update: requirements.txt

perf: 使用 tictoken 预先计算 token

CODE_OF_CONDUCT.md

@@ -0,0 +1,128 @@
+# Contributor Covenant Code of Conduct
+
+## Our Pledge
+
+We as members, contributors, and leaders pledge to make participation in our
+community a harassment-free experience for everyone, regardless of age, body
+size, visible or invisible disability, ethnicity, sex characteristics, gender
+identity and expression, level of experience, education, socio-economic status,
+nationality, personal appearance, race, religion, or sexual identity
+and orientation.
+
+We pledge to act and interact in ways that contribute to an open, welcoming,
+diverse, inclusive, and healthy community.
+
+## Our Standards
+
+Examples of behavior that contributes to a positive environment for our
+community include:
+
+* Demonstrating empathy and kindness toward other people
+* Being respectful of differing opinions, viewpoints, and experiences
+* Giving and gracefully accepting constructive feedback
+* Accepting responsibility and apologizing to those affected by our mistakes,
+  and learning from the experience
+* Focusing on what is best not just for us as individuals, but for the
+  overall community
+
+Examples of unacceptable behavior include:
+
+* The use of sexualized language or imagery, and sexual attention or
+  advances of any kind
+* Trolling, insulting or derogatory comments, and personal or political attacks
+* Public or private harassment
+* Publishing others' private information, such as a physical or email
+  address, without their explicit permission
+* Other conduct which could reasonably be considered inappropriate in a
+  professional setting
+
+## Enforcement Responsibilities
+
+Community leaders are responsible for clarifying and enforcing our standards of
+acceptable behavior and will take appropriate and fair corrective action in
+response to any behavior that they deem inappropriate, threatening, offensive,
+or harmful.
+
+Community leaders have the right and responsibility to remove, edit, or reject
+comments, commits, code, wiki edits, issues, and other contributions that are
+not aligned to this Code of Conduct, and will communicate reasons for moderation
+decisions when appropriate.
+
+## Scope
+
+This Code of Conduct applies within all community spaces, and also applies when
+an individual is officially representing the community in public spaces.
+Examples of representing our community include using an official e-mail address,
+posting via an official social media account, or acting as an appointed
+representative at an online or offline event.
+
+## Enforcement
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may be
+reported to the community leaders responsible for enforcement at
+SoulterL@outlook.com.
+All complaints will be reviewed and investigated promptly and fairly.
+
+All community leaders are obligated to respect the privacy and security of the
+reporter of any incident.
+
+## Enforcement Guidelines
+
+Community leaders will follow these Community Impact Guidelines in determining
+the consequences for any action they deem in violation of this Code of Conduct:
+
+### 1. Correction
+
+**Community Impact**: Use of inappropriate language or other behavior deemed
+unprofessional or unwelcome in the community.
+
+**Consequence**: A private, written warning from community leaders, providing
+clarity around the nature of the violation and an explanation of why the
+behavior was inappropriate. A public apology may be requested.
+
+### 2. Warning
+
+**Community Impact**: A violation through a single incident or series
+of actions.
+
+**Consequence**: A warning with consequences for continued behavior. No
+interaction with the people involved, including unsolicited interaction with
+those enforcing the Code of Conduct, for a specified period of time. This
+includes avoiding interactions in community spaces as well as external channels
+like social media. Violating these terms may lead to a temporary or
+permanent ban.
+
+### 3. Temporary Ban
+
+**Community Impact**: A serious violation of community standards, including
+sustained inappropriate behavior.
+
+**Consequence**: A temporary ban from any sort of interaction or public
+communication with the community for a specified period of time. No public or
+private interaction with the people involved, including unsolicited interaction
+with those enforcing the Code of Conduct, is allowed during this period.
+Violating these terms may lead to a permanent ban.
+
+### 4. Permanent Ban
+
+**Community Impact**: Demonstrating a pattern of violation of community
+standards, including sustained inappropriate behavior,  harassment of an
+individual, or aggression toward or disparagement of classes of individuals.
+
+**Consequence**: A permanent ban from any sort of public interaction within
+the community.
+
+## Attribution
+
+This Code of Conduct is adapted from the [Contributor Covenant][homepage],
+version 2.0, available at
+https://www.contributor-covenant.org/version/2/0/code_of_conduct.html.
+
+Community Impact Guidelines were inspired by [Mozilla's code of conduct
+enforcement ladder](https://github.com/mozilla/diversity).
+
+[homepage]: https://www.contributor-covenant.org
+
+For answers to common questions about this code of conduct, see the FAQ at
+https://www.contributor-covenant.org/faq. Translations are available at
+https://www.contributor-covenant.org/translations.

README.md

@@ -1,28 +1,60 @@
-## ⭐体验
+<div align="center">
 
-使用手机QQ扫码加入QQ频道(频道名: GPT机器人 | 频道号: x42d56aki2)
+<img src="https://socialify.git.ci/Soulter/QQChannelChatGPT/image?description=1&forks=1&issues=1&language=1&name=1&owner=1&pattern=Circuit%20Board&stargazers=1&theme=Light" alt="QQChannelChatGPT" width="600" height="300" />
 
-<img src="https://user-images.githubusercontent.com/37870767/230417115-9dd3c9d5-6b6b-4928-8fe3-82f559208aab.JPG" width="200"></img>
+<!-- [![Language](https://img.shields.io/badge/language-python-green.svg?style=plastic)](https://www.python.org/)
+[![License](https://img.shields.io/badge/license-AGPL3-orange.svg?style=plastic)](https://github.com/Soulter/QQChannelChatGPT/blob/master/LICENSE)
+![Python](https://img.shields.io/badge/python-3.9+-blue) -->
+ 
+基于go-cqhttp和官方QQ频道SDK的QQ机器人项目。支持ChatGPT、Claude、HuggingChat、Bard大模型。一次部署，同时使用。
 
-**Windows用户推荐Windows一键安装，请前往Release下载最新版本（Beta）**
-有报错请先看issue，解决不了再在频道内反馈。
+部署文档：https://github.com/Soulter/QQChannelChatGPT/wiki
+ 
+欢迎加群讨论 | **QQ群号：322154837** | **频道号: x42d56aki2** |
 
-## ⭐功能：
+<!-- <img src="https://user-images.githubusercontent.com/37870767/230417115-9dd3c9d5-6b6b-4928-8fe3-82f559208aab.JPG" width="300"></img> -->
 
-近期新功能：
-- 支持一键切换语言模型（使用/bing /revgpt /gpt分别可以切换newbing、逆向ChatGPT、官方ChatGPT模型）
-- 热更新
-- 接入QQ，支持在QQ上和QQ频道上同时聊天！https://github.com/Soulter/QQChannelChatGPT/issues/82
+</div>
 
-支持的AI语言模型（请在`configs/config.yaml`下配置）：
-- 逆向ChatGPT库
-- 官方ChatGPT AI
-- 文心一言（即将支持，链接https://github.com/Soulter/ERNIEBot 欢迎Star）
-- NewBing
-- Bard （即将支持）
+## 🤔您可能想了解的
+- **如何部署？** [帮助文档](https://github.com/Soulter/QQChannelChatGPT/wiki) (部署不成功欢迎进群捞人解决<3)
+- **go-cqhttp启动不成功、报登录失败？** [在这里搜索解决方法](https://github.com/Mrs4s/go-cqhttp/issues)
+- **程序闪退/机器人启动不成功？** [提交issue或加群反馈](https://github.com/Soulter/QQChannelChatGPT/issues)
+- **如何开启ChatGPT、Bard、Claude等语言模型？** [查看帮助](https://github.com/Soulter/QQChannelChatGPT/wiki/%E8%A1%A5%E5%85%85%EF%BC%9A%E5%A6%82%E4%BD%95%E5%BC%80%E5%90%AFChatGPT%E3%80%81Bard%E3%80%81Claude%E7%AD%89%E8%AF%AD%E8%A8%80%E6%A8%A1%E5%9E%8B%EF%BC%9F)
 
-部署QQ频道机器人教程链接：https://soulter.top/posts/qpdg.html
+## 🧩功能：
 
+🌍支持的AI语言模型一览：
+
+**文字模型**
+
+- OpenAI GPT-3模型（原生支持）
+- OpenAI GPT-3.5模型（原生支持）
+- OpenAI GPT-4模型（原生支持）
+- ChatGPT网页版 GPT-3.5模型（免费，原生支持）
+- ChatGPT网页版 GPT-4模型（需订阅Plus账户，原生支持）
+- Bing（免费，原生支持）
+- Claude模型（免费，由[LLMs插件](https://github.com/Soulter/llms)支持）
+- HuggingChat模型（免费，由[LLMs插件](https://github.com/Soulter/llms)支持）
+- Google Bard（免费，由[LLMs插件](https://github.com/Soulter/llms)支持）
+
+**图片生成**
+
+- NovelAI/Naifu (免费，由[AIDraw插件](https://github.com/Soulter/aidraw)支持)
+
+
+🌍机器人支持的能力一览：
+- 同时部署机器人到QQ和QQ频道
+- 大模型对话
+- 大模型网页搜索能力 **(目前仅支持OpenAI系的模型，最新版本下使用web on指令打开)**
+- 插件安装（在QQ或QQ频道聊天框内输入`plugin`了解详情）
+- 回复文字图片渲染（以图片markdown格式回复，**大幅度降低被风控概率**，需手动在`cmd_config.json`内开启qq_pic_mode）
+- 人格设置
+- 关键词回复
+- 热更新（更新本项目时**仅需**在QQ或QQ频道聊天框内输入`update latest r`）
+- Windows一键部署（https://github.com/Soulter/QQChatGPTLauncher/releases/latest）
+
+<!-- 
 ### 基本功能
 <details> 
  <summary>✅ 回复符合上下文</summary>
@@ -77,11 +109,35 @@
 
    -  QQ频道机器人框架为QQ官方开源的框架，稳定。
 
-</details>
+</details> -->
 
-> 关于token：token就相当于是AI中的单词数（但是不等于单词数），`text-davinci-003`模型中最大可以支持`4097`个token。在发送信息时，这个机器人会将用户的历史聊天记录打包发送给ChatGPT，因此，`token`也会相应的累加，为了保证聊天的上下文的逻辑性，就有了缓存token。
-### 指令功能
-需要先`@`机器人之后再输入指令
+<!-- > 关于token：token就相当于是AI中的单词数（但是不等于单词数），`text-davinci-003`模型中最大可以支持`4097`个token。在发送信息时，这个机器人会将用户的历史聊天记录打包发送给ChatGPT，因此，`token`也会相应的累加，为了保证聊天的上下文的逻辑性，就有了缓存token。 -->
+
+### 🛠️ 插件支持
+
+本项目支持接入插件。
+
+> 使用`plugin i 插件GitHub链接`即可安装。
+
+插件开发教程：https://github.com/Soulter/QQChannelChatGPT/wiki/%E5%9B%9B%E3%80%81%E5%BC%80%E5%8F%91%E6%8F%92%E4%BB%B6
+
+部分插件：
+
+- `LLMS`: https://github.com/Soulter/llms | Claude, HuggingChat 大语言模型接入。
+ 
+- `GoodPlugins`: https://github.com/Soulter/goodplugins | 随机动漫图片、搜番、喜报生成器等等
+
+- `sysstat`: https://github.com/Soulter/sysstatqcbot | 查看系统状态
+
+- `BiliMonitor`: https://github.com/Soulter/BiliMonitor | 订阅B站动态
+  
+- `liferestart`: https://github.com/Soulter/liferestart | 人生重开模拟器
+
+<!-- 
+### 指令
+
+#### OpenAI官方API
+在频道内需要先`@`机器人之后再输入指令；在QQ中暂时需要在消息前加上`ai `，不需要@
 - `/reset`重置prompt
 - `/his`查看历史记录（每个用户都有独立的会话）
 - `/his [页码数]`查看不同页码的历史记录。例如`/his 2`查看第2页
@@ -91,33 +147,196 @@
 - `/help` 查看帮助
 - `/key` 动态添加key
 - `/set` 人格设置面板
+- `/keyword nihao 你好` 设置关键词回复。nihao->你好
+- `/bing` 切换为bing
+- `/revgpt` 切换为ChatGPT逆向库
+- `/画` 画画
+
+#### Bing语言模型
+- `/reset`重置prompt
+- `/gpt` 切换为OpenAI官方API
+- `/revgpt` 切换为ChatGPT逆向库
+
+#### 逆向ChatGPT库语言模型
+- `/gpt` 切换为OpenAI官方API
+- `/bing` 切换为bing
+
+* 切换模型指令支持临时回复。如`/bing 你好`将会临时使用一次bing模型 -->
 
 ## 📰使用方法：
 
-**详细部署教程链接**https://soulter.top/posts/qpdg.html
+使用文档：https://github.com/Soulter/QQChannelChatGPT/wiki
 
+**Windows用户可以使用启动器一键安装，请前往Release下载最新版本（Beta）**
+<!-- 
 ### 安装第三方库
 
-使用Python的pip工具安装
-- `qq-botpy` （QQ频道官方Python SDK）
-- `openai` (OpenAI Python SDK)
-
 ```shell
 pip install -r requirements.txt
 ```
-> ⚠注意，由于qq-botpy库需要运行在`Python 3.8+`的版本上，因此本项目也需要在此之上运行
+> ⚠Python版本应>=3.9
 
 ### 配置
 
-- 获得 OpenAI的key [OpenAI](https://beta.openai.com/)
-- 获得 QQ开放平台下QQ频道机器人的token和appid [QQ开放平台](https://q.qq.com/)，一个QQ频道机器人（很容易创建~）
-- 在configs/config.yaml下进行配置
+**详细部署教程链接：**https://github.com/Soulter/QQChannelChatGPT/wiki
 
 ### 启动
-- 启动main.py
+- 启动main.py -->
+
+## 🙇‍感谢
+本项目使用了一下项目:
+
+[ChatGPT by acheong08](https://github.com/acheong08/ChatGPT)
+
+[EdgeGPT by acheong08](https://github.com/acheong08/EdgeGPT)
+
+[go-cqhttp by Mrs4s](https://github.com/Mrs4s/go-cqhttp)
+
+[nakuru-project by Lxns-Network](https://github.com/Lxns-Network/nakuru-project)
+
+<!-- ## 👀部分演示截图
+
+帮助中心（`help`指令）
+![)F%2VQA`O)`4BHTXZ653(~9](https://github.com/Soulter/QQChannelChatGPT/assets/37870767/57eaa8c6-6962-4940-823c-2e26b5206cf5)
+
+ -->
+## ⚙配置文件说明：
+```yaml
+# 如果你不知道怎么部署，请查看https://github.com/Soulter/QQChannelChatGPT/wiki
+# 不一定需要key了，如果你没有key但有openAI账号或者必应账号，可以考虑使用下面的逆向库
 
