14,10 @@ env_path = project_root / ".env.prod"
|
||||
root_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
|
||||
sys.path.append(root_path)
|
||||
|
||||
from src.plugins.personality.big5_test import BigFiveTest
|
||||
from src.plugins.personality.renqingziji import PersonalityEvaluator_direct
|
||||
from src.plugins.personality.questionnaire import FACTOR_DESCRIPTIONS, PERSONALITY_QUESTIONS
|
||||
from src.plugins.personality.big5_test import BigFiveTest # noqa: E402
|
||||
from src.plugins.personality.renqingziji import PersonalityEvaluator_direct # noqa: E402
|
||||
from src.plugins.personality.questionnaire import FACTOR_DESCRIPTIONS, PERSONALITY_QUESTIONS # noqa: E402
|
||||
|
||||
|
||||
class CombinedPersonalityTest:
|
||||
def __init__(self):
|
||||
@@ -51,10 +52,7 @@ class CombinedPersonalityTest:
|
||||
questionnaire_results = self.run_questionnaire()
|
||||
|
||||
# 转换问卷结果格式以便比较
|
||||
questionnaire_scores = {
|
||||
factor: data["得分"]
|
||||
for factor, data in questionnaire_results.items()
|
||||
}
|
||||
questionnaire_scores = {factor: data["得分"] for factor, data in questionnaire_results.items()}
|
||||
|
||||
# 运行情景测试
|
||||
print("\n=== 第二部分：情景反应测评 ===")
|
||||
@@ -74,7 +72,7 @@ class CombinedPersonalityTest:
|
||||
def run_questionnaire(self):
|
||||
"""运行问卷测试部分"""
|
||||
# 创建题目序号到题目的映射
|
||||
questions_map = {q['id']: q for q in PERSONALITY_QUESTIONS}
|
||||
questions_map = {q["id"]: q for q in PERSONALITY_QUESTIONS}
|
||||
|
||||
# 获取所有题目ID并随机打乱顺序
|
||||
question_ids = list(questions_map.keys())
|
||||
@@ -107,35 +105,25 @@ class CombinedPersonalityTest:
|
||||
def calculate_questionnaire_scores(self, answers):
|
||||
"""计算问卷测试的维度得分"""
|
||||
results = {}
|
||||
factor_questions = {
|
||||
"外向性": [],
|
||||
"神经质": [],
|
||||
"严谨性": [],
|
||||
"开放性": [],
|
||||
"宜人性": []
|
||||
}
|
||||
factor_questions = {"外向性": [], "神经质": [], "严谨性": [], "开放性": [], "宜人性": []}
|
||||
|
||||
# 将题目按因子分类
|
||||
for q in PERSONALITY_QUESTIONS:
|
||||
factor_questions[q['factor']].append(q)
|
||||
factor_questions[q["factor"]].append(q)
|
||||
|
||||
# 计算每个维度的得分
|
||||
for factor, questions in factor_questions.items():
|
||||
total_score = 0
|
||||
for q in questions:
|
||||
score = answers[q['id']]
|
||||
score = answers[q["id"]]
|
||||
# 处理反向计分题目
|
||||
if q['reverse_scoring']:
|
||||
if q["reverse_scoring"]:
|
||||
score = 7 - score # 6分量表反向计分为7减原始分
|
||||
total_score += score
|
||||
|
||||
# 计算平均分
|
||||
avg_score = round(total_score / len(questions), 2)
|
||||
results[factor] = {
|
||||
"得分": avg_score,
|
||||
"题目数": len(questions),
|
||||
"总分": total_score
|
||||
}
|
||||
results[factor] = {"得分": avg_score, "题目数": len(questions), "总分": total_score}
|
||||
|
||||
return results
|
||||
|
||||
@@ -160,11 +148,7 @@ class CombinedPersonalityTest:
|
||||
continue
|
||||
|
||||
print("\n正在评估您的描述...")
|
||||
scores = self.scenario_test.evaluate_response(
|
||||
scenario_data["场景"],
|
||||
response,
|
||||
scenario_data["评估维度"]
|
||||
)
|
||||
scores = self.scenario_test.evaluate_response(scenario_data["场景"], response, scenario_data["评估维度"])
|
||||
|
||||
# 更新分数
|
||||
for dimension, score in scores.items():
|
||||
@@ -186,10 +170,7 @@ class CombinedPersonalityTest:
|
||||
# 计算平均分
|
||||
for dimension in final_scores:
|
||||
if dimension_counts[dimension] > 0:
|
||||
final_scores[dimension] = round(
|
||||
final_scores[dimension] / dimension_counts[dimension],
|
||||
2
|
||||
)
|
||||
final_scores[dimension] = round(final_scores[dimension] / dimension_counts[dimension], 2)
|
||||
|
||||
return final_scores
|
||||
|
||||
@@ -225,9 +206,13 @@ class CombinedPersonalityTest:
|
||||
std_diff = (sum((x - mean_diff) ** 2 for x in diffs) / (len(diffs) - 1)) ** 0.5
|
||||
|
||||
# 计算效应量 (Cohen's d)
|
||||
pooled_std = ((sum((x - sum(questionnaire_values)/len(questionnaire_values))**2 for x in questionnaire_values) +
|
||||
sum((x - sum(scenario_values)/len(scenario_values))**2 for x in scenario_values)) /
|
||||
(2 * len(self.dimensions) - 2)) ** 0.5
|
||||
pooled_std = (
|
||||
(
|
||||
sum((x - sum(questionnaire_values) / len(questionnaire_values)) ** 2 for x in questionnaire_values)
|
||||
+ sum((x - sum(scenario_values) / len(scenario_values)) ** 2 for x in scenario_values)
|
||||
)
|
||||
/ (2 * len(self.dimensions) - 2)
|
||||
) ** 0.5
|
||||
|
||||
if pooled_std != 0:
|
||||
cohens_d = abs(mean_diff / pooled_std)
|
||||
@@ -244,7 +229,7 @@ class CombinedPersonalityTest:
|
||||
|
||||
# 对所有维度进行整体t检验
|
||||
t_stat, p_value = stats.ttest_rel(questionnaire_values, scenario_values)
|
||||
print(f"\n整体统计分析:")
|
||||
print("\n整体统计分析:")
|
||||
print(f"平均差异: {mean_diff:.3f}")
|
||||
print(f"差异标准差: {std_diff:.3f}")
|
||||
print(f"效应量(Cohen's d): {cohens_d:.3f}")
|
||||
@@ -269,12 +254,14 @@ class CombinedPersonalityTest:
|
||||
for dimension in self.dimensions:
|
||||
diff = abs(questionnaire_scores[dimension] - scenario_scores[dimension])
|
||||
if diff >= 1.0: # 差异大于等于1分视为显著
|
||||
significant_diffs.append({
|
||||
significant_diffs.append(
|
||||
{
|
||||
"dimension": dimension,
|
||||
"diff": diff,
|
||||
"questionnaire": questionnaire_scores[dimension],
|
||||
"scenario": scenario_scores[dimension]
|
||||
})
|
||||
"scenario": scenario_scores[dimension],
|
||||
}
|
||||
)
|
||||
|
||||
if significant_diffs:
|
||||
print("\n\n显著差异分析：")
|
||||
@@ -286,7 +273,7 @@ class CombinedPersonalityTest:
|
||||
print(f"差异值：{diff['diff']:.2f}")
|
||||
|
||||
# 分析可能的原因
|
||||
if diff['questionnaire'] > diff['scenario']:
|
||||
if diff["questionnaire"] > diff["scenario"]:
|
||||
print("可能原因：在问卷中的自我评价较高，但在具体情景中的表现较为保守。")
|
||||
else:
|
||||
print("可能原因：在具体情景中表现出更多该维度特征，而在问卷自评时较为保守。")
|
||||
@@ -297,7 +284,7 @@ class CombinedPersonalityTest:
|
||||
"测试时间": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
|
||||
"问卷测评结果": questionnaire_scores,
|
||||
"情景测评结果": scenario_scores,
|
||||
"维度说明": FACTOR_DESCRIPTIONS
|
||||
"维度说明": FACTOR_DESCRIPTIONS,
|
||||
}
|
||||
|
||||
# 确保目录存在
|
||||
@@ -312,6 +299,7 @@ class CombinedPersonalityTest:
|
||||
|
||||
print(f"\n完整的测评结果已保存到：{filename}")
|
||||
|
||||
|
||||
def load_existing_results():
|
||||
"""检查并加载已有的测试结果"""
|
||||
results_dir = "results"
|
||||
@@ -319,15 +307,13 @@ def load_existing_results():
|
||||
return None
|
||||
|
||||
# 获取所有personality_combined开头的文件
|
||||
result_files = [f for f in os.listdir(results_dir)
|
||||
if f.startswith("personality_combined_") and f.endswith(".json")]
|
||||
result_files = [f for f in os.listdir(results_dir) if f.startswith("personality_combined_") and f.endswith(".json")]
|
||||
|
||||
if not result_files:
|
||||
return None
|
||||
|
||||
# 按文件修改时间排序，获取最新的结果文件
|
||||
latest_file = max(result_files,
|
||||
key=lambda f: os.path.getmtime(os.path.join(results_dir, f)))
|
||||
latest_file = max(result_files, key=lambda f: os.path.getmtime(os.path.join(results_dir, f)))
|
||||
|
||||
print(f"\n发现已有的测试结果：{latest_file}")
|
||||
try:
|
||||
@@ -338,6 +324,7 @@ def load_existing_results():
|
||||
print(f"读取结果文件时出错：{str(e)}")
|
||||
return None
|
||||
|
||||
|
||||
def main():
|
||||
test = CombinedPersonalityTest()
|
||||
|
||||
@@ -357,5 +344,6 @@ def main():
|
||||
print("\n未找到已有的测试结果，开始新的测试...")