 
-## DEMO
-![1.jpg](screenshots/1.jpg)
-![3.jpg](screenshots/3.jpg)
-![2.jpg](screenshots/2.jpg)
+###############平台设置#################
+
+# QQ频道机器人
+# QQ开放平台的appid和令牌
+# q.qq.com
+# enable为true则启用，false则不启用
+qqbot:
+  enable: true
+  appid: 
+  token: 
+
+# QQ机器人
+# enable为true则启用，false则不启用
+# 需要安装GO-CQHTTP配合使用。
+# 文档：https://docs.go-cqhttp.org/
+# 请将go-cqhttp的配置文件的sever部分粘贴为以下内容，否则无法使用
+# 请先启动go-cqhttp再启动本程序
+# 
+# servers:
+#   - http:
+#       host: 127.0.0.1
+#       version: 0
+#       port: 5700
+#       timeout: 5
+#   - ws:
+#       address: 127.0.0.1:6700
+#       middlewares:
+#         <<: *default
+gocqbot:
+  enable: false
+
+# 设置是否一个人一个会话
+uniqueSessionMode: false
+# QChannelBot 的版本，请勿修改此字段，否则可能产生一些bug
+version: 3.0
+# [Beta] 转储历史记录时间间隔(分钟)
+dump_history_interval: 10
+# 一个用户只能在time秒内发送count条消息
+limit:
+  time: 60
+  count: 5
+# 公告
+notice: "此机器人由Github项目QQChannelChatGPT驱动。"
+# 是否打开私信功能
+# 设置为true则频道成员可以私聊机器人。
+# 设置为false则频道成员不能私聊机器人。
+direct_message_mode: true
+
+# 系统代理
+# http_proxy: http://localhost:7890
+# https_proxy: http://localhost:7890
+
+# 自定义回复前缀，如[Rev]或其他，务必加引号以防止不必要的bug。
+reply_prefix:
+  openai_official: "[GPT]"
+  rev_chatgpt: "[Rev]"
+  rev_edgegpt: "[RevBing]"
+
+# 百度内容审核服务
+# 新用户免费5万次调用。https://cloud.baidu.com/doc/ANTIPORN/index.html
+baidu_aip:
+  enable: false
+  app_id: 
+  api_key: 
+  secret_key: 
+
+
+
+
+###############语言模型设置#################
+
+
+# OpenAI官方API
+# 注意：已支持多key自动切换，方法：
+# key:
+#   - sk-xxxxxx
+#   - sk-xxxxxx
+# 在下方非注释的地方使用以上格式
+# 关于api_base：可以使用一些云函数（如腾讯、阿里）来避免国内被墙的问题。
+# 详见：
+# https://github.com/Ice-Hazymoon/openai-scf-proxy
+# https://github.com/Soulter/QQChannelChatGPT/issues/42
+# 设置为none则表示使用官方默认api地址
+openai:
+  key: 
+    - 
+  api_base: none
+  # 这里是GPT配置，语言模型默认使用gpt-3.5-turbo
+  chatGPTConfigs:
+    model: gpt-3.5-turbo
+    max_tokens: 3000
+    temperature: 0.9
+    top_p: 1
+    frequency_penalty: 0
+    presence_penalty: 0
+    
+  total_tokens_limit: 5000
+
+# 逆向文心一言【暂时不可用，请勿使用】
+rev_ernie:
+  enable: false
+
+# 逆向New Bing
+# 需要在项目根目录下创建cookies.json并粘贴cookies进去。
+# 详见：https://soulter.top/posts/qpdg.html
+rev_edgegpt:
+  enable: false
+
+# 逆向ChatGPT库
+# https://github.com/acheong08/ChatGPT
+# 优点：免费（无免费额度限制）；
+# 缺点：速度相对慢。OpenAI 速率限制：免费帐户每小时 50 个请求。您可以通过多帐户循环来绕过它
+# enable设置为true后，将会停止使用上面正常的官方API调用而使用本逆向项目
+#
+# 多账户可以保证每个请求都能得到及时的回复。
+# 关于account的格式
+# account:
+#   - email: 第1个账户
+#     password: 第1个账户密码
+#   - email: 第2个账户
+#     password: 第2个账户密码
+#   - ....
+# 支持使用access_token登录
+# 例：
+# - session_token: xxxxx
+# - access_token: xxxx
+# 请严格按照上面这个格式填写。
+# 逆向ChatGPT库的email-password登录方式不工作，建议使用access_token登录
+# 获取access_token的方法，详见：https://soulter.top/posts/qpdg.html
+rev_ChatGPT:
+  enable: false
+  account:
+    - access_token: 
+```

addons/plugins/helloworld/README.md

@@ -0,0 +1,5 @@
+# helloworld
+
+QQChannelChatGPT项目的测试插件
+
+A test plugin for QQChannelChatGPT plugin feature

addons/plugins/helloworld/helloworld.py

@@ -0,0 +1,79 @@
+from nakuru.entities.components import *
+from nakuru import (
+    GroupMessage,
+    FriendMessage
+)
+from botpy.message import Message, DirectMessage
+from model.platform.qq import QQ
+import time
+import threading
+
+class HelloWorldPlugin:
+    """
+    初始化函数, 可以选择直接pass
+    """
+    def __init__(self) -> None:
+        self.myThread = None # 线程对象，如果要使用线程，需要在此处定义。在run处定义会被释放掉
+        print("这是HelloWorld测试插件, 发送 helloworld 即可触发此插件。")
+
+    """
+    入口函数，机器人会调用此函数。
+    参数规范: message: 消息文本; role: 身份; platform: 消息平台; message_obj: 消息对象; qq_platform: QQ平台对象，可以通过调用qq_platform.send()直接发送消息。详见Helloworld插件示例
+    参数详情: role为admin或者member; platform为qqchan或者gocq; message_obj为nakuru的GroupMessage对象或者FriendMessage对象或者频道的Message, DirectMessage对象。
+    返回规范: bool: 是否hit到此插件(所有的消息均会调用每一个载入的插件, 如果没有hit到, 则应返回False)
+             Tuple: None或者长度为3的元组。当没有hit到时, 返回None. hit到时, 第1个参数为指令是否调用成功, 第2个参数为返回的消息文本或者gocq的消息链列表, 第3个参数为指令名称
+    例子：做一个名为"yuanshen"的插件；当接收到消息为“原神 可莉”, 如果不想要处理此消息，则返回False, None；如果想要处理，但是执行失败了，返回True, tuple([False, "请求失败啦~", "yuanshen"])
+          ；执行成功了，返回True, tuple([True, "结果文本", "yuanshen"])
+    """
+    def run(self, message: str, role: str, platform: str, message_obj, qq_platform: QQ):
+
+        if platform == "gocq":
+            """
+            QQ平台指令处理逻辑
+            """
+            img_url = "https://gchat.qpic.cn/gchatpic_new/905617992/720871955-2246763964-C6EE1A52CC668EC982453065C4FA8747/0?term=2&is_origin=0"
+            if message == "helloworld":
+                return True, tuple([True, [Plain("Hello World!!"), Image.fromURL(url=img_url)], "helloworld"])
+            elif message == "hiloop":
+                if self.myThread is None:
+                    self.myThread = threading.Thread(target=self.helloworldThread, args=(message_obj, qq_platform))
+                    self.myThread.start()
+                return True, tuple([True, [Plain("A lot of Helloworlds!!"), Image.fromURL(url=img_url)], "helloworld"]) 
+            else:
+                return False, None
+        elif platform == "qqchan":
+            """
+            频道处理逻辑(频道暂时只支持回复字符串类型的信息，返回的信息都会被转成字符串，如果不想处理某一个平台的信息，直接返回False, None就行)
+            """
+            if message == "helloworld":
+                return True, tuple([True, "Hello World!!", "helloworld"])
+            else:
+                return False, None
+    """
+    帮助函数，当用户输入 plugin v 插件名称 时，会调用此函数，返回帮助信息
+    返回参数要求(必填)：dict{
+        "name": str, # 插件名称
+        "desc": str, # 插件简短描述
+        "help": str, # 插件帮助信息
+        "version": str, # 插件版本
+        "author": str, # 插件作者
+    }
+    """        
+    def info(self):
+        return {
+            "name": "helloworld",
+            "desc": "测试插件",
+            "help": "测试插件, 回复helloworld即可触发",
+            "version": "v1.0.1 beta",
+            "author": "Soulter"
+        }
+    
+    def helloworldThread(self, meseage_obj, qq_platform: QQ):
+        while True:
+            qq_platform.send(meseage_obj, [Plain("Hello World!!")]) # 第一个参数可以是message_obj, 也可以是qq群号
+            time.sleep(3) # 睡眠3秒。 用while True一定要记得sleep，不然会卡死
+        
+
+        # 热知识：检测消息开头指令，使用以下方法
+        # if message.startswith("原神"):
+        #     pass

cores/qqbot/global_object.py

@@ -0,0 +1,16 @@
+class GlobalObject():
+    '''
+    存放一些公用的数据，用于在不同模块(如core与command)之间传递
+    '''
+    def __init__(self):
+        self.nick = None # gocq 的昵称
+        self.base_config = None # config.yaml
+        self.cached_plugins = {} # 缓存的插件
+        self.web_search = False # 是否开启了网页搜索
+        self.reply_prefix = None
+        self.admin_qq = "123456"
+        self.admin_qqchan = "123456"
+        self.uniqueSession = False
+        self.cnt_total = 0
+        self.platform_qq = None
+        self.platform_qqchan = None

launcher.py

@@ -1,56 +0,0 @@
-# -*- coding: utf-8 -*-
-
-from git.repo import Repo
-import git
-import os
-# import zipfile
-
-if __name__ == "__main__":
-    try:
-        # 检测文件夹
-        if not os.path.exists('QQChannelChatGPT'):
-            os.mkdir('QQChannelChatGPT')
-            
-        project_path = os.path.join('QQChannelChatGPT')
-        try:
-            repo = Repo(project_path)
-            # 检查当前commit的hash值
-            commit_hash = repo.head.object.hexsha
-            print("当前commit的hash值为: " + commit_hash)
-
-            # 得到远程仓库的origin的commit的列表
-            origin = repo.remotes.origin
-            try:
-                origin.fetch()
-            except:
-                pass
-            # 得到远程仓库的commit的hash值
-            remote_commit_hash = origin.refs.master.commit.hexsha
-            print("https://github.com/Soulter/QQChannelChatGPT的commit的hash值为: " + remote_commit_hash)
-            # 比较两个commit的hash值
-            if commit_hash != remote_commit_hash:
-                res = input("检测到项目有更新, 是否更新? (y/n): ")
-                if res == 'y':
-                    repo.remotes.origin.pull()
-                    print("项目更新完毕")
-                if res == 'n':
-                    print("已取消更新")
-        except:
-            print("正在从https://github.com/Soulter/QQChannelChatGPT.git拉取项目...")
-            Repo.clone_from('https://github.com/Soulter/QQChannelChatGPT.git',to_path=project_path,branch='master')
-            print("项目拉取完毕")
-            print("【重要提醒】如果你没有Python（版本>=3.8）或者Git环境, 请先安装, 否则接下来的操作会造成闪退。")
-            print("【重要提醒】Python下载地址: https://npm.taobao.org/mirrors/python/3.9.7/python-3.9.7-amd64.exe ")
-            print("【重要提醒】Git下载地址: https://registry.npmmirror.com/-/binary/git-for-windows/v2.39.2.windows.1/Git-2.39.2-64-bit.exe")
-            print("【重要提醒】安装时, 请务必勾选“Add Python to PATH”选项。")
-            input("已确保安装了Python3.9+的版本，按下回车继续...")
-            print("正在安装依赖库")
-            os.system('python -m pip install -r QQChannelChatGPT\\requirements.txt')
-            print("依赖库安装完毕")
-            input("初次启动, 请先在QQChannelChatGPT/configs/config.yaml填写相关配置! 按任意键继续...")
-        finally:
-            print("正在启动项目...")
-            os.system('python QQChannelChatGPT\main.py')
-    except BaseException as e:
-        print(e)
-        input("程序出错。可以截图发给QQ：905617992.按下回车键退出...")

main.py

@@ -1,27 +1,53 @@
-import threading
-import asyncio
 import os, sys
+from pip._internal import main as pipmain
+import warnings
+import traceback
 
+warnings.filterwarnings("ignore")
 abs_path = os.path.dirname(os.path.realpath(sys.argv[0])) + '/'
 
-def main(loop, event):
-    import cores.qqbot.core as qqBot
-    import yaml
-    ymlfile =  open(abs_path+"configs/config.yaml", 'r', encoding='utf-8')
-    cfg = yaml.safe_load(ymlfile)
+def main():
 
+    # config.yaml 配置文件加载和环境确认
+    try:
+        import cores.qqbot.core as qqBot
+        import yaml
+        from yaml.scanner import ScannerError
+        import util.general_utils as gu
+        ymlfile =  open(abs_path+"configs/config.yaml", 'r', encoding='utf-8')
+        cfg = yaml.safe_load(ymlfile)
+    except ImportError as import_error:
+        print(import_error)
+        input("第三方库未完全安装完毕，请退出程序重试。")
+    except FileNotFoundError as file_not_found:
+        print(file_not_found)
+        input("配置文件不存在，请检查是否已经下载配置文件。")
+    except ScannerError as e:
+        print(traceback.format_exc())
+        input("config.yaml 配置文件格式错误，请遵守 yaml 格式。")
+
+    # 设置代理
     if 'http_proxy' in cfg:
         os.environ['HTTP_PROXY'] = cfg['http_proxy']
     if 'https_proxy' in cfg:
         os.environ['HTTPS_PROXY'] = cfg['https_proxy']
+    
+    os.environ['NO_PROXY'] = 'cn.bing.com,https://api.sgroup.qq.com'
 
+    # 检查并创建 temp 文件夹
+    if not os.path.exists(abs_path + "temp"):
+        os.mkdir(abs_path+"temp")
+
+    # 选择默认模型
     provider = privider_chooser(cfg)
-    print('[System] 当前语言模型提供商: ' + str(provider))
-    # 执行Bot
+    if len(provider) == 0:
+        gu.log("注意：您目前未开启任何语言模型。", gu.LEVEL_WARNING)
+    print('[System] 开启的语言模型: ' + str(provider))
+
+    # 启动主程序（cores/qqbot/core.py）
     qqBot.initBot(cfg, provider)
 
 # 语言模型提供商选择器
-# 目前有：OpenAI官方API、逆向库
 def privider_chooser(cfg):
     l = []
     if 'rev_ChatGPT' in cfg and cfg['rev_ChatGPT']['enable']:
@@ -34,33 +60,44 @@ def privider_chooser(cfg):
         l.append('openai_official')
     return l
 
-def check_env():
-    if not (sys.version_info.major == 3 and sys.version_info.minor >= 8):
-        print("请使用Python3.8运行本项目")
-        input("按任意键退出...")
-        exit()
-    try:
-        print("检查依赖库中...")
-        if os.path.exists('requirements.txt'):
-            os.system("pip3 install -r requirements.txt")
-        elif os.path.exists('QQChannelChatGPT'+ os.sep +'requirements.txt'):
-            os.system('QQChannelChatGPT'+ os.sep +'requirements.txt')
-        os.system("clear")
-        print("安装依赖库完毕...")
-    except BaseException as e:
-        print("安装依赖库失败，请手动安装依赖库。")
-        print(e)
+def check_env(ch_mirror=False):
+    if not (sys.version_info.major == 3 and sys.version_info.minor >= 9):
+        print("请使用Python3.9+运行本项目")
         input("按任意键退出...")
         exit()
     
-    # 检查key
-    with open(abs_path+"configs/config.yaml", 'r', encoding='utf-8') as ymlfile:
-        import yaml
-        cfg = yaml.safe_load(ymlfile)
-        if cfg['openai']['key'] == '' or cfg['openai']['key'] == None:
-            print("请先在configs/config.yaml下添加一个可用的OpenAI Key。详情请前往https://beta.openai.com/account/api-keys")
-        if cfg['qqbot']['appid'] == '' or cfg['qqbot']['token'] == '' or cfg['qqbot']['appid'] == None or cfg['qqbot']['token'] == None: 
-            print("请先在configs/config.yaml下完善appid和token令牌(在https://q.qq.com/上注册一个QQ机器人即可获得)")
+    if os.path.exists('requirements.txt'):
+        pth = 'requirements.txt'
+    else:
+        pth = 'QQChannelChatGPT'+ os.sep +'requirements.txt'
+    print("正在检查更新第三方库...")
+    try:
+        if ch_mirror:
+            print("使用阿里云镜像")
+            pipmain(['install', '-r', pth, '-i', 'https://mirrors.aliyun.com/pypi/simple/', '--quiet'])
+        else:
+            pipmain(['install', '-r', pth, '--quiet'])
+    except BaseException as e:
+        print(e)
+        while True:
+            res = input("安装失败。\n如报错ValueError: check_hostname requires server_hostname，请尝试先关闭代理后重试。\n1.输入y回车重试\n2. 输入c回车使用国内镜像源下载\n3. 输入其他按键回车继续往下执行。")
+            if res == "y":
+                try:
+                    pipmain(['install', '-r', pth])
+                    break
+                except BaseException as e:
+                    print(e)
+                    continue
+            elif res == "c":
+                try:
+                    pipmain(['install', '-r', pth, '-i', 'https://mirrors.aliyun.com/pypi/simple/'])
+                    break
+                except BaseException as e:
+                    print(e)
+                    continue
+            else:
+                break
+    print("第三方库检查完毕。")
 
 def get_platform():
     import platform
@@ -75,7 +112,20 @@ def get_platform():
         print("other")
 
 if __name__ == "__main__":
-    check_env()
-    bot_event = threading.Event()
-    loop = asyncio.get_event_loop()
-    main(loop, bot_event)
+
+    args = sys.argv
+
+    if '-cn' in args:
+        check_env(True)
+    else:
+        check_env()
+
+    if '-replit' in args:
+        print("[System] 启动Replit Web保活服务...")
+        try:
+            from webapp_replit import keep_alive
+            keep_alive()
+        except BaseException as e:
+            print(e)
+            print(f"[System-err] Replit Web保活服务启动失败:{str(e)}")
+    main()

model/command/command.py

@@ -1,22 +1,357 @@
-import abc
 import json
-import git.exc
-from git.repo import Repo
+
+has_git = True
+try:
+    import git.exc
+    from git.repo import Repo
+except BaseException as e:
+    print("你正运行在无Git环境下，暂时将无法使用插件、热更新功能。")
+    has_git = False
+
 import os
 import sys
-
+import requests
 from model.provider.provider import Provider
+import json
+import util.plugin_util as putil
+import shutil
+import importlib
+from util import general_utils as gu
+from util.cmd_config import CmdConfig as cc
+from model.platform.qq import QQ
+import stat
+from nakuru.entities.components import (
+    Plain,
+    Image
+)
+from PIL import Image as PILImage
+from cores.qqbot.global_object import GlobalObject
+from pip._internal import main as pipmain
 
+PLATFORM_QQCHAN = 'qqchan'
+PLATFORM_GOCQ = 'gocq'
+
+# 指令功能的基类，通用的（不区分语言模型）的指令就在这实现
 class Command:
-    def __init__(self, provider: Provider):
-        self.provider = Provider
+    def __init__(self, provider: Provider, global_object: GlobalObject = None):
+        self.provider = provider
+        self.global_object = global_object
 
-    @abc.abstractmethod
-    def check_command(self, message):
-        if message.startswith("help") or message.startswith("帮助"):
-            return True, self.help()
+    def check_command(self, 
+                      message, 
+                      session_id: str,
+                      role, 
+                      platform, 
+                      message_obj):
+        # 插件
+        cached_plugins = self.global_object.cached_plugins
+        for k, v in cached_plugins.items():
+            try:
+                hit, res = v["clsobj"].run(message, role, platform, message_obj, self.global_object.platform_qq)
+                if hit:
+                    return True, res
+            except BaseException as e:
+                gu.log(f"{k}插件加载出现问题，原因: {str(e)}\n已安装插件: {cached_plugins.keys}\n如果你没有相关装插件的想法, 请直接忽略此报错, 不影响其他功能的运行。", level=gu.LEVEL_WARNING)
+
+        if self.command_start_with(message, "nick"):
+            return True, self.set_nick(message, platform, role)
+        
+        if self.command_start_with(message, "plugin"):
+            return True, self.plugin_oper(message, role, cached_plugins, platform)
+        
+        if self.command_start_with(message, "myid") or self.command_start_with(message, "!myid"):
+            return True, self.get_my_id(message_obj)
+        if self.command_start_with(message, "nconf") or self.command_start_with(message, "newconf"):
+            return True, self.get_new_conf(message, role)
+        if self.command_start_with(message, "web"): # 网页搜索
+            return True, self.web_search(message)
+        if self.command_start_with(message, "keyword"):
+            return True, self.keyword(message_obj, role)
+        
         return False, None
     
+    def web_search(self, message):
+        if message == "web on":
+            self.global_object.web_search = True
+            return True, "已开启网页搜索", "web"
+        elif message == "web off":
+            self.global_object.web_search = False
+            return True, "已关闭网页搜索", "web"
+        return True, f"网页搜索功能当前状态: {self.global_object.web_search}", "web"
+
+    def get_my_id(self, message_obj):
+        return True, f"你的ID：{str(message_obj.sender.tiny_id)}", "plugin"
+            
+    def get_new_conf(self, message, role):
+        if role != "admin":
+            return False, f"你的身份组{role}没有权限使用此指令。", "newconf"
+        l = message.split(" ")
+        if len(l) <= 1:
+            obj = cc.get_all()
+            p = gu.create_text_image("【cmd_config.json】", json.dumps(obj, indent=4, ensure_ascii=False))
+            return True, [Image.fromFileSystem(p)], "newconf"
+            
+    def get_plugin_modules(self):
+        plugins = []
+        try:
+            if os.path.exists("addons/plugins"):
+                plugins = putil.get_modules("addons/plugins")
+                return plugins
+            elif os.path.exists("QQChannelChatGPT/addons/plugins"):
+                plugins = putil.get_modules("QQChannelChatGPT/addons/plugins")
+                return plugins
+            else:
+                return None
+        except BaseException as e:
+            raise e
+
+    def plugin_reload(self, cached_plugins: dict, target: str = None, all: bool = False):
+        plugins = self.get_plugin_modules()
+        fail_rec = ""
+        if plugins is None:
+            return False, "未找到任何插件模块"
+        
+        print(plugins)
+
+        for plugin in plugins:
+            try:
+                p = plugin['module']
+                root_dir_name = plugin['pname']
+                if p not in cached_plugins or p == target or all:
+                    module = __import__("addons.plugins." + root_dir_name + "." + p, fromlist=[p])
+                    if p in cached_plugins:
+                        module = importlib.reload(module)
+                    cls = putil.get_classes(p, module)
+                    obj = getattr(module, cls[0])()
+                    try:
+                        info = obj.info()
+                        if 'name' not in info or 'desc' not in info or 'version' not in info or 'author' not in info:
+                            fail_rec += f"载入插件{p}失败，原因: 插件信息不完整\n"
+                            continue
+                        if isinstance(info, dict) == False:
+                            fail_rec += f"载入插件{p}失败，原因: 插件信息格式不正确\n"
+                            continue
+                    except BaseException as e:
+                        fail_rec += f"调用插件{p} info失败, 原因: {str(e)}\n"
+                        continue
+                    cached_plugins[info['name']] = {
+                        "module": module,
+                        "clsobj": obj,
+                        "info": info,
+                        "name": info['name'],
+                        "root_dir_name": root_dir_name,
+                    }
+            except BaseException as e:
+                fail_rec += f"加载{p}插件出现问题，原因 {str(e)}\n"
+        if fail_rec == "":
+            return True, None
+        else:
+            return False, fail_rec
+    
+    '''
+    插件指令
+    '''
+    def plugin_oper(self, message: str, role: str, cached_plugins: dict, platform: str):
+        if not has_git:
+            return False, "你正在运行在无Git环境下，暂时将无法使用插件、热更新功能。", "plugin"
+        l = message.split(" ")
+        if len(l) < 2:
+            p = gu.create_text_image("【插件指令面板】", "安装插件: \nplugin i 插件Github地址\n卸载插件: \nplugin d 插件名 \n重载插件: \nplugin reload\n查看插件列表：\nplugin l\n更新插件: plugin u 插件名\n")
+            return True, [Image.fromFileSystem(p)], "plugin"
+        else:
+            ppath = ""
+            if os.path.exists("addons/plugins"):
+                ppath = "addons/plugins"
+            elif os.path.exists("QQChannelChatGPT/addons/plugins"):
+                ppath = "QQChannelChatGPT/addons/plugins"
+            else:
+                return False, "未找到插件目录", "plugin"
+            if l[1] == "i":
+                if role != "admin":
+                    return False, f"你的身份组{role}没有权限安装插件", "plugin"
+                try:
+                    # 删除末尾的/
+                    if l[2].endswith("/"):
+                        l[2] = l[2][:-1]
+                    # 得到url的最后一段
+                    d = l[2].split("/")[-1]
+                    # 转换非法字符：-
+                    d = d.replace("-", "_")
+                    # 创建文件夹
+                    plugin_path = os.path.join(ppath, d)
+                    if os.path.exists(plugin_path):
+                        shutil.rmtree(plugin_path)
+                    os.mkdir(plugin_path)
+                    Repo.clone_from(l[2],to_path=plugin_path,branch='master')
+
+                    # 读取插件的requirements.txt
+                    if os.path.exists(os.path.join(plugin_path, "requirements.txt")):
+                        mm = pipmain(['install', '-r', os.path.join(plugin_path, "requirements.txt")])
+                        if mm != 0:
+                            return False, "插件依赖安装失败，需要您手动pip安装对应插件的依赖。", "plugin"
+                    # 加载没缓存的插件
+                    ok, err = self.plugin_reload(cached_plugins, target=d)
+                    if ok:
+                        return True, "插件拉取并载入成功~", "plugin"
+                    else:
+                        # if os.path.exists(plugin_path):
+                        #     shutil.rmtree(plugin_path)
+                        return False, f"插件拉取载入失败。\n跟踪: \n{err}", "plugin"
+                except BaseException as e:
+                    return False, f"拉取插件失败，原因: {str(e)}", "plugin"
+            elif l[1] == "d":
+                if role != "admin":
+                    return False, f"你的身份组{role}没有权限删除插件", "plugin"
+                if l[2] not in cached_plugins:
+                    return False, "未找到该插件", "plugin"
+
+                try:
+                    root_dir_name = cached_plugins[l[2]]["root_dir_name"]
+                    self.remove_dir(os.path.join(ppath, root_dir_name))
+                    del cached_plugins[l[2]]
+                    return True, "插件卸载成功~", "plugin"
+                except BaseException as e:
+                    return False, f"卸载插件失败，原因: {str(e)}", "plugin"
+            elif l[1] == "u":
+                if l[2] not in cached_plugins:
+                    return False, "未找到该插件", "plugin"
+                root_dir_name = cached_plugins[l[2]]["root_dir_name"]
+                plugin_path = os.path.join(ppath, root_dir_name)
+                try:
+                    repo = Repo(path = plugin_path)
+                    repo.remotes.origin.pull()
+                    # 读取插件的requirements.txt
+                    if os.path.exists(os.path.join(plugin_path, "requirements.txt")):
+                        mm = pipmain(['install', '-r', os.path.join(plugin_path, "requirements.txt")])
+                        if mm != 0:
+                            return False, "插件依赖安装失败，需要您手动pip安装对应插件的依赖。", "plugin"
+
+                    ok, err = self.plugin_reload(cached_plugins, target=l[2])
+                    if ok:
+                        return True, "\n更新插件成功!!", "plugin"
+                    else:
+                        return False, "更新插件成功，但是重载插件失败。\n问题跟踪: \n"+err, "plugin"
+                except BaseException as e:
+                    return False, "更新插件失败, 请使用plugin i指令覆盖安装", "plugin"
+
+            elif l[1] == "l":
+                try:
+                    plugin_list_info = "\n".join([f"{k}: \n名称: {v['info']['name']}\n简介: {v['info']['desc']}\n版本: {v['info']['version']}\n作者: {v['info']['author']}\n" for k, v in cached_plugins.items()])
+                    p = gu.create_text_image("【已激活插件列表】", plugin_list_info + "\n使用plugin v 插件名 查看插件帮助\n")
+                    return True, [Image.fromFileSystem(p)], "plugin"
+                except BaseException as e:
+                    return False, f"获取插件列表失败，原因: {str(e)}", "plugin"
+            elif l[1] == "v":
+                try:
+                    if l[2] in cached_plugins:
+                        info = cached_plugins[l[2]]["info"]
+                        p = gu.create_text_image(f"【插件信息】", f"名称: {info['name']}\n{info['desc']}\n版本: {info['version']}\n作者: {info['author']}\n\n帮助:\n{info['help']}")
+                        return True, [Image.fromFileSystem(p)], "plugin"
+                    else:
+                        return False, "未找到该插件", "plugin"
+                except BaseException as e:
+                    return False, f"获取插件信息失败，原因: {str(e)}", "plugin"
+            elif l[1] == "reload":
+                if role != "admin":
+                    return False, f"你的身份组{role}没有权限重载插件", "plugin"
+                for plugin in cached_plugins:
+                    try:
+                        print(f"更新插件 {plugin} 依赖...")
+                        plugin_path = os.path.join(ppath, cached_plugins[plugin]["root_dir_name"])
+                        if os.path.exists(os.path.join(plugin_path, "requirements.txt")):
+                            mm = pipmain(['install', '-r', os.path.join(plugin_path, "requirements.txt"), "--quiet"])
+                            if mm != 0:
+                                return False, "插件依赖安装失败，需要您手动pip安装对应插件的依赖。", "plugin"
+                    except BaseException as e:
+                        print(f"插件{plugin}依赖安装失败，原因: {str(e)}")
+                try:
+                    ok, err = self.plugin_reload(cached_plugins, all = True)
+                    if ok:
+                        return True, "\n重载插件成功~", "plugin"
+                    else:
+                        # if os.path.exists(plugin_path):
+                        #     shutil.rmtree(plugin_path)
+                        return False, f"插件重载失败。\n跟踪: \n{err}", "plugin"
+                except BaseException as e:
+                    return False, f"插件重载失败，原因: {str(e)}", "plugin"
+                
+            elif l[1] == "dev":
+                if role != "admin":
+                    return False, f"你的身份组{role}没有权限开发者模式", "plugin"
+                return True, "cached_plugins: \n" + str(cached_plugins), "plugin"
+    
+    def remove_dir(self, file_path):        
+        while 1:
+            if not os.path.exists(file_path):
+                break
+            try:
+                shutil.rmtree(file_path)
+            except PermissionError as e:
+                err_file_path = str(e).split("\'", 2)[1]
+                if os.path.exists(err_file_path):
+                    os.chmod(err_file_path, stat.S_IWUSR)
+
+    '''
+    nick: 存储机器人的昵称
+    '''
+    def set_nick(self, message: str, platform: str, role: str = "member"):
+        if role != "admin":
+            return True, "你无权使用该指令 :P", "nick"
+        if platform == PLATFORM_GOCQ:
+            l = message.split(" ")
+            if len(l) == 1:
+                return True, "【设置机器人昵称】示例：\n支持多昵称\nnick 昵称1 昵称2 昵称3", "nick"
+            nick = l[1:]
+            cc.put("nick_qq", nick)
+            self.global_object.nick = tuple(nick)
+            return True, f"设置成功！现在你可以叫我这些昵称来提问我啦~", "nick"
+        elif platform == PLATFORM_QQCHAN:
+            nick = message.split(" ")[2]
+            return False, "QQ频道平台不支持为机器人设置昵称。", "nick"
+
+    def general_commands(self):
+        return {
+            "help": "帮助",
+            "keyword": "设置关键词/关键指令回复",
+            "update": "更新面板",
+            "update latest": "更新到最新版本",
+            "update r": "重启机器人",
+            "reset": "重置会话",
+            "nick": "设置机器人昵称",
+            "plugin": "插件安装、卸载和重载",
+            "web on/off": "启动或关闭网页搜索能力",
+            "/bing": "切换到bing模型",
+            "/gpt": "切换到OpenAI ChatGPT API",
+            "/revgpt": "切换到网页版ChatGPT",
+        }
+    
+    def help_messager(self, commands: dict, platform: str, cached_plugins: dict = None):
+        try:
+            resp = requests.get("https://soulter.top/channelbot/notice.json").text
+            notice = json.loads(resp)["notice"]
+        except BaseException as e:
+            notice = ""
+        msg = "# Help Center\n## 指令列表\n"
+        # msg = "Github项目名QQChannelChatGPT, 有问题提交issue, 欢迎Star\n【指令列表】\n"
+        for key, value in commands.items():
+            msg += f"`{key}` - {value}\n"
+        # plugins
+        if cached_plugins != None:
+            plugin_list_info = "\n".join([f"`{k}` {v['info']['name']}\n{v['info']['desc']}\n" for k, v in cached_plugins.items()])
+            if plugin_list_info.strip() != "":
+                msg += "\n## 插件列表\n> 使用plugin v 插件名 查看插件帮助\n"
+                msg += plugin_list_info
+        msg += notice
+
+        try:
+            # p = gu.create_text_image("【Help Center】", msg)
+            p = gu.create_markdown_image(msg)
+            return [Image.fromFileSystem(p)]
+        except BaseException as e:
+            gu.log(str(e))
+        finally:
+            return msg
+    
     # 接受可变参数
     def command_start_with(self, message: str, *args):
         for arg in args:
@@ -24,27 +359,80 @@ class Command:
                 return True
         return False
     