|
||||
test.run_combined_test()
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
@@ -1,5 +1,9 @@
|
||||
# 人格测试问卷题目 王孟成, 戴晓阳, & 姚树桥. (2011). 中国大五人格问卷的初步编制Ⅲ:简式版的制定及信效度检验. 中国临床心理学杂志, 19(04), Article 04.
|
||||
# 王孟成, 戴晓阳, & 姚树桥. (2010). 中国大五人格问卷的初步编制Ⅰ:理论框架与信度分析. 中国临床心理学杂志, 18(05), Article 05.
|
||||
# 人格测试问卷题目
|
||||
# 王孟成, 戴晓阳, & 姚树桥. (2011).
|
||||
# 中国大五人格问卷的初步编制Ⅲ:简式版的制定及信效度检验. 中国临床心理学杂志, 19(04), Article 04.
|
||||
|
||||
# 王孟成, 戴晓阳, & 姚树桥. (2010).
|
||||
# 中国大五人格问卷的初步编制Ⅰ:理论框架与信度分析. 中国临床心理学杂志, 18(05), Article 05.
|
||||
|
||||
PERSONALITY_QUESTIONS = [
|
||||
# 神经质维度 (F1)
|
||||
@@ -11,7 +15,6 @@ PERSONALITY_QUESTIONS = [
|
||||
{"id": 6, "content": "在面对压力时，我有种快要崩溃的感觉", "factor": "神经质", "reverse_scoring": False},
|
||||
{"id": 7, "content": "我常担忧一些无关紧要的事情", "factor": "神经质", "reverse_scoring": False},
|
||||
{"id": 8, "content": "我常常感到内心不踏实", "factor": "神经质", "reverse_scoring": False},
|
||||
|
||||
# 严谨性维度 (F2)
|
||||
{"id": 9, "content": "在工作上，我常只求能应付过去便可", "factor": "严谨性", "reverse_scoring": True},
|
||||
{"id": 10, "content": "一旦确定了目标，我会坚持努力地实现它", "factor": "严谨性", "reverse_scoring": False},
|
||||
@@ -21,9 +24,13 @@ PERSONALITY_QUESTIONS = [
|
||||
{"id": 14, "content": "我喜欢一开头就把事情计划好", "factor": "严谨性", "reverse_scoring": False},
|
||||
{"id": 15, "content": "我工作或学习很勤奋", "factor": "严谨性", "reverse_scoring": False},
|
||||
{"id": 16, "content": "我是个倾尽全力做事的人", "factor": "严谨性", "reverse_scoring": False},
|
||||
|
||||
# 宜人性维度 (F3)
|
||||
{"id": 17, "content": "尽管人类社会存在着一些阴暗的东西（如战争、罪恶、欺诈），我仍然相信人性总的来说是善良的", "factor": "宜人性", "reverse_scoring": False},
|
||||
{
|
||||
"id": 17,
|
||||
"content": "尽管人类社会存在着一些阴暗的东西（如战争、罪恶、欺诈），我仍然相信人性总的来说是善良的",
|
||||
"factor": "宜人性",
|
||||
"reverse_scoring": False,
|
||||
},
|
||||
{"id": 18, "content": "我觉得大部分人基本上是心怀善意的", "factor": "宜人性", "reverse_scoring": False},
|
||||
{"id": 19, "content": "虽然社会上有骗子，但我觉得大部分人还是可信的", "factor": "宜人性", "reverse_scoring": False},
|
||||
{"id": 20, "content": "我不太关心别人是否受到不公正的待遇", "factor": "宜人性", "reverse_scoring": True},
|
||||
@@ -31,7 +38,6 @@ PERSONALITY_QUESTIONS = [
|
||||
{"id": 22, "content": "我常为那些遭遇不幸的人感到难过", "factor": "宜人性", "reverse_scoring": False},
|
||||
{"id": 23, "content": "我是那种只照顾好自己，不替别人担忧的人", "factor": "宜人性", "reverse_scoring": True},
|
||||
{"id": 24, "content": "当别人向我诉说不幸时，我常感到难过", "factor": "宜人性", "reverse_scoring": False},
|
||||
|
||||
# 开放性维度 (F4)
|
||||
{"id": 25, "content": "我的想象力相当丰富", "factor": "开放性", "reverse_scoring": False},
|
||||
{"id": 26, "content": "我头脑中经常充满生动的画面", "factor": "开放性", "reverse_scoring": False},
|
||||
@@ -39,9 +45,18 @@ PERSONALITY_QUESTIONS = [
|
||||
{"id": 28, "content": "我喜欢冒险", "factor": "开放性", "reverse_scoring": False},
|
||||
{"id": 29, "content": "我是个勇于冒险，突破常规的人", "factor": "开放性", "reverse_scoring": False},
|
||||
{"id": 30, "content": "我身上具有别人没有的冒险精神", "factor": "开放性", "reverse_scoring": False},
|
||||
{"id": 31, "content": "我渴望学习一些新东西，即使它们与我的日常生活无关", "factor": "开放性", "reverse_scoring": False},
|
||||
{"id": 32, "content": "我很愿意也很容易接受那些新事物、新观点、新想法", "factor": "开放性", "reverse_scoring": False},
|
||||
|
||||
{
|
||||
"id": 31,
|
||||
"content": "我渴望学习一些新东西，即使它们与我的日常生活无关",
|
||||
"factor": "开放性",
|
||||
"reverse_scoring": False,
|
||||
},
|
||||
{
|
||||
"id": 32,
|
||||
"content": "我很愿意也很容易接受那些新事物、新观点、新想法",
|
||||
"factor": "开放性",
|
||||
"reverse_scoring": False,
|
||||
},
|
||||
# 外向性维度 (F5)
|
||||
{"id": 33, "content": "我喜欢参加社交与娱乐聚会", "factor": "外向性", "reverse_scoring": False},
|
||||
{"id": 34, "content": "我对人多的聚会感到乏味", "factor": "外向性", "reverse_scoring": True},
|
||||
@@ -50,61 +65,78 @@ PERSONALITY_QUESTIONS = [
|
||||
{"id": 37, "content": "有我在的场合一般不会冷场", "factor": "外向性", "reverse_scoring": False},
|
||||
{"id": 38, "content": "我希望成为领导者而不是被领导者", "factor": "外向性", "reverse_scoring": False},
|
||||
{"id": 39, "content": "在一个团体中，我希望处于领导地位", "factor": "外向性", "reverse_scoring": False},
|
||||
{"id": 40, "content": "别人多认为我是一个热情和友好的人", "factor": "外向性", "reverse_scoring": False}
|
||||
{"id": 40, "content": "别人多认为我是一个热情和友好的人", "factor": "外向性", "reverse_scoring": False},
|
||||
]
|
||||
|
||||
# 因子维度说明
|
||||
FACTOR_DESCRIPTIONS = {
|
||||
"外向性": {
|
||||
"description": "反映个体神经系统的强弱和动力特征。外向性主要表现为个体在人际交往和社交活动中的倾向性，包括对社交活动的兴趣、对人群的态度、社交互动中的主动程度以及在群体中的影响力。高分者倾向于积极参与社交活动，乐于与人交往，善于表达自我，并往往在群体中发挥领导作用；低分者则倾向于独处，不喜欢热闹的社交场合，表现出内向、安静的特征。",
|
||||
"description": "反映个体神经系统的强弱和动力特征。外向性主要表现为个体在人际交往和社交活动中的倾向性，"
|
||||
"包括对社交活动的兴趣、"
|
||||
"对人群的态度、社交互动中的主动程度以及在群体中的影响力。高分者倾向于积极参与社交活动，乐于与人交往，善于表达自我，"
|
||||
"并往往在群体中发挥领导作用；低分者则倾向于独处，不喜欢热闹的社交场合，表现出内向、安静的特征。",
|
||||
"trait_words": ["热情", "活力", "社交", "主动"],
|
||||
"subfactors": {
|
||||
"合群性": "个体愿意与他人聚在一起，即接近人群的倾向；高分表现乐群、好交际，低分表现封闭、独处",
|
||||
"热情": "个体对待别人时所表现出的态度；高分表现热情好客，低分表现冷淡",
|
||||
"支配性": "个体喜欢指使、操纵他人，倾向于领导别人的特点；高分表现好强、发号施令，低分表现顺从、低调",
|
||||
"活跃": "个体精力充沛，活跃、主动性等特点；高分表现活跃，低分表现安静"
|
||||
}
|
||||
"活跃": "个体精力充沛，活跃、主动性等特点；高分表现活跃，低分表现安静",
|
||||
},
|
||||
},
|
||||
"神经质": {
|
||||
"description": "反映个体情绪的状态和体验内心苦恼的倾向性。这个维度主要关注个体在面对压力、挫折和日常生活挑战时的情绪稳定性和适应能力。它包含了对焦虑、抑郁、愤怒等负面情绪的敏感程度，以及个体对这些情绪的调节和控制能力。高分者容易体验负面情绪，对压力较为敏感，情绪波动较大；低分者则表现出较强的情绪稳定性，能够较好地应对压力和挫折。",
|
||||
"description": "反映个体情绪的状态和体验内心苦恼的倾向性。这个维度主要关注个体在面对压力、"
|
||||
"挫折和日常生活挑战时的情绪稳定性和适应能力。它包含了对焦虑、抑郁、愤怒等负面情绪的敏感程度，"
|
||||
"以及个体对这些情绪的调节和控制能力。高分者容易体验负面情绪，对压力较为敏感，情绪波动较大；"
|
||||
"低分者则表现出较强的情绪稳定性，能够较好地应对压力和挫折。",
|
||||
"trait_words": ["稳定", "沉着", "从容", "坚韧"],
|
||||
"subfactors": {
|
||||
"焦虑": "个体体验焦虑感的个体差异；高分表现坐立不安，低分表现平静",
|
||||
"抑郁": "个体体验抑郁情感的个体差异；高分表现郁郁寡欢，低分表现平静",
|
||||
"敏感多疑": "个体常常关注自己的内心活动，行为和过于意识人对自己的看法、评价；高分表现敏感多疑，低分表现淡定、自信",
|
||||
"敏感多疑": "个体常常关注自己的内心活动，行为和过于意识人对自己的看法、评价；高分表现敏感多疑，"
|
||||
"低分表现淡定、自信",
|
||||
"脆弱性": "个体在危机或困难面前无力、脆弱的特点；高分表现无能、易受伤、逃避，低分表现坚强",
|
||||
"愤怒-敌意": "个体准备体验愤怒，及相关情绪的状态；高分表现暴躁易怒，低分表现平静"
|
||||
}
|
||||
"愤怒-敌意": "个体准备体验愤怒，及相关情绪的状态；高分表现暴躁易怒，低分表现平静",
|
||||
},
|
||||
},
|
||||
"严谨性": {
|
||||
"description": "反映个体在目标导向行为上的组织、坚持和动机特征。这个维度体现了个体在工作、学习等目标性活动中的自我约束和行为管理能力。它涉及到个体的责任感、自律性、计划性、条理性以及完成任务的态度。高分者往往表现出强烈的责任心、良好的组织能力、谨慎的决策风格和持续的努力精神；低分者则可能表现出随意性强、缺乏规划、做事马虎或易放弃的特点。",
|
||||
"description": "反映个体在目标导向行为上的组织、坚持和动机特征。这个维度体现了个体在工作、"
|
||||
"学习等目标性活动中的自我约束和行为管理能力。它涉及到个体的责任感、自律性、计划性、条理性以及完成任务的态度。"
|
||||
"高分者往往表现出强烈的责任心、良好的组织能力、谨慎的决策风格和持续的努力精神；低分者则可能表现出随意性强、"
|
||||
"缺乏规划、做事马虎或易放弃的特点。",
|
||||
"trait_words": ["负责", "自律", "条理", "勤奋"],
|
||||
"subfactors": {
|
||||
"责任心": "个体对待任务和他人认真负责，以及对自己承诺的信守；高分表现有责任心、负责任，低分表现推卸责任、逃避处罚",
|
||||
"责任心": "个体对待任务和他人认真负责，以及对自己承诺的信守；高分表现有责任心、负责任，"
|
||||
"低分表现推卸责任、逃避处罚",
|
||||
"自我控制": "个体约束自己的能力，及自始至终的坚持性；高分表现自制、有毅力，低分表现冲动、无毅力",
|
||||
"审慎性": "个体在采取具体行动前的心理状态；高分表现谨慎、小心，低分表现鲁莽、草率",
|
||||
"条理性": "个体处理事务和工作的秩序，条理和逻辑性；高分表现整洁、有秩序，低分表现混乱、遗漏",
|
||||
"勤奋": "个体工作和学习的努力程度及为达到目标而表现出的进取精神；高分表现勤奋、刻苦，低分表现懒散"
|
||||
}
|
||||
"勤奋": "个体工作和学习的努力程度及为达到目标而表现出的进取精神；高分表现勤奋、刻苦，低分表现懒散",
|
||||
},
|
||||
},
|
||||
"开放性": {
|
||||
"description": "反映个体对新异事物、新观念和新经验的接受程度，以及在思维和行为方面的创新倾向。这个维度体现了个体在认知和体验方面的广度、深度和灵活性。它包括对艺术的欣赏能力、对知识的求知欲、想象力的丰富程度，以及对冒险和创新的态度。高分者往往具有丰富的想象力、广泛的兴趣、开放的思维方式和创新的倾向；低分者则倾向于保守、传统，喜欢熟悉和常规的事物。",
|
||||
"description": "反映个体对新异事物、新观念和新经验的接受程度，以及在思维和行为方面的创新倾向。"
|
||||
"这个维度体现了个体在认知和体验方面的广度、深度和灵活性。它包括对艺术的欣赏能力、对知识的求知欲、想象力的丰富程度，"
|
||||
"以及对冒险和创新的态度。高分者往往具有丰富的想象力、广泛的兴趣、开放的思维方式和创新的倾向；低分者则倾向于保守、"
|
||||
"传统，喜欢熟悉和常规的事物。",
|
||||
"trait_words": ["创新", "好奇", "艺术", "冒险"],
|
||||
"subfactors": {
|
||||
"幻想": "个体富于幻想和想象的水平；高分表现想象力丰富，低分表现想象力匮乏",
|
||||
"审美": "个体对于艺术和美的敏感与热爱程度；高分表现富有艺术气息，低分表现一般对艺术不敏感",
|
||||
"好奇心": "个体对未知事物的态度；高分表现兴趣广泛、好奇心浓，低分表现兴趣少、无好奇心",
|
||||
"冒险精神": "个体愿意尝试有风险活动的个体差异；高分表现好冒险，低分表现保守",
|
||||
"价值观念": "个体对新事物、新观念、怪异想法的态度；高分表现开放、坦然接受新事物，低分则相反"
|
||||
}
|
||||
"价值观念": "个体对新事物、新观念、怪异想法的态度；高分表现开放、坦然接受新事物，低分则相反",
|
||||
},
|
||||
},
|
||||
"宜人性": {
|
||||
"description": "反映个体在人际关系中的亲和倾向，体现了对他人的关心、同情和合作意愿。这个维度主要关注个体与他人互动时的态度和行为特征，包括对他人的信任程度、同理心水平、助人意愿以及在人际冲突中的处理方式。高分者通常表现出友善、富有同情心、乐于助人的特质，善于与他人建立和谐关系；低分者则可能表现出较少的人际关注，在社交互动中更注重自身利益，较少考虑他人感受。",
|
||||
"description": "反映个体在人际关系中的亲和倾向，体现了对他人的关心、同情和合作意愿。"
|
||||
"这个维度主要关注个体与他人互动时的态度和行为特征，包括对他人的信任程度、同理心水平、"
|
||||
"助人意愿以及在人际冲突中的处理方式。高分者通常表现出友善、富有同情心、乐于助人的特质，善于与他人建立和谐关系；"
|
||||
"低分者则可能表现出较少的人际关注，在社交互动中更注重自身利益，较少考虑他人感受。",
|
||||
"trait_words": ["友善", "同理", "信任", "合作"],
|
||||
"subfactors": {
|
||||
"信任": "个体对他人和/或他人言论的相信程度；高分表现信任他人，低分表现怀疑",
|
||||
"体贴": "个体对别人的兴趣和需要的关注程度；高分表现体贴、温存，低分表现冷漠、不在乎",
|
||||
"同情": "个体对处于不利地位的人或物的态度；高分表现富有同情心，低分表现冷漠"
|
||||
}
|
||||
}
|
||||
"同情": "个体对处于不利地位的人或物的态度；高分表现富有同情心，低分表现冷漠",
|
||||
},
|
||||
},
|
||||
}
|
||||
@@ -1,10 +1,12 @@
|
||||
'''
|
||||
The definition of artificial personality in this paper follows the dispositional para-digm and adapts a definition of personality developed for humans [17]:
|
||||
"""
|
||||
The definition of artificial personality in this paper follows the dispositional para-digm and adapts a definition of
|
||||
personality developed for humans [17]:
|
||||
Personality for a human is the "whole and organisation of relatively stable tendencies and patterns of experience and
|
||||
behaviour within one person (distinguishing it from other persons)". This definition is modified for artificial personality:
|
||||
behaviour within one person (distinguishing it from other persons)". This definition is modified for artificial
|
||||
personality:
|
||||
Artificial personality describes the relatively stable tendencies and patterns of behav-iour of an AI-based machine that
|
||||
can be designed by developers and designers via different modalities, such as language, creating the impression
|
||||
of individuality of a humanized social agent when users interact with the machine.'''
|
||||
of individuality of a humanized social agent when users interact with the machine."""