-    def keyword(self, message: str, role: str):
+    # keyword: 关键字
+    def keyword(self, message_obj, role: str):
         if role != "admin":
             return True, "你没有权限使用该指令", "keyword"
-        if len(message.split(" ")) != 3:
-            return True, "【设置关键词/关键指令回复】示例：\nkeyword hi 你好\n当发送hi的时候会回复你好\nkeyword /hi 你好\n当发送/hi时会回复你好", "keyword"
         
-        l = message.split(" ")
+        plain_text = ""
+        image_url = ""
+
+        for comp in message_obj.message:
+            if isinstance(comp, Plain):
+                plain_text += comp.text
+            elif isinstance(comp, Image) and image_url == "":
+                if comp.url is None:
+                    image_url = comp.file
+                else:
+                    image_url = comp.url
+
+        l = plain_text.split(" ")
+
+        if len(l) < 3 and image_url == "":
+            return True, """
+【设置关键词回复】示例：
+1. keyword hi 你好
+当发送hi的时候会回复你好
+2. keyword /hi 你好
+当发送/hi时会回复你好
+3. keyword d hi
+删除hi关键词的回复
+4. keyword hi <图片>
+当发送hi时会回复图片
+""", "keyword"
+        
+        del_mode = False
+        if l[1] == "d":
+            del_mode = True    
+
         try:
             if os.path.exists("keyword.json"):
                 with open("keyword.json", "r", encoding="utf-8") as f:
                     keyword = json.load(f)
-                    keyword[l[1]] = l[2]
+                    if del_mode:
+                        # 删除关键词
+                        if l[2] not in keyword:
+                            return False, "该关键词不存在", "keyword"
+                        else: del keyword[l[2]]
+                    else:
+                        keyword[l[1]] = {
+                            "plain_text": " ".join(l[2:]),
+                            "image_url": image_url
+                        }
             else:
-                keyword = {l[1]: l[2]}
+                if del_mode:
+                    return False, "该关键词不存在", "keyword"
+                keyword = {
+                    l[1]: {
+                    "plain_text": " ".join(l[2:]),
+                    "image_url": image_url
+                    }
+                }
             with open("keyword.json", "w", encoding="utf-8") as f:
                 json.dump(keyword, f, ensure_ascii=False, indent=4)
-            return True, "设置成功: "+l[1]+" -> "+l[2], "keyword"
+                f.flush()
+            if del_mode:
+                return True, "删除成功: "+l[2], "keyword"
+            if image_url == "":
+                return True, "设置成功: "+l[1]+" "+" ".join(l[2:]), "keyword"
+            else:
+                return True, [Plain("设置成功: "+l[1]+" "+" ".join(l[2:])), Image.fromURL(image_url)], "keyword"
         except BaseException as e:
             return False, "设置失败: "+str(e), "keyword"
     
     def update(self, message: str, role: str):
+        if not has_git:
+            return False, "你正在运行在无Git环境下，暂时将无法使用插件、热更新功能。", "update"
         if role != "admin":
             return True, "你没有权限使用该指令", "keyword"
         l = message.split(" ")
@@ -63,7 +451,10 @@ class Command:
             commits_log = ''
             index = 1
             for commit in commits:
-                commits_log += f"[{index}] {commit.message}\n-----------\n"
+                if commit.message.endswith("\n"):
+                    commits_log += f"[{index}] {commit.message}-----------\n"
+                else:
+                    commits_log += f"[{index}] {commit.message}\n-----------\n"
                 index+=1
             remote_commit_hash = origin.refs.master.commit.hexsha[:6]
 
@@ -80,28 +471,24 @@ class Command:
                     repo.remotes.origin.pull()
 
                     try:
-                        os.system("pip3 install -r "+pash_tag+"requirements.txt")
+                        origin = repo.remotes.origin
+                        origin.fetch()
+                        commits = list(repo.iter_commits('master', max_count=1))
+                        commit_log = commits[0].message
                     except BaseException as e:
-                        print(str(e))
+                        commit_log = "无法获取commit信息"
 
-                    py = sys.executable
-                    os.execl(py, py, *sys.argv)
+                    tag = "update"
+                    if len(l) == 3 and l[2] == "r":
+                        tag = "update latest r"
 
-                    # 检查是否是windows环境
-                    # if platform.system().lower() == "windows":
-                    #     if os.path.exists("launcher.exe"):
-                    #         os.system("start launcher.exe")
-                    #     elif os.path.exists("QQChannelChatGPT\\main.py"):
-                    #         os.system("start python QQChannelChatGPT\\main.py")
-                    #     else:
-                    #         return True, "更新成功，未发现启动项，因此需要手动重启程序。"
-                    #     exit()
-                    # else:
-                    #     py = sys.executable
-                    #     os.execl(py, py, *sys.argv)
+                    return True, f"更新成功。新版本内容: \n{commit_log}\nps:重启后生效。输入update r重启（重启指令不返回任何确认信息）。", tag
                     
                 except BaseException as e:
                     return False, "更新失败: "+str(e), "update"
+            if l[1] == "r":
+                py = sys.executable
+                os.execl(py, py, *sys.argv)
 
     def reset(self):
         return False
@@ -116,7 +503,7 @@ class Command:
         return False
     
     def help(self):
-        return True, f"[Github项目名: QQChannelChatGPT，有问题请前往提交issue，欢迎Star此项目~]\n\n指令面板：\nstatus 查看机器人key状态\ncount 查看机器人统计信息\nreset 重置会话\nhis 查看历史记录\ntoken 查看会话token数\nhelp 查看帮助\nset 人格指令菜单\nkey 动态添加key", "help"
+        return False
     
     def status(self):
         return False

model/command/command_openai_official.py

@@ -2,15 +2,39 @@ from model.command.command import Command
 from model.provider.provider_openai_official import ProviderOpenAIOfficial
 from cores.qqbot.personality import personalities
 
+from model.platform.qq import QQ
+from util import general_utils as gu
+from cores.qqbot.global_object import GlobalObject
+
 class CommandOpenAIOfficial(Command):
-    def __init__(self, provider: ProviderOpenAIOfficial):
+    def __init__(self, provider: ProviderOpenAIOfficial, global_object: GlobalObject):
         self.provider = provider
+        self.cached_plugins = {}
+        self.global_object = global_object
+        self.personality_str = ""
+        super().__init__(provider, global_object)
         
-    def check_command(self, message: str, session_id: str, user_name: str, role):
+    def check_command(self, 
+                      message: str, 
+                      session_id: str, 
+                      role: str, 
+                      platform: str,
+                      message_obj):
+        self.platform = platform
+        hit, res = super().check_command(
+            message,
+            session_id,
+            role,
+            platform,
+            message_obj
+        )
+        
+        if hit:
+            return True, res
         if self.command_start_with(message, "reset", "重置"):
-            return True, self.reset(session_id)
+            return True, self.reset(session_id, message)
         elif self.command_start_with(message, "his", "历史"):
-            return True, self.his(message, session_id, user_name)
+            return True, self.his(message, session_id)
         elif self.command_start_with(message, "token"):
             return True, self.token(session_id)
         elif self.command_start_with(message, "gpt"):
@@ -27,21 +51,42 @@ class CommandOpenAIOfficial(Command):
             return True, self.set(message, session_id)
         elif self.command_start_with(message, "update"):
             return True, self.update(message, role)
-        elif self.command_start_with(message, "画"):
+        elif self.command_start_with(message, "画", "draw"):
             return True, self.draw(message)
-        elif self.command_start_with(message, "keyword"):
-            return True, self.keyword(message, role)
         elif self.command_start_with(message, "key"):
-            return True, self.key(message, user_name)
+            return True, self.key(message)
+        elif self.command_start_with(message, "switch"):
+            return True, self.switch(message)
         
         return False, None
+    
+    def help(self):
+        commands = super().general_commands()
+        commands['画'] = '画画'
+        commands['key'] = '添加OpenAI key'
+        commands['set'] = '人格设置面板'
+        commands['gpt'] = '查看gpt配置信息'
+        commands['status'] = '查看key使用状态'
+        commands['token'] = '查看本轮会话token'
+        return True, super().help_messager(commands, self.platform, self.global_object.cached_plugins), "help"
 
+        
+    def reset(self, session_id: str, message: str = "reset"):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "reset"
+        l = message.split(" ")
+        if len(l) == 1:
+            self.provider.forget(session_id)
+            return True, "重置成功", "reset"
+        if len(l) == 2 and l[1] == "p":
+            self.provider.forget(session_id)
+            if self.personality_str != "":
+                self.set(self.personality_str, session_id) # 重新设置人格
+            return True, "重置成功", "reset"
     
-    def reset(self, session_id: str):
-        self.provider.forget(session_id)
-        return True, "重置成功", "reset"
-    
-    def his(self, message: str, session_id: str, name: str):
+    def his(self, message: str, session_id: str):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "his"
         #分页，每页5条
         msg = ''
         size_per_page = 3
@@ -58,12 +103,18 @@ class CommandOpenAIOfficial(Command):
         return True, f"历史记录如下：\n{p}\n第{page}页 | 共{max_page}页\n*输入/his 2跳转到第2页", "his"
     
     def token(self, session_id: str):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "token"
         return True, f"会话的token数: {self.provider.get_user_usage_tokens(self.provider.session_dict[session_id])}\n系统最大缓存token数: {self.provider.max_tokens}", "token"
 
     def gpt(self):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "gpt"
         return True, f"OpenAI GPT配置:\n {self.provider.chatGPT_configs}", "gpt"
     
     def status(self):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "status"
         chatgpt_cfg_str = ""
         key_stat = self.provider.get_key_stat()
         index = 1
@@ -79,37 +130,75 @@ class CommandOpenAIOfficial(Command):
                 continue
             if 'sponsor' in key_stat[key]:
                 sponsor = key_stat[key]['sponsor']
-            chatgpt_cfg_str += f"  |-{index}: {key_stat[key]['used']}/{max} {sponsor}赞助{tag}\n"
+            chatgpt_cfg_str += f"  |-{index}: {key[-8:]} {key_stat[key]['used']}/{max} {sponsor}{tag}\n"
             index += 1
-        return True, f"⭐使用情况({str(gg_count)}个已用):\n{chatgpt_cfg_str}⏰全频道已用{total}tokens", "status"
+        return True, f"⭐使用情况({str(gg_count)}个已用):\n{chatgpt_cfg_str}", "status"
 
     def count(self):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型。", "reset"
         guild_count, guild_msg_count, guild_direct_msg_count, session_count = self.provider.get_stat()
-        return True, f"当前会话数: {len(self.provider.session_dict)}\n共有频道数: {guild_count} \n共有消息数: {guild_msg_count}\n私信数: {guild_direct_msg_count}\n历史会话数: {session_count}", "count"
+        return True, f"【本指令部分统计可能已经过时】\n当前会话数: {len(self.provider.session_dict)}\n共有频道数: {guild_count} \n共有消息数: {guild_msg_count}\n私信数: {guild_direct_msg_count}\n历史会话数: {session_count}", "count"
 
-    def key(self, message: str, user_name: str):
+    def key(self, message: str):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "reset"
         l = message.split(" ")
         if len(l) == 1:
             msg = "感谢您赞助key，key为官方API使用，请以以下格式赞助:\n/key xxxxx"
             return True, msg, "key"
         key = l[1]
         if self.provider.check_key(key):
-            self.provider.append_key(key, user_name)
-            return True, f"*★,°*:.☆(￣▽￣)/$:*.°★* 。\n该Key被验证为有效。感谢{user_name}赞助~"
+            self.provider.append_key(key)
+            return True, f"*★,°*:.☆(￣▽￣)/$:*.°★* 。\n该Key被验证为有效。感谢你的赞助~"
         else:
             return True, "该Key被验证为无效。也许是输入错误了，或者重试。", "key"
 