|
||||
|
||||
from typing import Dict, List
|
||||
import json
|
||||
@@ -13,9 +15,9 @@ from pathlib import Path
|
||||
from dotenv import load_dotenv
|
||||
import sys
|
||||
|
||||
'''
|
||||
"""
|
||||
第一种方案：基于情景评估的人格测定
|
||||
'''
|
||||
"""
|
||||
current_dir = Path(__file__).resolve().parent
|
||||
project_root = current_dir.parent.parent.parent
|
||||
env_path = project_root / ".env.prod"
|
||||
@@ -23,9 +25,9 @@ env_path = project_root / ".env.prod"
|
||||
root_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
|
||||
sys.path.append(root_path)
|
||||
|
||||
from src.plugins.personality.scene import get_scene_by_factor,get_all_scenes,PERSONALITY_SCENES
|
||||
from src.plugins.personality.questionnaire import PERSONALITY_QUESTIONS,FACTOR_DESCRIPTIONS
|
||||
from src.plugins.personality.offline_llm import LLMModel
|
||||
from src.plugins.personality.scene import get_scene_by_factor, PERSONALITY_SCENES # noqa: E402
|
||||
from src.plugins.personality.questionnaire import FACTOR_DESCRIPTIONS # noqa: E402
|
||||
from src.plugins.personality.offline_llm import LLMModel # noqa: E402
|
||||
|
||||
# 加载环境变量
|
||||
if env_path.exists():
|
||||
@@ -49,6 +51,7 @@ class PersonalityEvaluator_direct:
|
||||
|
||||
# 从每个维度选择3个场景
|
||||
import random
|
||||
|
||||
scene_keys = list(scenes.keys())
|
||||
selected_scenes = random.sample(scene_keys, min(3, len(scene_keys)))
|
||||
|
||||
@@ -60,11 +63,9 @@ class PersonalityEvaluator_direct:
|
||||
other_traits = [t for t in PERSONALITY_SCENES if t != trait]
|
||||
secondary_trait = random.choice(other_traits)
|
||||
|
||||
self.scenarios.append({
|
||||
"场景": scene["scenario"],
|
||||
"评估维度": [trait, secondary_trait],
|
||||
"场景编号": scene_key
|
||||
})
|
||||
self.scenarios.append(
|
||||
{"场景": scene["scenario"], "评估维度": [trait, secondary_trait], "场景编号": scene_key}
|
||||
)
|
||||
|
||||
self.llm = LLMModel()
|
||||
|
||||
@@ -178,11 +179,7 @@ def main():
|
||||
print(f"测试场景数：{dimension_counts[trait]}")
|
||||
|
||||
# 保存结果
|
||||
result = {
|
||||
"final_scores": final_scores,
|
||||
"dimension_counts": dimension_counts,
|
||||
"scenarios": evaluator.scenarios
|
||||
}
|
||||
result = {"final_scores": final_scores, "dimension_counts": dimension_counts, "scenarios": evaluator.scenarios}
|
||||
|
||||
# 确保目录存在
|
||||
os.makedirs("results", exist_ok=True)
|
||||
|
||||
@@ -1,4 +1,4 @@
|
||||
from typing import Dict, List
|
||||
from typing import Dict
|
||||
|
||||
PERSONALITY_SCENES = {
|
||||
"外向性": {
|
||||
@@ -8,7 +8,7 @@ PERSONALITY_SCENES = {
|
||||
同事：「嗨！你是新来的同事吧？我是市场部的小林。」
|
||||
|
||||
同事看起来很友善，还主动介绍说：「待会午饭时间，我们部门有几个人准备一起去楼下新开的餐厅，你要一起来吗？可以认识一下其他同事。」""",
|
||||
"explanation": "这个场景通过职场社交情境，观察个体对于新环境、新社交圈的态度和反应倾向。"
|
||||
"explanation": "这个场景通过职场社交情境，观察个体对于新环境、新社交圈的态度和反应倾向。",
|
||||
},
|
||||
"场景2": {
|
||||
"scenario": """在大学班级群里，班长发起了一个组织班级联谊活动的投票：
|
||||
@@ -16,7 +16,7 @@ PERSONALITY_SCENES = {
|
||||
班长：「大家好！下周末我们准备举办一次班级联谊活动，地点在学校附近的KTV。想请大家报名参加，也欢迎大家邀请其他班级的同学！」
|
||||
|
||||
已经有几个同学在群里积极响应，有人@你问你要不要一起参加。""",
|
||||
"explanation": "通过班级活动场景，观察个体对群体社交活动的参与意愿。"
|
||||
"explanation": "通过班级活动场景，观察个体对群体社交活动的参与意愿。",
|
||||
},
|
||||
"场景3": {
|
||||
"scenario": """你在社交平台上发布了一条动态，收到了很多陌生网友的评论和私信：
|
||||
@@ -24,13 +24,14 @@ PERSONALITY_SCENES = {
|
||||
网友A：「你说的这个观点很有意思！想和你多交流一下。」
|
||||
|
||||
网友B：「我也对这个话题很感兴趣，要不要建个群一起讨论？」""",
|
||||
"explanation": "通过网络社交场景，观察个体对线上社交的态度。"
|
||||
"explanation": "通过网络社交场景，观察个体对线上社交的态度。",
|
||||
},
|
||||
"场景4": {
|
||||
"scenario": """你暗恋的对象今天主动来找你：
|
||||
|
||||
对方：「那个...我最近在准备一个演讲比赛，听说你口才很好。能不能请你帮我看看演讲稿，顺便给我一些建议？如果你有时间的话，可以一起吃个饭聊聊。」""",
|
||||
"explanation": "通过恋爱情境，观察个体在面对心仪对象时的社交表现。"
|
||||
对方：「那个...我最近在准备一个演讲比赛，听说你口才很好。能不能请你帮我看看演讲稿，顺便给我一些建议？"""
|
||||
"""如果你有时间的话，可以一起吃个饭聊聊。」""",
|
||||
"explanation": "通过恋爱情境，观察个体在面对心仪对象时的社交表现。",
|
||||
},
|
||||
"场景5": {
|
||||
"scenario": """在一次线下读书会上，主持人突然点名让你分享读后感：
|
||||
@@ -38,18 +39,18 @@ PERSONALITY_SCENES = {
|
||||
主持人：「听说你对这本书很有见解，能不能和大家分享一下你的想法？」
|
||||
|
||||
现场有二十多个陌生的读书爱好者，都期待地看着你。""",
|
||||
"explanation": "通过即兴发言场景，观察个体的社交表现欲和公众表达能力。"
|
||||
}
|
||||
"explanation": "通过即兴发言场景，观察个体的社交表现欲和公众表达能力。",
|
||||
},
|
||||
},
|
||||
|
||||
"神经质": {
|
||||
"场景1": {
|
||||
"scenario": """你正在准备一个重要的项目演示，这关系到你的晋升机会。就在演示前30分钟，你收到了主管发来的消息：
|
||||
"scenario": """你正在准备一个重要的项目演示，这关系到你的晋升机会。"""
|
||||
"""就在演示前30分钟，你收到了主管发来的消息：
|
||||
|
||||
主管：「临时有个变动，CEO也会来听你的演示。他对这个项目特别感兴趣。」
|
||||
|
||||
正当你准备回复时，主管又发来一条：「对了，能不能把演示时间压缩到15分钟？CEO下午还有其他安排。你之前准备的是30分钟的版本对吧？」""",
|
||||
"explanation": "这个场景通过突发的压力情境，观察个体在面对计划外变化时的情绪反应和调节能力。"
|
||||
"explanation": "这个场景通过突发的压力情境，观察个体在面对计划外变化时的情绪反应和调节能力。",
|
||||
},
|
||||
"场景2": {
|
||||
"scenario": """期末考试前一天晚上，你收到了好朋友发来的消息：
|
||||
@@ -57,7 +58,7 @@ PERSONALITY_SCENES = {
|
||||
好朋友：「不好意思这么晚打扰你...我看你平时成绩很好，能不能帮我解答几个问题？我真的很担心明天的考试。」
|
||||
|
||||
你看了看时间，已经是晚上11点，而你原本计划的复习还没完成。""",
|
||||
"explanation": "通过考试压力场景，观察个体在时间紧张时的情绪管理。"
|
||||
"explanation": "通过考试压力场景，观察个体在时间紧张时的情绪管理。",
|
||||
},
|
||||
"场景3": {
|
||||
"scenario": """你在社交媒体上发表的一个观点引发了争议，有不少人开始批评你：
|
||||
@@ -67,7 +68,7 @@ PERSONALITY_SCENES = {
|
||||
网友B：「建议楼主先去补补课再来发言。」
|
||||
|
||||
评论区里的负面评论越来越多，还有人开始人身攻击。""",
|
||||
"explanation": "通过网络争议场景，观察个体面对批评时的心理承受能力。"
|
||||
"explanation": "通过网络争议场景，观察个体面对批评时的心理承受能力。",
|
||||
},
|
||||
"场景4": {
|
||||
"scenario": """你和恋人约好今天一起看电影，但在约定时间前半小时，对方发来消息：
|
||||
@@ -77,7 +78,7 @@ PERSONALITY_SCENES = {
|
||||
二十分钟后，对方又发来消息：「可能要再等等，抱歉！」
|
||||
|
||||
电影快要开始了，但对方还是没有出现。""",
|
||||
"explanation": "通过恋爱情境，观察个体对不确定性的忍耐程度。"
|
||||
"explanation": "通过恋爱情境，观察个体对不确定性的忍耐程度。",
|
||||
},
|
||||
"场景5": {
|
||||
"scenario": """在一次重要的小组展示中，你的组员在演示途中突然卡壳了：
|
||||
@@ -85,10 +86,9 @@ PERSONALITY_SCENES = {
|
||||
组员小声对你说：「我忘词了，接下来的部分是什么来着...」
|
||||
|
||||
台下的老师和同学都在等待，气氛有些尴尬。""",
|
||||
"explanation": "通过公开场合的突发状况，观察个体的应急反应和压力处理能力。"
|
||||
}
|
||||
"explanation": "通过公开场合的突发状况，观察个体的应急反应和压力处理能力。",
|
||||
},
|
||||
},
|
||||
|
||||
"严谨性": {
|
||||
"场景1": {
|
||||
"scenario": """你是团队的项目负责人，刚刚接手了一个为期两个月的重要项目。在第一次团队会议上：
|
||||
@@ -98,7 +98,7 @@ PERSONALITY_SCENES = {
|
||||
小张：「要不要先列个时间表？不过感觉太详细的计划也没必要，点到为止就行。」
|
||||
|
||||
小李：「客户那边说如果能提前完成有奖励，我觉得我们可以先做快一点的部分。」""",
|
||||
"explanation": "这个场景通过项目管理情境，体现个体在工作方法、计划性和责任心方面的特征。"
|
||||
"explanation": "这个场景通过项目管理情境，体现个体在工作方法、计划性和责任心方面的特征。",
|
||||
},
|
||||
"场景2": {
|
||||
"scenario": """期末小组作业，组长让大家分工完成一份研究报告。在截止日期前三天：
|
||||
@@ -108,7 +108,7 @@ PERSONALITY_SCENES = {
|
||||
组员B：「我这边可能还要一天才能完成，最近太忙了。」
|
||||
|
||||
组员C发来一份没有任何引用出处、可能存在抄袭的内容：「我写完了，你们看看怎么样？」""",
|
||||
"explanation": "通过学习场景，观察个体对学术规范和质量要求的重视程度。"
|
||||
"explanation": "通过学习场景，观察个体对学术规范和质量要求的重视程度。",
|
||||
},
|
||||
"场景3": {
|
||||
"scenario": """你在一个兴趣小组的群聊中，大家正在讨论举办一次线下活动：
|
||||
@@ -118,7 +118,7 @@ PERSONALITY_SCENES = {
|
||||
成员B：「对啊，随意一点挺好的。」
|
||||
|
||||
成员C：「人来了自然就热闹了。」""",
|
||||
"explanation": "通过活动组织场景，观察个体对活动计划的态度。"
|
||||
"explanation": "通过活动组织场景，观察个体对活动计划的态度。",
|
||||
},
|
||||
"场景4": {
|
||||
"scenario": """你和恋人计划一起去旅游，对方说：
|
||||
@@ -126,7 +126,7 @@ PERSONALITY_SCENES = {
|
||||
恋人：「我们就随心而行吧！订个目的地，其他的到了再说，这样更有意思。」
|
||||
|
||||
距离出发还有一周时间，但机票、住宿和具体行程都还没有确定。""",
|
||||
"explanation": "通过旅行规划场景，观察个体的计划性和对不确定性的接受程度。"
|
||||
"explanation": "通过旅行规划场景，观察个体的计划性和对不确定性的接受程度。",
|
||||
},
|
||||
"场景5": {
|
||||
"scenario": """在一个重要的团队项目中，你发现一个同事的工作存在明显错误：
|
||||
@@ -134,18 +134,19 @@ PERSONALITY_SCENES = {
|
||||
同事：「差不多就行了，反正领导也看不出来。」
|
||||
|
||||
这个错误可能不会立即造成问题，但长期来看可能会影响项目质量。""",
|
||||
"explanation": "通过工作质量场景，观察个体对细节和标准的坚持程度。"
|
||||
}
|
||||
"explanation": "通过工作质量场景，观察个体对细节和标准的坚持程度。",
|
||||
},
|
||||
},
|
||||
|
||||
"开放性": {
|
||||
"场景1": {
|
||||
"scenario": """周末下午，你的好友小美兴致勃勃地给你打电话：
|
||||
|
||||
小美：「我刚发现一个特别有意思的沉浸式艺术展！不是传统那种挂画的展览，而是把整个空间都变成了艺术品。观众要穿特制的服装，还要带上VR眼镜，好像还有AI实时互动！」
|
||||
小美：「我刚发现一个特别有意思的沉浸式艺术展！不是传统那种挂画的展览，而是把整个空间都变成了艺术品。"""
|
||||
"""观众要穿特制的服装，还要带上VR眼镜，好像还有AI实时互动！」
|
||||
|
||||
小美继续说：「虽然票价不便宜，但听说体验很独特。网上评价两极分化，有人说是前所未有的艺术革新，也有人说是哗众取宠。要不要周末一起去体验一下？」""",
|
||||
"explanation": "这个场景通过新型艺术体验，反映个体对创新事物的接受程度和尝试意愿。"
|
||||
小美继续说：「虽然票价不便宜，但听说体验很独特。网上评价两极分化，有人说是前所未有的艺术革新，也有人说是哗众取宠。"""
|
||||
"""要不要周末一起去体验一下？」""",
|
||||
"explanation": "这个场景通过新型艺术体验，反映个体对创新事物的接受程度和尝试意愿。",
|
||||
},
|
||||
"场景2": {
|
||||
"scenario": """在一节创意写作课上，老师提出了一个特别的作业：
|
||||
@@ -153,15 +154,16 @@ PERSONALITY_SCENES = {
|
||||
老师：「下周的作业是用AI写作工具协助创作一篇小说。你们可以自由探索如何与AI合作，打破传统写作方式。」
|
||||
|
||||
班上随即展开了激烈讨论，有人认为这是对创作的亵渎，也有人对这种新形式感到兴奋。""",
|
||||
"explanation": "通过新技术应用场景，观察个体对创新学习方式的态度。"
|
||||
"explanation": "通过新技术应用场景，观察个体对创新学习方式的态度。",
|
||||
},
|
||||
"场景3": {
|
||||
"scenario": """在社交媒体上，你看到一个朋友分享了一种新的生活方式：
|
||||
|
||||
「最近我在尝试'数字游牧'生活，就是一边远程工作一边环游世界。没有固定住所，住青旅或短租，认识来自世界各地的朋友。虽然有时会很不稳定，但这种自由的生活方式真的很棒！」
|
||||
「最近我在尝试'数字游牧'生活，就是一边远程工作一边环游世界。"""
|
||||
"""没有固定住所，住青旅或短租，认识来自世界各地的朋友。虽然有时会很不稳定，但这种自由的生活方式真的很棒！」
|
||||
|
||||
评论区里争论不断，有人向往这种生活，也有人觉得太冒险。""",
|
||||
"explanation": "通过另类生活方式，观察个体对非传统选择的态度。"
|
||||
"explanation": "通过另类生活方式，观察个体对非传统选择的态度。",
|
||||
},
|
||||
"场景4": {
|
||||
"scenario": """你的恋人突然提出了一个想法：
|
||||
@@ -169,7 +171,7 @@ PERSONALITY_SCENES = {
|
||||
恋人：「我们要不要尝试一下开放式关系？就是在保持彼此关系的同时，也允许和其他人发展感情。现在国外很多年轻人都这样。」
|
||||
|
||||
这个提议让你感到意外，你之前从未考虑过这种可能性。""",
|
||||
"explanation": "通过感情观念场景，观察个体对非传统关系模式的接受度。"
|
||||
"explanation": "通过感情观念场景，观察个体对非传统关系模式的接受度。",
|
||||
},
|
||||
"场景5": {
|
||||
"scenario": """在一次朋友聚会上，大家正在讨论未来职业规划：
|
||||
@@ -179,10 +181,9 @@ PERSONALITY_SCENES = {
|
||||
朋友B：「我想去学习生物科技，准备转行做人造肉研发。」
|
||||
|
||||
朋友C：「我在考虑加入一个区块链创业项目，虽然风险很大。」""",
|
||||
"explanation": "通过职业选择场景，观察个体对新兴领域的探索意愿。"
|
||||
}
|
||||
"explanation": "通过职业选择场景，观察个体对新兴领域的探索意愿。",
|
||||
},
|
||||
},
|
||||
|
||||
"宜人性": {
|
||||
"场景1": {
|
||||
"scenario": """在回家的公交车上，你遇到这样一幕：
|
||||
@@ -194,7 +195,7 @@ PERSONALITY_SCENES = {
|
||||
年轻人B：「现在的老年人真是...我看她包里还有菜，肯定是去菜市场买完菜回来的，这么多人都不知道叫子女开车接送。」
|
||||
|
||||
就在这时，老奶奶一个趔趄，差点摔倒。她扶住了扶手，但包里的东西洒了一些出来。""",
|
||||
"explanation": "这个场景通过公共场合的助人情境，体现个体的同理心和对他人需求的关注程度。"
|
||||
"explanation": "这个场景通过公共场合的助人情境，体现个体的同理心和对他人需求的关注程度。",
|
||||
},
|
||||
"场景2": {
|
||||
"scenario": """在班级群里，有同学发起为生病住院的同学捐款：
|
||||
@@ -204,7 +205,7 @@ PERSONALITY_SCENES = {
|
||||
同学B：「我觉得这是他家里的事，我们不方便参与吧。」
|
||||
|
||||
同学C：「但是都是同学一场，帮帮忙也是应该的。」""",
|
||||
"explanation": "通过同学互助场景，观察个体的助人意愿和同理心。"
|
||||
"explanation": "通过同学互助场景，观察个体的助人意愿和同理心。",
|
||||
},
|
||||
"场景3": {
|
||||
"scenario": """在一个网络讨论组里，有人发布了求助信息：
|
||||
@@ -215,7 +216,7 @@ PERSONALITY_SCENES = {
|
||||
「生活本来就是这样，想开点！」
|
||||
「你这样子太消极了，要积极面对。」
|
||||
「谁还没点烦心事啊，过段时间就好了。」""",
|
||||
"explanation": "通过网络互助场景，观察个体的共情能力和安慰方式。"
|
||||
"explanation": "通过网络互助场景，观察个体的共情能力和安慰方式。",
|
||||
},
|
||||
"场景4": {
|
||||
"scenario": """你的恋人向你倾诉工作压力：
|
||||
@@ -223,7 +224,7 @@ PERSONALITY_SCENES = {
|
||||
恋人：「最近工作真的好累，感觉快坚持不下去了...」
|
||||
|
||||
但今天你也遇到了很多烦心事，心情也不太好。""",
|
||||
"explanation": "通过感情关系场景，观察个体在自身状态不佳时的关怀能力。"
|
||||
"explanation": "通过感情关系场景，观察个体在自身状态不佳时的关怀能力。",
|
||||
},
|
||||
"场景5": {
|
||||
"scenario": """在一次团队项目中，新来的同事小王因为经验不足，造成了一个严重的错误。在部门会议上：
|
||||
@@ -231,11 +232,12 @@ PERSONALITY_SCENES = {
|
||||
主管：「这个错误造成了很大的损失，是谁负责的这部分？」
|
||||
|
||||
小王看起来很紧张，欲言又止。你知道是他造成的错误，同时你也是这个项目的共同负责人。""",
|
||||
"explanation": "通过职场情境，观察个体在面对他人过错时的态度和处理方式。"
|
||||
}
|
||||
}
|
||||
"explanation": "通过职场情境，观察个体在面对他人过错时的态度和处理方式。",
|
||||
},
|
||||
},
|
||||
}
|
||||
|
||||
|
||||
def get_scene_by_factor(factor: str) -> Dict:
|
||||
"""
|
||||
根据人格因子获取对应的情景测试
|
||||
@@ -248,6 +250,7 @@ def get_scene_by_factor(factor: str) -> Dict:
|
||||
"""
|
||||
return PERSONALITY_SCENES.get(factor, None)
|
||||
|
||||
|
||||
def get_all_scenes() -> Dict:
|
||||
"""
|
||||
获取所有情景测试
|
||||
|
||||
123
src/plugins/schedule/offline_llm.py
Normal file
123
src/plugins/schedule/offline_llm.py
Normal file
@@ -0,0 +1,123 @@
|
||||
import asyncio
|
||||
import os
|
||||
import time
|
||||
from typing import Tuple, Union
|
||||
|
||||
import aiohttp
|
||||
import requests
|
||||
from src.common.logger import get_module_logger
|
||||
|
||||
logger = get_module_logger("offline_llm")
|
||||
|
||||
|
||||
class LLMModel:
|
||||
def __init__(self, model_name="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")
|
||||
|