+    def switch(self, message: str):
+        '''
+        切换账号
+        '''
+        l = message.split(" ")
+        if len(l) == 1:
+            _, ret, _ = self.status()
+            curr_ = self.provider.get_curr_key()
+            if curr_ is None:
+                ret += "当前您未选择账号。输入/switch <账号序号>切换账号。"
+            else:
+                ret += f"当前您选择的账号为：{curr_[-8:]}。输入/switch <账号序号>切换账号。"
+            return True, ret, "switch"
+        elif len(l) == 2:
+            try:
+                key_stat = self.provider.get_key_stat()
+                index = int(l[1])
+                if index > len(key_stat) or index < 1:
+                    return True, "账号序号不合法。", "switch"
+                else:
+                    ret = self.provider.check_key(list(key_stat.keys())[index-1])
+                    if ret:
+                        return True, f"账号切换成功。", "switch"
+                    else:
+                        return True, f"账号切换失败，可能超额或超频。", "switch"
+            except BaseException as e:
+                return True, "未知错误: "+str(e), "switch"
+        else:
+            return True, "参数过多。", "switch"
+
     def unset(self, session_id: str):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "unset"
         self.provider.now_personality = {}
         self.provider.forget(session_id)
         return True, "已清除人格并重置历史记录。", "unset"
 
     def set(self, message: str, session_id: str):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "set"
         l = message.split(" ")
         if len(l) == 1:
-            return True, f"【由Github项目QQChannelChatGPT支持】\n\n【人格文本由PlexPt开源项目awesome-chatgpt-pr \
-        ompts-zh提供】\n\n这个是人格设置指令。\n设置人格: \n/set 人格名。例如/set 编剧\n人格列表: /set list\n人格详细信息: \
-        /set view 人格名\n自定义人格: /set 人格文本\n清除人格: /unset\n【当前人格】: {str(self.provider.now_personality)}", "set"
+            return True, f"【人格文本由PlexPt开源项目awesome-chatgpt-pr \
+        ompts-zh提供】\n设置人格: \n/set 人格名。例如/set 编剧\n人格列表: /set list\n人格详细信息: \
+        /set view 人格名\n自定义人格: /set 人格文本\n重置会话(清除人格): /reset\n重置会话(保留人格): /reset p\n【当前人格】: {str(self.provider.now_personality)}", "set"
         elif l[1] == "list":
             msg = "人格列表：\n"
             for key in personalities.keys():
@@ -137,14 +226,20 @@ class CommandOpenAIOfficial(Command):
                 self.provider.session_dict[session_id] = []
                 new_record = {
                     "user": {
-                        "role": "system",
+                        "role": "user",
                         "content": personalities[ps],
                     },
+                    "AI": {
+                        "role": "assistant",
+                        "content": "好的，接下来我会扮演这个角色。"
+                    },
+                    'type': "personality",
                     'usage_tokens': 0,
                     'single-tokens': 0
                 }
                 self.provider.session_dict[session_id].append(new_record)
-                return True, f"人格{ps}已设置.", "set"
+                self.personality_str = message
+                return True, f"人格{ps}已设置。", "set"
             else:
                 self.provider.now_personality = {
                     'name': '自定义人格',
@@ -152,17 +247,29 @@ class CommandOpenAIOfficial(Command):
                 }
                 new_record = {
                     "user": {
-                        "role": "system",
+                        "role": "user",
                         "content": ps,
                     },
+                    "AI": {
+                        "role": "assistant",
+                        "content": "好的，接下来我会扮演这个角色。"
+                    },
+                    'type': "personality",
                     'usage_tokens': 0,
                     'single-tokens': 0
                 }
                 self.provider.session_dict[session_id] = []
                 self.provider.session_dict[session_id].append(new_record)
+                self.personality_str = message
                 return True, f"自定义人格已设置。 \n人格信息: {ps}", "set"
             
     def draw(self, message):
+        if self.provider is None:
+            return False, "未启动OpenAI ChatGPT语言模型.", "draw"
+        if message.startswith("/画"):
+            message = message[2:]
+        elif message.startswith("画"):
+            message = message[1:]
         try:
             # 画图模式传回3个参数
             img_url = self.provider.image_chat(message)

model/command/command_rev_chatgpt.py

@@ -1,19 +1,134 @@
 from model.command.command import Command
 from model.provider.provider_rev_chatgpt import ProviderRevChatGPT
+from model.platform.qq import QQ
+from cores.qqbot.personality import personalities
+from cores.qqbot.global_object import GlobalObject
 
 class CommandRevChatGPT(Command):
-    def __init__(self, provider: ProviderRevChatGPT):
+    def __init__(self, provider: ProviderRevChatGPT, global_object: GlobalObject):
         self.provider = provider
+        self.cached_plugins = {}
+        self.global_object = global_object
+        self.personality_str = ""
+        super().__init__(provider, global_object)
+
+    def check_command(self, 
+                      message: str, 
+                      session_id: str,
+                      role: str, 
+                      platform: str,
+                      message_obj):
+        self.platform = platform
+        hit, res = super().check_command(
+            message,
+            session_id,
+            role,
+            platform,
+            message_obj
+        )
         
-    def check_command(self, message: str, role):
+        if hit:
+            return True, res
         if self.command_start_with(message, "help", "帮助"):
             return True, self.help()
+        elif self.command_start_with(message, "reset"):
+            return True, self.reset(session_id, message)
         elif self.command_start_with(message, "update"):
             return True, self.update(message, role)
-        elif self.command_start_with(message, "keyword"):
-            return True, self.keyword(message, role)
+        elif self.command_start_with(message, "set"):
+            return True, self.set(message, session_id)
+        elif self.command_start_with(message, "switch"):
+            return True, self.switch(message, session_id)
         return False, None
+
+    def reset(self, session_id, message: str):
+        l = message.split(" ")
+        if len(l) == 1:
+            self.provider.forget(session_id)
+            return True, "重置完毕。", "reset"
+        if len(l) == 2 and l[1] == "p":
+            self.provider.forget(session_id)
+            ret = self.provider.text_chat(self.personality_str)
+            return True, f"重置完毕（保留人格）。\n\n{ret}", "reset"
     
+    def set(self, message: str, session_id: str):
+        l = message.split(" ")
+        if len(l) == 1:
+            return True, f"设置人格: \n/set 人格名或人格文本。例如/set 编剧\n人格列表: /set list\n人格详细信息: \
+        /set view 人格名\n重置会话(清除人格): /reset\n重置会话(保留人格): /reset p", "set"
+        elif l[1] == "list":
+            msg = "人格列表：\n"
+            for key in personalities.keys():
+                msg += f"  |-{key}\n"
+            msg += '\n\n*输入/set view 人格名查看人格详细信息'
+            msg += '\n*不定时更新人格库，请及时更新本项目。'
+            return True, msg, "set"
+        elif l[1] == "view":
+            if len(l) == 2:
+                return True, "请输入/set view 人格名", "set"
+            ps = l[2].strip()
+            if ps in personalities:
+                msg = f"人格【{ps}】详细信息：\n"
+                msg += f"{personalities[ps]}\n"
+            else:
+                msg = f"人格【{ps}】不存在。"
+            return True, msg, "set"
+        else:
+            ps = l[1].strip()
+            if ps in personalities:
+                self.reset(session_id, "reset")
+                self.personality_str = personalities[ps]
+                ret = self.provider.text_chat(self.personality_str, session_id)
+                return True, f"人格【{ps}】已设置。\n\n{ret}", "set"
+            else:
+                self.reset(session_id, "reset")
+                self.personality_str = ps
+                ret = self.provider.text_chat(ps, session_id)
+                return True, f"人格信息已设置。\n\n{ret}", "set"
+
+    def switch(self, message: str, session_id: str):
+        '''
+        切换账号
+        '''
+        l = message.split(" ")
+        rev_chatgpt = self.provider.get_revchatgpt()
+        if len(l) == 1:
+            ret = "当前账号：\n"
+            index = 0
+            curr_ = None
+            for revstat in rev_chatgpt:
+                index += 1
+                ret += f"[{index}]. {revstat['id']}\n"
+                # if session_id in revstat['user']:
+                #     curr_ = revstat['id']
+                for user in revstat['user']:
+                    if session_id == user['id']:
+                        curr_ = revstat['id']
+                        break
+            if curr_ is None:
+                ret += "当前您未选择账号。输入/switch <账号序号>切换账号。"
+            else:
+                ret += f"当前您选择的账号为：{curr_}。输入/switch <账号序号>切换账号。"
+            return True, ret, "switch"
+        elif len(l) == 2:
+            try:
+                index = int(l[1])
+                if index > len(self.provider.rev_chatgpt) or index < 1:
+                    return True, "账号序号不合法。", "switch"
+                else:
+                    # pop
+                    for revstat in self.provider.rev_chatgpt:
+                        if session_id in revstat['user']:
+                            revstat['user'].remove(session_id)
+                    # append
+                    self.provider.rev_chatgpt[index - 1]['user'].append(session_id)
+                    return True, f"切换账号成功。当前账号为：{self.provider.rev_chatgpt[index - 1]['id']}", "switch"
+            except BaseException:
+                return True, "账号序号不合法。", "switch"
+        else:
+            return True, "参数过多。", "switch"
+
     def help(self):
-        return True, "[Github项目名: QQChannelChatGPT，有问题请前往提交issue，欢迎Star此项目~]\n\nRevChatGPT指令面板：\n当前语言模型RevChatGPT未实现任何指令\n", "help"
-    
+        commands = super().general_commands()
+        commands['set'] = '设置人格'
+        return True, super().help_messager(commands, self.platform, self.global_object.cached_plugins), "help"

model/command/command_rev_edgegpt.py

@@ -1,22 +1,46 @@
 from model.command.command import Command
 from model.provider.provider_rev_edgegpt import ProviderRevEdgeGPT
 import asyncio
+from model.platform.qq import QQ
+from cores.qqbot.global_object import GlobalObject
+
 class CommandRevEdgeGPT(Command):
-    def __init__(self, provider: ProviderRevEdgeGPT):
+    def __init__(self, provider: ProviderRevEdgeGPT, global_object: GlobalObject):
         self.provider = provider
+        self.cached_plugins = {}
+        self.global_object = global_object
+        super().__init__(provider, global_object)
+
+    def check_command(self, 
+                      message: str, 
+                      session_id: str, 
+                      role: str, 
+                      platform: str,
+                      message_obj):
+        self.platform = platform
+
+        hit, res = super().check_command(
+            message,
+            session_id,
+            role,
+            platform,
+            message_obj
+        )
         
-    def check_command(self, message: str, loop, role):
+        if hit:
+            return True, res
         if self.command_start_with(message, "reset"):
-            return True, self.reset(loop)
+            return True, self.reset()
         elif self.command_start_with(message, "help"):
             return True, self.help()
         elif self.command_start_with(message, "update"):
             return True, self.update(message, role)
-        elif self.command_start_with(message, "keyword"):
-            return True, self.keyword(message, role)
+
         return False, None
     
-    def reset(self, loop):
+    def reset(self, loop = None):
+        if self.provider is None:
+            return False, "未启动Bing语言模型.", "reset"
         res = asyncio.run_coroutine_threadsafe(self.provider.forget(), loop).result()
         print(res)
         if res:
@@ -25,5 +49,4 @@ class CommandRevEdgeGPT(Command):
             return res, "重置失败", "reset"
     
     def help(self):
-        return True, "[Github项目名: QQChannelChatGPT，有问题请前往提交issue，欢迎Star此项目~]\n\nRevBing指令面板:\nreset: 重置\nhelp: 帮助", "help"
-        
+        return True, super().help_messager(super().general_commands(), self.platform, self.global_object.cached_plugins), "help"

model/platform/qq.py

@@ -1,13 +1,183 @@
-from nakuru.entities.components import Plain
+from nakuru.entities.components import Plain, At, Image, Node
+from util import general_utils as gu
+from util.cmd_config import CmdConfig
+import asyncio
+from nakuru import (
+    CQHTTP,
+    GuildMessage
+)
+import time
 
+
+class FakeSource:
+    def __init__(self, type, group_id):
+        self.type = type
+        self.group_id = group_id
+    
 class QQ:
+    def __init__(self, is_start: bool, cc: CmdConfig = None, gocq_loop = None) -> None:
+        self.is_start = is_start
+        self.gocq_loop = gocq_loop
+        self.cc = cc
+        self.waiting = {}
+        self.gocq_cnt = 0
+
     def run_bot(self, gocq):
-        self.client = gocq
+        self.client: CQHTTP = gocq
         self.client.run()
 