||||
logger.info(f"API URL: {self.base_url}") # 使用 logger 记录 base_url
|
||||
|
||||
def generate_response(self, prompt: str) -> Union[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"
|
||||
logger.info(f"Request URL: {api_url}") # 记录请求的 URL
|
||||
|
||||
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
|
||||
logger.warning(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 Exception as e:
|
||||
if retry < max_retries - 1: # 如果还有重试机会
|
||||
wait_time = base_wait_time * (2**retry)
|
||||
logger.error(f"[回复]请求失败，等待{wait_time}秒后重试... 错误: {str(e)}")
|
||||
time.sleep(wait_time)
|
||||
else:
|
||||
logger.error(f"请求失败: {str(e)}")
|
||||
return f"请求失败: {str(e)}", ""
|
||||
|
||||
logger.error("达到最大重试次数，请求仍然失败")
|
||||
return "达到最大重试次数，请求仍然失败", ""
|
||||
|
||||
async def generate_response_async(self, prompt: str) -> Union[str, Tuple[str, str]]:
|
||||
"""异步方式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"
|
||||
logger.info(f"Request URL: {api_url}") # 记录请求的 URL
|
||||
|
||||
max_retries = 3
|
||||
base_wait_time = 15
|
||||
|
||||
async with aiohttp.ClientSession() as session:
|
||||
for retry in range(max_retries):
|
||||
try:
|
||||
async with session.post(api_url, headers=headers, json=data) as response:
|
||||
if response.status == 429:
|
||||
wait_time = base_wait_time * (2**retry) # 指数退避
|
||||
logger.warning(f"遇到请求限制(429)，等待{wait_time}秒后重试...")
|
||||
await asyncio.sleep(wait_time)
|
||||
continue
|
||||
|
||||
response.raise_for_status() # 检查其他响应状态
|
||||
|
||||
result = await 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 Exception as e:
|
||||
if retry < max_retries - 1: # 如果还有重试机会
|
||||
wait_time = base_wait_time * (2**retry)
|
||||
logger.error(f"[回复]请求失败，等待{wait_time}秒后重试... 错误: {str(e)}")
|
||||
await asyncio.sleep(wait_time)
|
||||
else:
|
||||
logger.error(f"请求失败: {str(e)}")
|
||||
return f"请求失败: {str(e)}", ""
|
||||
|
||||
logger.error("达到最大重试次数，请求仍然失败")
|
||||
return "达到最大重试次数，请求仍然失败", ""
|
||||
191
src/plugins/schedule/schedule_generator copy.py
Normal file
191
src/plugins/schedule/schedule_generator copy.py
Normal file
@@ -0,0 +1,191 @@
|
||||
import datetime
|
||||
import json
|
||||
import re
|
||||
import os
|
||||
import sys
|
||||
from typing import Dict, Union
|
||||
|
||||
|
||||
# 添加项目根目录到 Python 路径
|
||||
root_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../.."))
|
||||
sys.path.append(root_path)
|
||||
|
||||
from src.common.database import db # noqa: E402
|
||||
from src.common.logger import get_module_logger # noqa: E402
|
||||
from src.plugins.schedule.offline_llm import LLMModel # noqa: E402
|
||||
from src.plugins.chat.config import global_config # noqa: E402
|
||||
|
||||
logger = get_module_logger("scheduler")
|
||||
|
||||
|
||||
class ScheduleGenerator:
|
||||
enable_output: bool = True
|
||||
|
||||
def __init__(self):
|
||||
# 使用离线LLM模型
|
||||
self.llm_scheduler = LLMModel(model_name="Pro/deepseek-ai/DeepSeek-V3", temperature=0.9)
|
||||
self.today_schedule_text = ""
|
||||
self.today_schedule = {}
|
||||
self.tomorrow_schedule_text = ""
|
||||
self.tomorrow_schedule = {}
|
||||
self.yesterday_schedule_text = ""
|
||||
self.yesterday_schedule = {}
|
||||
|
||||
async def initialize(self):
|
||||
today = datetime.datetime.now()
|
||||
tomorrow = datetime.datetime.now() + datetime.timedelta(days=1)
|
||||
yesterday = datetime.datetime.now() - datetime.timedelta(days=1)
|
||||
|
||||
self.today_schedule_text, self.today_schedule = await self.generate_daily_schedule(target_date=today)
|
||||
self.tomorrow_schedule_text, self.tomorrow_schedule = await self.generate_daily_schedule(
|
||||
target_date=tomorrow, read_only=True
|
||||
)
|
||||
self.yesterday_schedule_text, self.yesterday_schedule = await self.generate_daily_schedule(
|
||||
target_date=yesterday, read_only=True
|
||||
)
|
||||
|
||||
async def generate_daily_schedule(
|
||||
self, target_date: datetime.datetime = None, read_only: bool = False
|
||||
) -> Dict[str, str]:
|
||||
date_str = target_date.strftime("%Y-%m-%d")
|
||||
weekday = target_date.strftime("%A")
|
||||
|
||||
schedule_text = str
|
||||
|
||||
existing_schedule = db.schedule.find_one({"date": date_str})
|
||||
if existing_schedule:
|
||||
if self.enable_output:
|
||||
logger.debug(f"{date_str}的日程已存在:")
|
||||
schedule_text = existing_schedule["schedule"]
|
||||
# print(self.schedule_text)
|
||||
|
||||
elif not read_only:
|
||||
logger.debug(f"{date_str}的日程不存在，准备生成新的日程。")
|
||||
prompt = (
|
||||
f"""我是{global_config.BOT_NICKNAME}，{global_config.PROMPT_SCHEDULE_GEN}，请为我生成{date_str}（{weekday}）的日程安排，包括："""
|
||||
+ """
|
||||
1. 早上的学习和工作安排
|
||||
2. 下午的活动和任务
|
||||
3. 晚上的计划和休息时间
|
||||
请按照时间顺序列出具体时间点和对应的活动，用一个时间点而不是时间段来表示时间，用JSON格式返回日程表，
|
||||
仅返回内容，不要返回注释，不要添加任何markdown或代码块样式，时间采用24小时制，
|
||||
格式为{"时间": "活动","时间": "活动",...}。"""
|
||||
)
|
||||
|
||||
try:
|
||||
schedule_text, _ = self.llm_scheduler.generate_response(prompt)
|
||||
db.schedule.insert_one({"date": date_str, "schedule": schedule_text})
|
||||
self.enable_output = True
|
||||
except Exception as e:
|
||||
logger.error(f"生成日程失败: {str(e)}")
|
||||
schedule_text = "生成日程时出错了"
|
||||
# print(self.schedule_text)
|
||||
else:
|
||||
if self.enable_output:
|
||||
logger.debug(f"{date_str}的日程不存在。")
|
||||
schedule_text = "忘了"
|
||||
|
||||
return schedule_text, None
|
||||
|
||||
schedule_form = self._parse_schedule(schedule_text)
|
||||
return schedule_text, schedule_form
|
||||
|
||||
def _parse_schedule(self, schedule_text: str) -> Union[bool, Dict[str, str]]:
|
||||
"""解析日程文本，转换为时间和活动的字典"""
|
||||
try:
|
||||
reg = r"\{(.|\r|\n)+\}"
|
||||
matched = re.search(reg, schedule_text)[0]
|
||||
schedule_dict = json.loads(matched)
|
||||
return schedule_dict
|
||||
except json.JSONDecodeError:
|
||||
logger.exception("解析日程失败: {}".format(schedule_text))
|
||||
return False
|
||||
|
||||
def _parse_time(self, time_str: str) -> str:
|
||||
"""解析时间字符串，转换为时间"""
|
||||
return datetime.datetime.strptime(time_str, "%H:%M")
|
||||
|
||||
def get_current_task(self) -> str:
|
||||
"""获取当前时间应该进行的任务"""
|
||||
current_time = datetime.datetime.now().strftime("%H:%M")
|
||||
|
||||
# 找到最接近当前时间的任务
|
||||
closest_time = None
|
||||
min_diff = float("inf")
|
||||
|
||||
# 检查今天的日程
|
||||
if not self.today_schedule:
|
||||
return "摸鱼"
|
||||
for time_str in self.today_schedule.keys():
|
||||
diff = abs(self._time_diff(current_time, time_str))
|
||||
if closest_time is None or diff < min_diff:
|
||||
closest_time = time_str
|
||||
min_diff = diff
|
||||
|
||||
# 检查昨天的日程中的晚间任务
|
||||
if self.yesterday_schedule:
|
||||
for time_str in self.yesterday_schedule.keys():
|
||||
if time_str >= "20:00": # 只考虑晚上8点之后的任务
|
||||
# 计算与昨天这个时间点的差异（需要加24小时）
|
||||
diff = abs(self._time_diff(current_time, time_str))
|
||||
if diff < min_diff:
|
||||
closest_time = time_str
|
||||
min_diff = diff
|
||||
return closest_time, self.yesterday_schedule[closest_time]
|
||||
|
||||
if closest_time:
|
||||
return closest_time, self.today_schedule[closest_time]
|
||||
return "摸鱼"
|
||||
|
||||
def _time_diff(self, time1: str, time2: str) -> int:
|
||||
"""计算两个时间字符串之间的分钟差"""
|
||||
if time1 == "24:00":
|
||||
time1 = "23:59"
|
||||
if time2 == "24:00":
|
||||
time2 = "23:59"
|
||||
t1 = datetime.datetime.strptime(time1, "%H:%M")
|
||||
t2 = datetime.datetime.strptime(time2, "%H:%M")
|
||||
diff = int((t2 - t1).total_seconds() / 60)
|
||||
# 考虑时间的循环性
|
||||
if diff < -720:
|
||||
diff += 1440 # 加一天的分钟
|
||||
elif diff > 720:
|
||||
diff -= 1440 # 减一天的分钟
|
||||
# print(f"时间1[{time1}]: 时间2[{time2}]，差值[{diff}]分钟")
|
||||
return diff
|
||||
|
||||
def print_schedule(self):
|
||||
"""打印完整的日程安排"""
|
||||
if not self._parse_schedule(self.today_schedule_text):
|
||||
logger.warning("今日日程有误，将在下次运行时重新生成")
|
||||
db.schedule.delete_one({"date": datetime.datetime.now().strftime("%Y-%m-%d")})
|
||||
else:
|
||||
logger.info("=== 今日日程安排 ===")
|
||||
for time_str, activity in self.today_schedule.items():
|
||||
logger.info(f"时间[{time_str}]: 活动[{activity}]")
|
||||
logger.info("==================")
|
||||
self.enable_output = False
|
||||
|
||||
|
||||
async def main():
|
||||
# 使用示例
|
||||
scheduler = ScheduleGenerator()
|
||||
await scheduler.initialize()
|
||||
scheduler.print_schedule()
|
||||
print("\n当前任务：")
|
||||
print(await scheduler.get_current_task())
|
||||
|
||||
print("昨天日程：")
|
||||
print(scheduler.yesterday_schedule)
|
||||
print("今天日程：")
|
||||
print(scheduler.today_schedule)
|
||||
print("明天日程：")
|
||||
print(scheduler.tomorrow_schedule)
|
||||
|
||||
# 当作为组件导入时使用的实例
|
||||
bot_schedule = ScheduleGenerator()
|
||||
|
||||
if __name__ == "__main__":
|
||||
import asyncio
|
||||
# 当直接运行此文件时执行
|
||||
asyncio.run(main())
|
||||
@@ -5,8 +5,9 @@ from typing import Dict, Union
|
||||
|
||||
from nonebot import get_driver
|
||||
|
||||
from src.plugins.chat.config import global_config
|
||||
# 添加项目根目录到 Python 路径
|
||||
|
||||
from src.plugins.chat.config import global_config
|
||||
from ...common.database import db # 使用正确的导入语法
|
||||
from ..models.utils_model import LLM_request
|
||||
from src.common.logger import get_module_logger
|
||||
@@ -73,7 +74,7 @@ class ScheduleGenerator:
|
||||
)
|
||||
|
||||
try:
|
||||
schedule_text, _ = await self.llm_scheduler.generate_response(prompt)
|
||||
schedule_text, _, _ = await self.llm_scheduler.generate_response(prompt)
|
||||
db.schedule.insert_one({"date": date_str, "schedule": schedule_text})
|
||||
self.enable_output = True
|
||||
except Exception as e:
|
||||
@@ -165,24 +166,5 @@ class ScheduleGenerator:
|
||||
logger.info(f"时间[{time_str}]: 活动[{activity}]")
|
||||
logger.info("==================")
|
||||
self.enable_output = False
|
||||
|
||||
|
||||
# def main():
|
||||
# # 使用示例
|
||||
# scheduler = ScheduleGenerator()
|
||||
# # new_schedule = scheduler.generate_daily_schedule()
|
||||
# scheduler.print_schedule()
|
||||
# print("\n当前任务：")
|
||||
# print(scheduler.get_current_task())
|
||||
|
||||
# print("昨天日程：")
|
||||
# print(scheduler.yesterday_schedule)
|
||||
# print("今天日程：")
|
||||
# print(scheduler.today_schedule)
|
||||
# print("明天日程：")
|
||||
# print(scheduler.tomorrow_schedule)
|
||||
|
||||
# if __name__ == "__main__":
|
||||
# main()
|
||||
|
||||
# 当作为组件导入时使用的实例
|
||||
bot_schedule = ScheduleGenerator()
|
||||
|
||||
@@ -1,8 +1,6 @@
|
||||
HOST=127.0.0.1
|
||||
PORT=8080
|
||||
|
||||
ENABLE_ADVANCE_OUTPUT=false
|
||||
|
||||
# 插件配置
|
||||
PLUGINS=["src2.plugins.chat"]
|
||||
|
||||
@@ -31,6 +29,7 @@ CHAT_ANY_WHERE_KEY=
|
||||
SILICONFLOW_KEY=
|
||||
|
||||
# 定义日志相关配置
|
||||
SIMPLE_OUTPUT=true # 精简控制台输出格式
|
||||
CONSOLE_LOG_LEVEL=INFO # 自定义日志的默认控制台输出日志级别
|
||||
FILE_LOG_LEVEL=DEBUG # 自定义日志的默认文件输出日志级别
|
||||
DEFAULT_CONSOLE_LOG_LEVEL=SUCCESS # 原生日志的控制台输出日志级别（nonebot就是这一类）
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
[inner]
|
||||
version = "0.0.10"
|
||||
version = "0.0.11"
|
||||
|
||||
#以下是给开发人员阅读的，一般用户不需要阅读
|
||||
#如果你想要修改配置文件，请在修改后将version的值进行变更
|
||||
@@ -66,12 +66,15 @@ model_r1_distill_probability = 0.1 # 麦麦回答时选择次要回复模型3
|
||||
max_response_length = 1024 # 麦麦回答的最大token数
|
||||
|
||||
[willing]
|
||||
willing_mode = "classical"
|
||||
# willing_mode = "dynamic"
|
||||
# willing_mode = "custom"
|
||||
willing_mode = "classical" # 回复意愿模式 经典模式
|
||||
# willing_mode = "dynamic" # 动态模式(可能不兼容)
|
||||
# willing_mode = "custom" # 自定义模式（可自行调整
|
||||
|
||||
[memory]
|
||||
build_memory_interval = 2000 # 记忆构建间隔 单位秒 间隔越低，麦麦学习越多，但是冗余信息也会增多
|
||||
build_memory_distribution = [4,2,0.6,24,8,0.4] # 记忆构建分布，参数：分布1均值，标准差，权重，分布2均值，标准差，权重
|
||||
build_memory_sample_num = 10 # 采样数量，数值越高记忆采样次数越多
|
||||
build_memory_sample_length = 20 # 采样长度，数值越高一段记忆内容越丰富
|
||||
memory_compress_rate = 0.1 # 记忆压缩率 控制记忆精简程度 建议保持默认,调高可以获得更多信息，但是冗余信息也会增多
|
||||
|
||||
forget_memory_interval = 1000 # 记忆遗忘间隔 单位秒 间隔越低，麦麦遗忘越频繁，记忆更精简，但更难学习
|
||||
@@ -109,9 +112,7 @@ tone_error_rate=0.2 # 声调错误概率
|
||||
word_replace_rate=0.006 # 整词替换概率
|
||||
|
||||
[others]
|
||||
enable_advance_output = false # 是否启用高级输出
|
||||
enable_kuuki_read = true # 是否启用读空气功能