-    async def send_qq_msg(self, source, res):
-        print("[System-Info] 回复QQ消息中..."+res)
-        # 通过消息链处理
-        await self.client.sendGroupMessage(source.group_id, [
-            Plain(text=res)
-        ])
+    def get_msg_loop(self):
+        return self.gocq_loop
+    
+    def get_cnt(self):
+        return self.gocq_cnt
+    
+    def set_cnt(self, cnt):
+        self.gocq_cnt = cnt
+
+    async def send_qq_msg(self, 
+                          source, 
+                          res,
+                          image_mode=None):
+        self.gocq_cnt += 1
+        if not self.is_start:
+            raise Exception("管理员未启动GOCQ平台")
+        """
+         res可以是一个数组, 也就是gocq的消息链。
+         插件开发者请使用send方法, 可以不用直接调用这个方法。
+        """
+        gu.log("回复GOCQ消息: "+str(res), level=gu.LEVEL_INFO, tag="GOCQ", max_len=300)
+
+        if isinstance(source, int):
+            source = FakeSource("GroupMessage", source)
+        
+        # str convert to CQ Message Chain
+        if isinstance(res, str):
+            res_str = res
+            res = []
+            if source.type == "GroupMessage" and not isinstance(source, FakeSource):
+                res.append(At(qq=source.user_id))
+            res.append(Plain(text=res_str))
+
+        # if image mode, put all Plain texts into a new picture.
+        if image_mode is None:
+            image_mode = self.cc.get('qq_pic_mode', False)
+        if image_mode and isinstance(res, list):
+            plains = []
+            news = []
+            for i in res:
+                if isinstance(i, Plain):
+                    plains.append(i.text)
+                else:
+                    news.append(i)
+            plains_str = "".join(plains).strip()
+            if plains_str != "" and len(plains_str) > 50:
+                p = gu.create_markdown_image("".join(plains))
+                news.append(Image.fromFileSystem(p))
+                res = news
+
+        # 回复消息链
+        if isinstance(res, list) and len(res) > 0:
+            if source.type == "GuildMessage":
+                await self.client.sendGuildChannelMessage(source.guild_id, source.channel_id, res)
+                return
+            elif source.type == "FriendMessage":
+                await self.client.sendFriendMessage(source.user_id, res)
+                return
+            elif source.type == "GroupMessage":
+                # 过长时forward发送
+                plain_text_len = 0
+                image_num = 0
+                for i in res:
+                    if isinstance(i, Plain):
+                        plain_text_len += len(i.text)
+                    elif isinstance(i, Image):
+                        image_num += 1
+                if plain_text_len > self.cc.get('qq_forward_threshold', 200):
+                    # 删除At
+                    for i in res:
+                        if isinstance(i, At):
+                            res.remove(i)
+                    node = Node(res)
+                    # node.content = res
+                    node.uin = 123456
+                    node.name = f"bot"
+                    node.time = int(time.time())
+                    # print(node)
+                    nodes=[node]
+                    await self.client.sendGroupForwardMessage(source.group_id, nodes)
+                    return
+                await self.client.sendGroupMessage(source.group_id, res)
+                return
+
+    def send(self, 
+            to,
+            res,
+            image_mode=False,
+            ):
+        '''
+        提供给插件的发送QQ消息接口, 不用在外部await。
+        参数说明：第一个参数可以是消息对象，也可以是QQ群号。第二个参数是消息内容（消息内容可以是消息链列表，也可以是纯文字信息）。
+        第三个参数是是否开启图片模式，如果开启，那么所有纯文字信息都会被合并成一张图片。
+        '''
+        try:
+            asyncio.run_coroutine_threadsafe(self.send_qq_msg(to, res, image_mode), self.gocq_loop).result()
+        except BaseException as e:
+            raise e
+        
+    def send_guild(self, 
+            message_obj,
+            res,
+            ):
+        '''
+        提供给插件的发送GOCQ QQ频道消息接口, 不用在外部await。
+        参数说明：第一个参数必须是消息对象, 第二个参数是消息内容（消息内容可以是消息链列表，也可以是纯文字信息）。
+        '''
+        try:
+            asyncio.run_coroutine_threadsafe(self.send_qq_msg(message_obj, res), self.gocq_loop).result()
+        except BaseException as e:
+            raise e
+
+    def create_text_image(title: str, text: str, max_width=30, font_size=20):
+        '''
+        文本转图片。
+        title: 标题
+        text: 文本内容
+        max_width: 文本宽度最大值（默认30）
+        font_size: 字体大小（默认20）
+
+        返回：文件路径
+        '''
+        try:
+            img = gu.word2img(title, text, max_width, font_size)
+            p = gu.save_temp_img(img)
+            return p
+        except Exception as e:
+            raise e
+
+    def wait_for_message(self, group_id):
+        '''
+        等待下一条消息
+        '''
+        self.waiting[group_id] = ''
+        while True:
+            if group_id in self.waiting and self.waiting[group_id] != '':
+                # 去掉
+                ret = self.waiting[group_id]
+                del self.waiting[group_id]
+                return ret
+            time.sleep(0.5)
+
+    def get_client(self):
+        return self.client
+
+    def nakuru_method_invoker(self, func, *args, **kwargs):
+        """
+        返回一个方法调用器，可以用来立即调用nakuru的方法。
+        """
+        try:
+            ret = asyncio.run_coroutine_threadsafe(func(*args, **kwargs), self.gocq_loop).result()
+            return ret
+        except BaseException as e:
+            raise e
+

model/platform/qqchan.py

@@ -1,63 +1,153 @@
 import io
 import botpy
-from PIL import Image
+from PIL import Image as PILImage
+from botpy.message import Message, DirectMessage
 import re
 import asyncio
 import requests
 from cores.qqbot.personality import personalities
+from util import general_utils as gu
+from nakuru.entities.components import Plain, At, Image
+from botpy.types.message import Reference
 
+class NakuruGuildMember():
+    tiny_id: int
+    user_id: int
+    title: str
+    nickname: str
+    role: int
+    icon_url: str
+
+class NakuruGuildMessage():
+    type: str = "GuildMessage"
+    self_id: int
+    self_tiny_id: int
+    sub_type: str
+    message_id: str
+    guild_id: int
+    channel_id: int
+    user_id: int
+    message: list
+    sender: NakuruGuildMember
+    raw_message: Message
+
+    def __str__(self) -> str:
+        return str(self.__dict__)
 
 class QQChan():
+    def __init__(self, cnt: dict = None) -> None:
+        self.qqchan_cnt = 0
+
+    def get_cnt(self):
+        return self.qqchan_cnt
+    
+    def set_cnt(self, cnt):
+        self.qqchan_cnt = cnt
 
     def run_bot(self, botclient, appid, token):
         intents = botpy.Intents(public_guild_messages=True, direct_message=True) 
         self.client = botclient
         self.client.run(appid=appid, token=token)
 
-    def send_qq_msg(self, message, res, image_mode=False, msg_ref = None):
-        print("[System-Info] 回复QQ频道消息中..."+res)
-        if not image_mode:
-            try:
-                if msg_ref is not None:
-                    reply_res = asyncio.run_coroutine_threadsafe(message.reply(content=res, message_reference = msg_ref), self.client.loop)
+    # gocq-频道SDK兼容层（发）
+    def gocq_compatible_send(self, gocq_message_chain: list):
+        plain_text = ""
+        image_path = None # only one img supported
+        for i in gocq_message_chain:
+            if isinstance(i, Plain):
+                plain_text += i.text
+            elif isinstance(i, Image) and image_path == None:
+                if i.path is not None:
+                    image_path = i.path
                 else:
-                    reply_res = asyncio.run_coroutine_threadsafe(message.reply(content=res), self.client.loop)
-                reply_res.result()
-            except BaseException as e:
-                # 分割过长的消息
-                if "msg over length" in str(e):
-                    split_res = []
-                    split_res.append(res[:len(res)//2])      
-                    split_res.append(res[len(res)//2:])
-                    for i in split_res:
-                        if msg_ref is not None:
-                            reply_res = asyncio.run_coroutine_threadsafe(message.reply(content=i, message_reference = msg_ref), self.client.loop)
-                        else:
-                            reply_res = asyncio.run_coroutine_threadsafe(message.reply(content=i), self.client.loop)
-                        reply_res.result()
-                else:
-                    # 发送qq信息
+                    image_path = i.file
+        return plain_text, image_path
+    
+    # gocq-频道SDK兼容层（收）
+    def gocq_compatible_receive(self, message: Message) -> NakuruGuildMessage:
+        ngm = NakuruGuildMessage()
+        try:
+            ngm.self_id = message.mentions[0].id
+            ngm.self_tiny_id = message.mentions[0].id
+        except:
+            ngm.self_id = 0
+            ngm.self_tiny_id = 0
+
+        ngm.sub_type = "normal"
+        ngm.message_id = message.id
+        ngm.guild_id = int(message.channel_id)
+        ngm.channel_id = int(message.channel_id)
+        ngm.user_id = int(message.author.id)
+        msg = []
+        plain_content = message.content.replace("<@!"+str(ngm.self_id)+">", "").strip()
+        msg.append(Plain(plain_content))
+        if message.attachments:
+            for i in message.attachments:
+                if i.content_type.startswith("image"):
+                    url = i.url
+                    if not url.startswith("http"):
+                        url = "https://"+url
+                    img = Image.fromURL(url)
+                    msg.append(img)
+        ngm.message = msg
+        ngm.sender = NakuruGuildMember()
+        ngm.sender.tiny_id = int(message.author.id)
+        ngm.sender.user_id = int(message.author.id)
+        ngm.sender.title = ""
+        ngm.sender.nickname = message.author.username
+        ngm.sender.role = 0
+        ngm.sender.icon_url = message.author.avatar
+        ngm.raw_message = message
+        return ngm
+
+
+
+    def send_qq_msg(self, message: NakuruGuildMessage, res):
+        gu.log("回复QQ频道消息: "+str(res), level=gu.LEVEL_INFO, tag="QQ频道", max_len=500)
+        self.qqchan_cnt += 1
+        plain_text = ""
+        image_path = None
+        if isinstance(res, list):
+            # 兼容gocq
+            plain_text, image_path = self.gocq_compatible_send(res)
+        elif isinstance(res, str):
+            plain_text = res
+
+        # print(plain_text, image_path)
+        msg_ref = Reference(message_id=message.raw_message.id, ignore_get_message_error=False)
+        if image_path is not None:
+            msg_ref = None
+            if image_path.startswith("http"):
+                pic_res = requests.get(image_path, stream = True)
+                if pic_res.status_code == 200:
+                    image = PILImage.open(io.BytesIO(pic_res.content))
+                    image_path = gu.save_temp_img(image)
+
+        try:
+            reply_res = asyncio.run_coroutine_threadsafe(message.raw_message.reply(content=str(plain_text), message_reference = msg_ref, file_image=image_path), self.client.loop)
+            reply_res.result()
+        except BaseException as e:
+            # 分割过长的消息
+            if "msg over length" in str(e):
+                split_res = []
+                split_res.append(plain_text[:len(plain_text)//2])      
+                split_res.append(plain_text[len(plain_text)//2:])
+                for i in split_res:
+                    reply_res = asyncio.run_coroutine_threadsafe(message.raw_message.reply(content=str(i), message_reference = msg_ref, file_image=image_path), self.client.loop)
+                    reply_res.result()
+            else:
+                # 发送qq信息
+                try:
+                    # 防止被qq频道过滤消息
+                    plain_text = plain_text.replace(".", " . ")
+                    reply_res = asyncio.run_coroutine_threadsafe(message.raw_message.reply(content=str(plain_text), message_reference = msg_ref, file_image=image_path), self.client.loop)
+                    # 发送信息
+                except BaseException as e:
+                    print("QQ频道API错误: \n"+str(e))
                     try:
-                        # 防止被qq频道过滤消息
-                        res = res.replace(".", " . ")
-                        asyncio.run_coroutine_threadsafe(message.reply(content=res), self.client.loop).result()
-                        # 发送信息
+                        reply_res = asyncio.run_coroutine_threadsafe(message.raw_message.reply(content=str(str.join(" ", plain_text)), message_reference = msg_ref, file_image=image_path), self.client.loop)
                     except BaseException as e:
-                        print("QQ频道API错误: \n"+str(e))
-                        res = str.join(" ", res)
-                        try:
-                            asyncio.run_coroutine_threadsafe(message.reply(content=res), self.client.loop).result()
-                        except BaseException as e:
-                            # 如果还是不行则报出错误
-                            res = re.sub(r'(https|http)?:\/\/(\w|\.|\/|\?|\=|\&|\%)*\b', '[被隐藏的链接]', str(e), flags=re.MULTILINE)
-                            res = res.replace(".", "·")
-                            asyncio.run_coroutine_threadsafe(message.reply(content=res), self.client.loop).result()
-                            # send(message, f"QQ频道API错误：{str(e)}\n下面是格式化后的回答：\n{f_res}")
-        else:
-            pic_res = requests.get(str(res), stream=True)
-            if pic_res.status_code == 200:
-                # 将二进制数据转换成图片对象
-                image = Image.open(io.BytesIO(pic_res.content))
-                # 保存图片到本地
-                image.save('tmp_image.jpg')
-            asyncio.run_coroutine_threadsafe(message.reply(file_image='tmp_image.jpg', content=""), self.client.loop)
+                        plain_text = re.sub(r'(https|http)?:\/\/(\w|\.|\/|\?|\=|\&|\%)*\b', '[被隐藏的链接]', str(e), flags=re.MULTILINE)
+                        plain_text = plain_text.replace(".", "·")
+                        asyncio.run_coroutine_threadsafe(message.raw_message.reply(content=plain_text), self.client.loop).result()
+                        # send(message, f"QQ频道API错误：{str(e)}\n下面是格式化后的回答：\n{f_res}")

model/provider/provider.py

@@ -3,16 +3,11 @@ import abc
 class Provider:
     def __init__(self, cfg):
         pass
-
-    def text_chat(self, prompt):
-        pass
-
-    def image_chat(self, prompt):
-        pass
-
-    def memory(self):
+    
+    @abc.abstractmethod
+    def text_chat(self, prompt, session_id, image_url: None, function_call: None):
         pass
 
     @abc.abstractmethod
-    def forget(self) -> bool:
+    def forget(self, session_id = None) -> bool:
         pass

model/provider/provider_openai_official.py

@@ -1,4 +1,5 @@
-import openai
+from openai import OpenAI
+from openai.types.chat.chat_completion import ChatCompletion
 import json
 import time
 import os
@@ -6,26 +7,44 @@ import sys
 from cores.database.conn import dbConn
 from model.provider.provider import Provider
 import threading
+from util import general_utils as gu
+import traceback
+import tiktoken
 
 abs_path = os.path.dirname(os.path.realpath(sys.argv[0])) + '/'
-key_record_path = abs_path+'chatgpt_key_record'
+key_record_path = abs_path + 'chatgpt_key_record'
 
 class ProviderOpenAIOfficial(Provider):
     def __init__(self, cfg):
         self.key_list = []
-        if 'api_base' in cfg and cfg['api_base'] != 'none' and cfg['api_base'] != '':
-            openai.api_base = cfg['api_base']
+        # 如果 cfg['key']中有长度为1的字符串，那么是格式错误，直接报错
+        for key in cfg['key']:
+            if len(key) == 1:
+                input("检查到了长度为 1 的Key。配置文件中的 openai.key 处的格式错误 (符号 - 的后面要加空格)，请退出程序并检查配置文件，按回车跳过。")
+                raise BaseException("配置文件格式错误")
         if cfg['key'] != '' and cfg['key'] != None:
-            print("[System] 读取ChatGPT Key成功")
             self.key_list = cfg['key']
         else:
             input("[System] 请先去完善ChatGPT的Key。详情请前往https://beta.openai.com/account/api-keys")
+        if len(self.key_list) == 0:
+            raise Exception("您打开了 OpenAI 模型服务，但是未填写 key。请前往填写。")
         
-        # init key record
-        self.init_key_record()
+        self.key_stat = {}
+        for k in self.key_list:
+            self.key_stat[k] = {'exceed': False, 'used': 0}
 
-        self.chatGPT_configs = cfg['chatGPTConfigs']
-        print(f'[System] 加载ChatGPTConfigs: {self.chatGPT_configs}')
+        self.api_base = None
+        if 'api_base' in cfg and cfg['api_base'] != 'none' and cfg['api_base'] != '':
+            self.api_base = cfg['api_base']
+            print(f"设置 api_base 为: {self.api_base}")
+        # openai client
+        self.client = OpenAI(
+            api_key=self.key_list[0],
+            base_url=self.api_base
+        )
+
+        self.openai_model_configs: dict = cfg['chatGPTConfigs']
+        gu.log(f'加载 OpenAI Chat Configs: {self.openai_model_configs}')
         self.openai_configs = cfg
         # 会话缓存
         self.session_dict = {}
@@ -34,14 +53,17 @@ class ProviderOpenAIOfficial(Provider):
         # 历史记录持久化间隔时间
         self.history_dump_interval = 20
 
+        self.enc = tiktoken.get_encoding("cl100k_base")
+
         # 读取历史记录
         try:
             db1 = dbConn()
             for session in db1.get_all_session():
                 self.session_dict[session[0]] = json.loads(session[1])['data']
-            print("[System] 历史记录读取成功喵")
+            gu.log("读取历史记录成功。")
         except BaseException as e:
-            print("[System] 历史记录读取失败: " + str(e))
+            gu.log("读取历史记录失败，但不影响使用。", level=gu.LEVEL_ERROR)
+
 