|
||||
enable_debug_output = false # 是否启用调试输出
|
||||
enable_friend_chat = false # 是否启用好友聊天
|
||||
|
||||
[groups]
|
||||
@@ -120,59 +121,67 @@ talk_allowed = [
|
||||
123,
|
||||
] #可以回复消息的群
|
||||
talk_frequency_down = [] #降低回复频率的群
|
||||
ban_user_id = [] #禁止回复消息的QQ号
|
||||
ban_user_id = [] #禁止回复和读取消息的QQ号
|
||||
|
||||
[remote] #测试功能，发送统计信息，主要是看全球有多少只麦麦
|
||||
[remote] #发送统计信息，主要是看全球有多少只麦麦
|
||||
enable = true
|
||||
|
||||
|
||||
#下面的模型若使用硅基流动则不需要更改，使用ds官方则改成.env.prod自定义的宏，使用自定义模型则选择定位相似的模型自己填写
|
||||
#推理模型：
|
||||
#推理模型
|
||||
|
||||
[model.llm_reasoning] #回复模型1 主要回复模型
|
||||
name = "Pro/deepseek-ai/DeepSeek-R1"
|
||||
# name = "Qwen/QwQ-32B"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0 #模型的输入价格（非必填，可以记录消耗）
|
||||
pri_out = 0 #模型的输出价格（非必填，可以记录消耗）
|
||||
pri_in = 4 #模型的输入价格（非必填，可以记录消耗）
|
||||
pri_out = 16 #模型的输出价格（非必填，可以记录消耗）
|
||||
|
||||
[model.llm_reasoning_minor] #回复模型3 次要回复模型
|
||||
name = "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 1.26 #模型的输入价格（非必填，可以记录消耗）
|
||||
pri_out = 1.26 #模型的输出价格（非必填，可以记录消耗）
|
||||
|
||||
#非推理模型
|
||||
|
||||
[model.llm_normal] #V3 回复模型2 次要回复模型
|
||||
name = "Pro/deepseek-ai/DeepSeek-V3"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 2 #模型的输入价格（非必填，可以记录消耗）
|
||||
pri_out = 8 #模型的输出价格（非必填，可以记录消耗）
|
||||
|
||||
[model.llm_normal_minor] #V2.5
|
||||
name = "deepseek-ai/DeepSeek-V2.5"
|
||||
provider = "SILICONFLOW"
|
||||
|
||||
[model.llm_emotion_judge] #主题判断 0.7/m
|
||||
[model.llm_emotion_judge] #表情包判断
|
||||
name = "Qwen/Qwen2.5-14B-Instruct"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0.7
|
||||
pri_out = 0.7
|
||||
|
||||
[model.llm_topic_judge] #主题判断：建议使用qwen2.5 7b
|
||||
[model.llm_topic_judge] #记忆主题判断：建议使用qwen2.5 7b
|
||||
name = "Pro/Qwen/Qwen2.5-7B-Instruct"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0
|
||||
pri_out = 0
|
||||
|
||||
[model.llm_summary_by_topic] #建议使用qwen2.5 32b 及以上
|
||||
[model.llm_summary_by_topic] #概括模型，建议使用qwen2.5 32b 及以上
|
||||
name = "Qwen/Qwen2.5-32B-Instruct"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0
|
||||
pri_out = 0
|
||||
pri_in = 1.26
|
||||
pri_out = 1.26
|
||||
|
||||
[model.moderation] #内容审核 未启用
|
||||
[model.moderation] #内容审核，开发中
|
||||
name = ""
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0
|
||||
pri_out = 0
|
||||
pri_in = 1.0
|
||||
pri_out = 2.0
|
||||
|
||||
# 识图模型
|
||||
|
||||
[model.vlm] #图像识别 0.35/m
|
||||
name = "Pro/Qwen/Qwen2-VL-7B-Instruct"
|
||||
[model.vlm] #图像识别
|
||||
name = "Pro/Qwen/Qwen2.5-VL-7B-Instruct"
|
||||
provider = "SILICONFLOW"
|
||||
pri_in = 0.35
|
||||
pri_out = 0.35
|
||||
|
||||
#嵌入模型
|
||||
|
||||
|
||||
41
webui.py
41
webui.py
@@ -4,11 +4,14 @@ import toml
|
||||
import signal
|
||||
import sys
|
||||
import requests
|
||||
|
||||
try:
|
||||
from src.common.logger import get_module_logger
|
||||
|
||||
logger = get_module_logger("webui")
|
||||
except ImportError:
|
||||
from loguru import logger
|
||||
|
||||
# 检查并创建日志目录
|
||||
log_dir = "logs/webui"
|
||||
if not os.path.exists(log_dir):
|
||||
@@ -24,11 +27,13 @@ import ast
|
||||
from packaging import version
|
||||
from decimal import Decimal
|
||||
|
||||
|
||||
def signal_handler(signum, frame):
|
||||
"""处理 Ctrl+C 信号"""
|
||||
logger.info("收到终止信号，正在关闭 Gradio 服务器...")
|
||||
sys.exit(0)
|
||||
|
||||
|
||||
# 注册信号处理器
|
||||
signal.signal(signal.SIGINT, signal_handler)
|
||||
|
||||
@@ -66,9 +71,18 @@ else:
|
||||
|
||||
HAVE_ONLINE_STATUS_VERSION = version.parse("0.0.9")
|
||||
|
||||
# 定义意愿模式可选项
|
||||
WILLING_MODE_CHOICES = [
|
||||
"classical",
|
||||
"dynamic",
|
||||
"custom",
|
||||
]
|
||||
|
||||
|
||||
# 添加WebUI配置文件版本
|
||||
WEBUI_VERSION = version.parse("0.0.9")
|
||||
|
||||
|
||||
# ==============================================
|
||||
# env环境配置文件读取部分
|
||||
def parse_env_config(config_file):
|
||||
@@ -84,10 +98,14 @@ def parse_env_config(config_file):
|
||||
# 逐行处理配置
|
||||
for line in lines:
|
||||
line = line.strip()
|
||||
# 忽略空行和注释
|
||||
# 忽略空行和注释行
|
||||
if not line or line.startswith("#"):
|
||||
continue
|
||||
|
||||
# 处理行尾注释
|
||||
if "#" in line:
|
||||
line = line.split("#")[0].strip()
|
||||
|
||||
# 拆分键值对
|
||||
key, value = line.split("=", 1)
|
||||
|
||||
@@ -522,6 +540,7 @@ def save_message_and_emoji_config(
|
||||
|
||||
|
||||
def save_response_model_config(
|
||||
t_willing_mode,
|
||||
t_model_r1_probability,
|
||||
t_model_r2_probability,
|
||||
t_model_r3_probability,
|
||||
@@ -543,6 +562,8 @@ def save_response_model_config(
|
||||
t_vlm_model_name,
|
||||
t_vlm_model_provider,
|
||||
):
|
||||
if PARSED_CONFIG_VERSION >= version.parse("0.0.10"):
|
||||
config_data["willing"]["willing_mode"] = t_willing_mode
|
||||
config_data["response"]["model_r1_probability"] = t_model_r1_probability
|
||||
config_data["response"]["model_v3_probability"] = t_model_r2_probability
|
||||
config_data["response"]["model_r1_distill_probability"] = t_model_r3_probability
|
||||
@@ -1182,6 +1203,23 @@ with gr.Blocks(title="MaimBot配置文件编辑") as app:
|
||||
with gr.Column(scale=3):
|
||||
with gr.Row():
|
||||
gr.Markdown("""### 回复设置""")
|
||||
if PARSED_CONFIG_VERSION >= version.parse("0.0.10"):
|
||||
with gr.Row():
|
||||
gr.Markdown("""#### 回复意愿模式""")
|
||||
with gr.Row():
|
||||
gr.Markdown("""回复意愿模式说明：\n
|
||||
classical为经典回复意愿管理器\n
|
||||
dynamic为动态意愿管理器\n
|
||||
custom为自定义意愿管理器
|
||||
""")
|
||||
with gr.Row():
|
||||
willing_mode = gr.Dropdown(
|
||||
choices=WILLING_MODE_CHOICES,
|
||||
value=config_data["willing"]["willing_mode"],
|
||||
label="回复意愿模式",
|
||||
)
|
||||
else:
|
||||
willing_mode = gr.Textbox(visible=False, value="disabled")
|
||||
with gr.Row():
|
||||
model_r1_probability = gr.Slider(
|
||||
minimum=0,
|
||||
@@ -1355,6 +1393,7 @@ with gr.Blocks(title="MaimBot配置文件编辑") as app:
|
||||
save_model_btn.click(
|
||||
save_response_model_config,
|
||||
inputs=[
|
||||
willing_mode,
|
||||
model_r1_probability,
|
||||
model_r2_probability,
|
||||
model_r3_probability,
|
||||
|
||||
Reference in New Issue
Block a user