         # 读取统计信息
         if not os.path.exists(abs_path+"configs/stat"):
@@ -65,7 +87,7 @@ class ProviderOpenAIOfficial(Provider):
         self.now_personality = {}
 
 
-    # 转储历史记录的定时器~ Soulter
+    # 转储历史记录
     def dump_history(self):
         time.sleep(10)
         db = dbConn()
@@ -88,7 +110,11 @@ class ProviderOpenAIOfficial(Provider):
             # 每隔10分钟转储一次
             time.sleep(10*self.history_dump_interval)
 
-    def text_chat(self, prompt, session_id):
+    def text_chat(self, prompt, session_id = None, image_url = None, function_call=None):
+        if session_id is None:
+            session_id = "unknown"
+            if "unknown" in self.session_dict:
+                del self.session_dict["unknown"] 
         # 会话机制
         if session_id not in self.session_dict:
             self.session_dict[session_id] = []
@@ -106,43 +132,93 @@ class ProviderOpenAIOfficial(Provider):
                 f.write(json.dumps(fjson))
                 f.flush()
                 f.close()
+        
+        # 使用 tictoken 截断消息
+        _encoded_prompt = self.enc.encode(prompt)
+        prompt = self.enc.decode(_encoded_prompt[:self.openai_model_configs['max_tokens'] - 100])
+        gu.log(f"注意，有一部分 prompt 文本由于超出 token 限制而被截断。", level=gu.LEVEL_WARNING, max_len=300)
 
-        cache_data_list, new_record, req = self.wrap(prompt, session_id)
+        cache_data_list, new_record, req = self.wrap(prompt, session_id, image_url)
+        gu.log(f"CACHE_DATA_: {str(cache_data_list)}", level=gu.LEVEL_DEBUG, max_len=99999)
+        gu.log(f"OPENAI REQUEST: {str(req)}", level=gu.LEVEL_DEBUG, max_len=9999)
         retry = 0
         response = None
-        while retry < 5:
+        err = ''
+
+        # 截断倍率
+        truncate_rate = 0.75
+
+        use_gpt4v = False
+        for i in req:
+            if isinstance(i['content'], list):
+                use_gpt4v = True
+                break
+        if image_url is not None:
+            use_gpt4v = True
+        if use_gpt4v:
+            conf = self.openai_model_configs.copy()
+            conf['model'] = 'gpt-4-vision-preview'
+        else:
+            conf = self.openai_model_configs
+        print(req)
+        while retry < 10:
             try:
-                response = openai.ChatCompletion.create(
-                    messages=req,
-                    **self.chatGPT_configs
-                )
+                if function_call is None:
+                    response = self.client.chat.completions.create(
+                        messages=req,
+                        **conf
+                    )
+                else:
+                    response = self.client.chat.completions.create(
+                        messages=req,
+                        tools = function_call,
+                        **conf
+                    )
                 break
             except Exception as e:
-                print(e)
+                print(traceback.format_exc())
+                if 'Invalid content type. image_url is only supported by certain models.' in str(e):
+                    raise e
                 if 'You exceeded' in str(e) or 'Billing hard limit has been reached' in str(e) or 'No API key provided' in str(e) or 'Incorrect API key provided' in str(e):
-                    print("[System] 当前Key已超额或者不正常,正在切换")
-                    self.key_stat[openai.api_key]['exceed'] = True
-                    self.save_key_record()
-
-                    response, is_switched = self.handle_switch_key(req)
+                    gu.log("当前Key已超额或异常, 正在切换", level=gu.LEVEL_WARNING)
+                    self.key_stat[self.client.api_key]['exceed'] = True
+                    is_switched = self.handle_switch_key()
                     if not is_switched:
                         # 所有Key都超额或不正常
                         raise e
-                    else:
-                        break
-                if 'maximum context length' in str(e):
-                    print("token超限, 清空对应缓存")
+                    retry -= 1
+                elif 'maximum context length' in str(e):
+                    gu.log("token超限, 清空对应缓存，并进行消息截断")
                     self.session_dict[session_id] = []
+                    prompt = prompt[:int(len(prompt)*truncate_rate)]
+                    truncate_rate -= 0.05
                     cache_data_list, new_record, req = self.wrap(prompt, session_id)
-                retry+=1
-        if retry >= 5:
-            raise BaseException("连接超时")
-        
-        self.key_stat[openai.api_key]['used'] += response['usage']['total_tokens']
-        self.save_key_record()
-        print("[ChatGPT] "+str(response["choices"][0]["message"]["content"]))
-        chatgpt_res = str(response["choices"][0]["message"]["content"]).strip()
-        current_usage_tokens = response['usage']['total_tokens']
+
+                elif 'Limit: 3 / min. Please try again in 20s.' in str(e) or "OpenAI response error" in str(e):
+                    time.sleep(30)
+                    continue
+                else:
+                    gu.log(str(e), level=gu.LEVEL_ERROR)
+                time.sleep(2)
+                err = str(e)
+                retry += 1
+        if retry >= 10:
+            gu.log(r"如果报错, 且您的机器在中国大陆内, 请确保您的电脑已经设置好代理软件(梯子), 并在配置文件设置了系统代理地址。详见https://github.com/Soulter/QQChannelChatGPT/wiki/%E4%BA%8C%E3%80%81%E9%A1%B9%E7%9B%AE%E9%85%8D%E7%BD%AE%E6%96%87%E4%BB%B6%E9%85%8D%E7%BD%AE", max_len=999)
+            raise BaseException("连接出错: "+str(err))
+        assert isinstance(response, ChatCompletion)
+        gu.log(f"OPENAI RESPONSE: {response.usage}", level=gu.LEVEL_DEBUG, max_len=9999)
+
+        # 结果分类
+        choice = response.choices[0]
+        if choice.message.content != None:
+            # 文本形式
+            chatgpt_res = str(choice.message.content).strip()
+        elif choice.message.tool_calls != None and len(choice.message.tool_calls) > 0:
+            # tools call (function calling)
+            return choice.message.tool_calls[0].function
+
+        self.key_stat[self.client.api_key]['used'] += response.usage.total_tokens
+        current_usage_tokens = response.usage.total_tokens
 
         # 超过指定tokens， 尽可能的保留最多的条目，直到小于max_tokens
         if current_usage_tokens > self.max_tokens:
@@ -152,7 +228,7 @@ class ProviderOpenAIOfficial(Provider):
                 if index >= len(cache_data_list):
                     break
                 # 保留人格信息
-                if 'user' in cache_data_list[index] and cache_data_list[index]['user']['role'] != 'system':
+                if cache_data_list[index]['type'] != 'personality':
                     t -= int(cache_data_list[index]['single_tokens'])
                     del cache_data_list[index]
                 else:
@@ -171,6 +247,7 @@ class ProviderOpenAIOfficial(Provider):
             new_record['single_tokens'] = current_usage_tokens - int(cache_data_list[-1]['usage_tokens'])
         else:
             new_record['single_tokens'] = current_usage_tokens
+
         cache_data_list.append(new_record)
 
         self.session_dict[session_id] = cache_data_list
@@ -182,39 +259,35 @@ class ProviderOpenAIOfficial(Provider):
         image_url = ''
         while retry < 5:
             try:
-                # print("test1")
-                response = openai.Image.create(
+                response = self.client.images.generate(
                     prompt=prompt,
                     n=img_num,
                     size=img_size
                 )
-                # print("test2")
                 image_url = []
                 for i in range(img_num):
                     image_url.append(response['data'][i]['url'])
-                print(image_url)
                 break
             except Exception as e:
-                print(e)
+                gu.log(str(e), level=gu.LEVEL_ERROR)
                 if 'You exceeded' in str(e) or 'Billing hard limit has been reached' in str(
                         e) or 'No API key provided' in str(e) or 'Incorrect API key provided' in str(e):
-                    print("[System] 当前Key已超额或者不正常,正在切换")
-                    self.key_stat[openai.api_key]['exceed'] = True
-                    self.save_key_record()
-
-                    response, is_switched = self.handle_switch_key(req)
+                    gu.log("当前 Key 已超额或者不正常, 正在切换", level=gu.LEVEL_WARNING)
+                    self.key_stat[self.client.api_key]['exceed'] = True
+                    is_switched = self.handle_switch_key()
                     if not is_switched:
                         # 所有Key都超额或不正常
                         raise e
-                    else:
-                        break
-                retry += 1
+                else:
+                    retry += 1
         if retry >= 5:
             raise BaseException("连接超时")
                 
         return image_url
 
-    def forget(self, session_id) -> bool:
+    def forget(self, session_id = None) -> bool:
+        if session_id is None:
+            return False
         self.session_dict[session_id] = []
         return True
     
@@ -275,7 +348,20 @@ class ProviderOpenAIOfficial(Provider):
             return -1, -1, -1, -1
 
     # 包装信息
-    def wrap(self, prompt, session_id):
+    def wrap(self, prompt, session_id, image_url = None):
+        if image_url is not None:
+            prompt = [
+                {
+                    "type": "text",
+                    "text": prompt
+                },
+                {
+                    "type": "image_url",
+                    "image_url": {
+                        "url": image_url
+                    }
+                }
+            ]
         # 获得缓存信息
         context = self.session_dict[session_id]
         new_record = {
@@ -284,6 +370,7 @@ class ProviderOpenAIOfficial(Provider):
                 "content": prompt,
             },
             "AI": {},
+            'type': "common",
             'usage_tokens': 0,
         }
         req_list = []
@@ -295,104 +382,53 @@ class ProviderOpenAIOfficial(Provider):
         req_list.append(new_record['user'])
         return context, new_record, req_list
     
-    def handle_switch_key(self, req):
+    def handle_switch_key(self):
         # messages = [{"role": "user", "content": prompt}]
-        while True:
-            is_all_exceed = True
-            for key in self.key_stat:
-                if key == None:
-                    continue
-                if not self.key_stat[key]['exceed']:
-                    is_all_exceed = False
-                    openai.api_key = key
-                    print(f"[System] 切换到Key: {key}, 已使用token: {self.key_stat[key]['used']}")
-                    if len(req) > 0:
-                        try:
-                            response = openai.ChatCompletion.create(
-                                messages=req,
-                                **self.chatGPT_configs
-                            )
-                            return response, True
-                        except Exception as e:
-                            print(e)
-                            if 'You exceeded' in str(e):
-                                print("[System] 当前Key已超额,正在切换")
-                                self.key_stat[openai.api_key]['exceed'] = True
-                                self.save_key_record()
-                                time.sleep(1)
-                                continue
-                    else:
-                        return True
-            if is_all_exceed:
-                print("[System] 所有Key已超额")
-                return None, False
-            else:
-                print("[System] 在切换key时程序异常。")
-                return None, False
-                
-    def getConfigs(self):
+        is_all_exceed = True
+        for key in self.key_stat:
+            if key == None or self.key_stat[key]['exceed']:
+                continue
+            is_all_exceed = False
+            self.client.api_key = key
+            gu.log(f"切换到Key: {key}, 已使用token: {self.key_stat[key]['used']}", level=gu.LEVEL_INFO)
+            break
+        if is_all_exceed:
+            gu.log("所有Key已超额", level=gu.LEVEL_CRITICAL)
+            return False
+        return True
+        
+    def get_configs(self):
         return self.openai_configs
     
-    def save_key_record(self):
-        with open(key_record_path, 'w', encoding='utf-8') as f:
-            json.dump(self.key_stat, f)
-
     def get_key_stat(self):
         return self.key_stat
+    
     def get_key_list(self):
         return self.key_list
     
+    def get_curr_key(self):
+        return self.client.api_key
+    
     # 添加key
     def append_key(self, key, sponsor):
         self.key_list.append(key)
         self.key_stat[key] = {'exceed': False, 'used': 0, 'sponsor': sponsor}
-        self.save_key_record()
-        self.init_key_record()
 
     # 检查key是否可用
     def check_key(self, key):
-        pre_key = openai.api_key
-        openai.api_key = key
-        messages = [{"role": "user", "content": "1"}]
+        client_ = OpenAI(
+            api_key=key,
+            base_url=self.api_base
+        )
+        messages = [{"role": "user", "content": "please just echo `test`"}]
         try:
-            response = openai.ChatCompletion.create(
+            client_.chat.completions.create(
                 messages=messages,
-                **self.chatGPT_configs
+                **self.openai_model_configs
             )
-            openai.api_key = pre_key
             return True
         except Exception as e:
             pass
-        openai.api_key = pre_key
         return False
-
-    #将key_list的key转储到key_record中，并记录相关数据
-    def init_key_record(self):
-        if not os.path.exists(key_record_path):
-            with open(key_record_path, 'w', encoding='utf-8') as f:
-                json.dump({}, f)
-        with open(key_record_path, 'r', encoding='utf-8') as keyfile:
-            try:
-                self.key_stat = json.load(keyfile)
-            except Exception as e:
-                print(e)
-                self.key_stat = {}
-            finally:
-                for key in self.key_list:
-                    if key not in self.key_stat:
-                        self.key_stat[key] = {'exceed': False, 'used': 0}
-                        # if openai.api_key is None:
-                        #     openai.api_key = key
-                    else:
-                        # if self.key_stat[key]['exceed']:
-                        #     print(f"Key: {key} 已超额")
-                        #     continue
-                        # else:
-                        #     if openai.api_key is None:
-                        #         openai.api_key = key
-                        #         print(f"使用Key: {key}, 已使用token: {self.key_stat[key]['used']}")
-                        pass
-                if openai.api_key == None:
-                    self.handle_switch_key("")
-            self.save_key_record()
+        
 

model/provider/provider_rev_chatgpt.py

@@ -1,25 +1,64 @@
 from revChatGPT.V1 import Chatbot
+from revChatGPT import typings
 from model.provider.provider import Provider
+from util import general_utils as gu
+from util import cmd_config as cc
+import time
+
 
 class ProviderRevChatGPT(Provider):
-    def __init__(self, config):
-        self.rev_chatgpt = []
+    def __init__(self, config, base_url = None):
+        if base_url == "":
+            base_url = None
+        self.rev_chatgpt: list[dict] = []
+        self.cc = cc.CmdConfig()
         for i in range(0, len(config['account'])):
             try:
-                print(f"[System] 创建rev_ChatGPT负载{str(i)}: " + str(config['account'][i]))
+                gu.log(f"创建逆向ChatGPT负载{str(i+1)}中...", level=gu.LEVEL_INFO, tag="RevChatGPT")
+
                 if 'password' in config['account'][i]:
-                    config['account'][i]['password'] = str(config['account'][i]['password'])
+                    gu.log(f"创建逆向ChatGPT负载{str(i+1)}失败: 已不支持账号密码登录，请使用access_token方式登录。", level=gu.LEVEL_ERROR, tag="RevChatGPT")
+                    continue
+                rev_account_config = {
+                    'access_token': config['account'][i]['access_token'],
+                }
+                if self.cc.get("rev_chatgpt_model") != "":
+                    rev_account_config['model'] = self.cc.get("rev_chatgpt_model")
+                if len(self.cc.get("rev_chatgpt_plugin_ids")) > 0:
+                    rev_account_config['plugin_ids'] = self.cc.get("rev_chatgpt_plugin_ids")
+                if self.cc.get("rev_chatgpt_PUID") != "":
+                    rev_account_config['PUID'] = self.cc.get("rev_chatgpt_PUID")
+                if len(self.cc.get("rev_chatgpt_unverified_plugin_domains")) > 0:
+                    rev_account_config['unverified_plugin_domains'] = self.cc.get("rev_chatgpt_unverified_plugin_domains")
+                cb = Chatbot(config=rev_account_config, base_url=base_url)
+                # cb.captcha_solver = self.__captcha_solver
+                # 后八位c
+                g_id = rev_account_config['access_token'][-8:]
                 revstat = {
-                    'obj': Chatbot(config=config['account'][i]),
-                    'busy': False
+                    'id': g_id,
+                    'obj': cb,
+                    'busy': False,
+                    'user': []
                 }
                 self.rev_chatgpt.append(revstat)
             except BaseException as e:
-                print(f"[System] 创建rev_ChatGPT负载失败: {str(e)}")
+                gu.log(f"创建逆向ChatGPT负载{str(i+1)}失败: {str(e)}", level=gu.LEVEL_ERROR, tag="RevChatGPT")
 
-    def forget(self) -> bool:
+    def forget(self, session_id = None) -> bool:
+        for i in self.rev_chatgpt:
+            for user in i['user']:
+                if session_id == user['id']:
+                    try:
+                        i['obj'].reset_chat()
+                        return True
+                    except BaseException as e:
+                        gu.log(f"重置RevChatGPT失败。原因: {str(e)}", level=gu.LEVEL_ERROR, tag="RevChatGPT")
+                        return False
         return False
-    
+
+    def get_revchatgpt(self) -> list:
+        return self.rev_chatgpt
+        
     def request_text(self, prompt: str, bot) -> str:
         resp = ''
         err_count = 0
@@ -30,41 +69,145 @@ class ProviderRevChatGPT(Provider):
                 for data in bot.ask(prompt):
                     resp = data["message"]
                 break
+            except typings.Error as e:
+                if e.code == typings.ErrorType.INVALID_ACCESS_TOKEN_ERROR:
+                    raise e
+                if e.code == typings.ErrorType.EXPIRED_ACCESS_TOKEN_ERROR:
+                    raise e
+                if e.code == typings.ErrorType.PROHIBITED_CONCURRENT_QUERY_ERROR:
+                    raise e
+                if "Your authentication token has expired. Please try signing in again." in str(e):
+                    raise e
+                if "The message you submitted was too long" in str(e):
+                    raise e
+                if "You've reached our limit of messages per hour." in str(e):
+                    raise e
+                if "Rate limited by proxy" in str(e):
+                    gu.log(f"触发请求频率限制, 60秒后自动重试。", level=gu.LEVEL_WARNING, tag="RevChatGPT")
+                    time.sleep(60)
+                
+                err_count += 1
+                gu.log(f"请求异常: {str(e)}，正在重试。({str(err_count)})", level=gu.LEVEL_WARNING, tag="RevChatGPT")
+                if err_count >= retry_count:
+                    raise e
             except BaseException as e:
-                try:
-                    print("[RevChatGPT] 请求出现了一些问题, 正在重试。次数"+str(err_count))
-                    err_count += 1
-                    if err_count >= retry_count:
-                        raise e
-                except BaseException:
-                    err_count += 1
+                err_count += 1
+                gu.log(f"请求异常: {str(e)}，正在重试。({str(err_count)})", level=gu.LEVEL_WARNING, tag="RevChatGPT")
+                if err_count >= retry_count:
+                    raise e
+        if resp == '':
+            resp = "RevChatGPT请求异常。"
         
-        print("[RevChatGPT] "+str(resp))
+        # print("[RevChatGPT] "+str(resp))
         return resp
 
-    def text_chat(self, prompt):
+    def text_chat(self, prompt, session_id = None, image_url = None, function_call=None) -> str:
+
+        # 选择一个人少的账号。
+        selected_revstat = None
+        min_revstat = None
+        min_ = None
+        new_user = False
+        conversation_id = ''
+        parent_id = ''
+        for revstat in self.rev_chatgpt:
+            for user in revstat['user']:
+                if session_id == user['id']:
+                    selected_revstat = revstat
+                    conversation_id = user['conversation_id']
+                    parent_id = user['parent_id']
+                    break
+            if min_ is None:
+                min_ = len(revstat['user'])
+                min_revstat = revstat
+            elif len(revstat['user']) < min_:
+                min_ = len(revstat['user'])
+                min_revstat = revstat
+            # if session_id in revstat['user']:
+            #     selected_revstat = revstat
+            #     break
+        
+        if selected_revstat is None:
+            selected_revstat = min_revstat
+            selected_revstat['user'].append({
+                'id': session_id,
+                'conversation_id': '',
+                'parent_id': ''
+            })
+            new_user = True
+
+        gu.log(f"选择账号{str(selected_revstat)}", tag="RevChatGPT", level=gu.LEVEL_DEBUG)
+
+        while selected_revstat['busy']:
+            gu.log(f"账号忙碌，等待中...", tag="RevChatGPT", level=gu.LEVEL_DEBUG)
+            time.sleep(1)
+        selected_revstat['busy'] = True
+        
+        if not new_user:
+            # 非新用户，则使用其专用的会话
+            selected_revstat['obj'].conversation_id = conversation_id
+            selected_revstat['obj'].parent_id = parent_id
+        else:
+            # 新用户，则使用新的会话
+            selected_revstat['obj'].reset_chat()
+
         res = ''
-        print("[Debug] "+str(self.rev_chatgpt))
+        err_msg = ''
+        err_cnt = 0
+        while err_cnt < 15:
+            try:
+                res = self.request_text(prompt, selected_revstat['obj'])
+                selected_revstat['busy'] = False
+                # 记录新用户的会话
+                if new_user:
+                    i = 0
+                    for user in selected_revstat['user']:
+                        if user['id'] == session_id:
+                            selected_revstat['user'][i]['conversation_id'] = selected_revstat['obj'].conversation_id
+                            selected_revstat['user'][i]['parent_id'] = selected_revstat['obj'].parent_id
+                            break
+                        i += 1
+                return res.strip()
+            except BaseException as e:
+                if "Your authentication token has expired. Please try signing in again." in str(e):
+                    raise Exception(f"此账号(access_token后8位为{selected_revstat['id']})的access_token已过期，请重新获取，或者切换账号。")
+                if "The message you submitted was too long" in str(e):
+                    raise Exception("发送的消息太长，请分段发送。")
+                if "You've reached our limit of messages per hour." in str(e):
+                    raise Exception("触发RevChatGPT请求频率限制。请1小时后再试，或者切换账号。")
+                gu.log(f"请求异常: {str(e)}", level=gu.LEVEL_WARNING, tag="RevChatGPT")
+                err_cnt += 1
+                time.sleep(3)
+
+        raise Exception(f'回复失败。原因：{err_msg}。如果您设置了多个账号，可以使用/switch指令切换账号。输入/switch查看详情。')
+            
+
+        # while self.is_all_busy():
+        #     time.sleep(1)
+        # res = ''
+        # err_msg = ''
+        # cursor = 0
+        # for revstat in self.rev_chatgpt:
+        #     cursor += 1
+        #     if not revstat['busy']:
+        #         try:
+        #             revstat['busy'] = True
+        #             res = self.request_text(prompt, revstat['obj'])
+        #             revstat['busy'] = False
+        #             return res.strip()
+        #         # todo: 细化错误管理
+        #         except BaseException as e:
+        #             revstat['busy'] = False
+        #             gu.log(f"请求出现问题: {str(e)}", level=gu.LEVEL_WARNING, tag="RevChatGPT")
+        #             err_msg += f"账号{cursor} - 错误原因: {str(e)}"
+        #             continue
+        #     else:
+        #         err_msg += f"账号{cursor} - 错误原因: 忙碌"
+        #         continue
+        # raise Exception(f'回复失败。错误跟踪：{err_msg}')
+    
+    def is_all_busy(self) -> bool:
         for revstat in self.rev_chatgpt:
             if not revstat['busy']:
-                try:
-                    revstat['busy'] = True
-                    print("[Debug] 使用逆向ChatGPT回复ing", end='', flush=True)
-                    res = self.request_text(prompt, revstat['obj'])
-                    print("OK")
-                    revstat['busy'] = False
-                    # 处理结果文本
-                    chatgpt_res = res.strip()
-                    return res
-                except Exception as e:
-                    print("[System-Error] 逆向ChatGPT回复失败" + str(e))
-                    try:
-                        if e.code == 2:
-                            print("[System-Error] 频率限制，正在切换账号。"+ str(e))
-                            continue
-                        else:
-                            res = '所有的非忙碌OpenAI账号经过测试都暂时出现问题，请稍后再试或者联系管理员~'
-                            return res
-                    except BaseException:
-                        continue
-        res = '所有的OpenAI账号都有负载, 请稍后再试~'
+                return False
+        return True

model/provider/provider_rev_edgegpt.py

@@ -1,6 +1,12 @@
 from model.provider.provider import Provider
-from EdgeGPT import Chatbot, ConversationStyle
+# from EdgeGPT import Chatbot, ConversationStyle
 import json
+import os
+from util import general_utils as gu
+from util.cmd_config import CmdConfig as cc
+import time
+from EdgeGPT.EdgeUtils import Query, Cookie
+from EdgeGPT.EdgeGPT import Chatbot as EdgeChatbot, ConversationStyle, NotAllowedToAccess
 
 class ProviderRevEdgeGPT(Provider):
     def __init__(self):
@@ -8,21 +14,30 @@ class ProviderRevEdgeGPT(Provider):
         self.wait_stack = []
         with open('./cookies.json', 'r') as f:
             cookies = json.load(f)
-        self.bot = Chatbot(cookies=cookies)
+        proxy = cc.get("bing_proxy", None)
+        if proxy == "":
+            proxy = None
+        # q = Query("Hello, bing!", cookie_files="./cookies.json")
+        # print(q)
+        self.bot = EdgeChatbot(cookies=cookies, proxy = "http://127.0.0.1:7890")
+        ret = self.bot.ask_stream("Hello, bing!", conversation_style=ConversationStyle.creative, wss_link="wss://ai.nothingnessvoid.tech/sydney/ChatHub")
+        # self.bot = Chatbot(cookies=cookies, proxy = proxy)
+        for i in ret:
+            print(i, flush=True)
 
     def is_busy(self):
         return self.busy
 
-    async def forget(self):
+    async def forget(self, session_id = None):
         try:
             await self.bot.reset()
             return True
         except BaseException:
             return False
         
-    async def text_chat(self, prompt):
-        if self.busy:
-             return
+    async def text_chat(self, prompt, platform = 'none', image_url=None, function_call=None):
+        while self.busy:
+            time.sleep(1)
         self.busy = True
         resp = 'err'
         err_count = 0
@@ -31,22 +46,59 @@ class ProviderRevEdgeGPT(Provider):
         while err_count < retry_count:
             try:
                 resp = await self.bot.ask(prompt=prompt, conversation_style=ConversationStyle.creative)
-                resp = resp['item']['messages'][len(resp['item']['messages'])-1]['text']
-                if 'I\'m sorry but I prefer not to continue this conversation. I\'m still learning so I appreciate your understanding and patience.' in resp:
+                # print("[RevEdgeGPT] "+str(resp))
+                if 'messages' not in resp['item']:
+                    await self.bot.reset()
+                msj_obj = resp['item']['messages'][len(resp['item']['messages'])-1]
+                reply_msg = msj_obj['text']
+                if 'sourceAttributions' in msj_obj:
+                    reply_source = msj_obj['sourceAttributions']
+                else:
+                    reply_source = []
+                if 'throttling' in resp['item']:
+                    throttling = resp['item']['throttling']
+                    # print(throttling)
+                else:
+                    throttling = None
+                if 'I\'m sorry but I prefer not to continue this conversation. I\'m still learning so I appreciate your understanding and patience.' in reply_msg:
                     self.busy = False
                     return '', 0
-                if resp == prompt:
-                    resp += '\n\n如果你没有让我复述你的话，那代表我可能不想和你继续这个话题了，请输入reset重置会话😶'
+                if reply_msg == prompt:
+                    # resp += '\n\n如果你没有让我复述你的话，那代表我可能不想和你继续这个话题了，请输入reset重置会话😶'
+                    await self.forget()
+                    err_count += 1
+                    continue
+                if reply_source is None:
+                    # 不想答复
+                    return '', 0
+                else:
+                    if platform != 'qqchan':
+                        index = 1
+                        if len(reply_source) > 0:
+                            reply_msg += "\n\n信息来源:\n"
+                        for i in reply_source:
+                            reply_msg += f"[{str(index)}]: {i['seeMoreUrl']} | {i['providerDisplayName']}\n"
+                            index += 1
+                if throttling is not None:
+                    if throttling['numUserMessagesInConversation'] == throttling['maxNumUserMessagesInConversation']:
+                        # 达到上限，重置会话
+                        await self.forget()
+                    if throttling['numUserMessagesInConversation'] > throttling['maxNumUserMessagesInConversation']:
+                        await self.forget()
+                        err_count += 1
+                        continue
+                    reply_msg += f"\n[{throttling['numUserMessagesInConversation']}/{throttling['maxNumUserMessagesInConversation']}]"
                 break
             except BaseException as e:
-                print(e.with_traceback)
+                gu.log(str(e), level=gu.LEVEL_WARNING, tag="RevEdgeGPT")
                 err_count += 1
                 if err_count >= retry_count:
+                        gu.log(r"如果报错, 且您的机器在中国大陆内, 请确保您的电脑已经设置好代理软件(梯子), 并在配置文件设置了系统代理地址。详见https://github.com/Soulter/QQChannelChatGPT/wiki/%E4%BA%8C%E3%80%81%E9%A1%B9%E7%9B%AE%E9%85%8D%E7%BD%AE%E6%96%87%E4%BB%B6%E9%85%8D%E7%BD%AE", max_len=999)
                         self.busy = False
                         raise e
-                print("[RevEdgeGPT] 请求出现了一些问题, 正在重试。次数"+str(err_count))
+                gu.log("请求出现了一些问题, 正在重试。次数"+str(err_count), level=gu.LEVEL_WARNING, tag="RevEdgeGPT")
         self.busy = False
         
-        print("[RevEdgeGPT] "+str(resp))
-        return resp, 1
+        # print("[RevEdgeGPT] "+str(reply_msg))
+        return reply_msg, 1
     

requirements.txt

@@ -1,10 +1,15 @@
-requests
-openai~=0.27.4
+pydantic~=1.10.4
+requests~=2.28.1
+openai~=1.2.3
 qq-botpy
-revChatGPT~=4.0.8
-baidu-aip
-EdgeGPT~=0.1.22.1
-chardet
-Pillow
-GitPython
-git+https://github.com/Lxns-Network/nakuru-project.git
+chardet~=5.1.0
+Pillow~=9.4.0
+GitPython~=3.1.31
+nakuru-project
+beautifulsoup4
+googlesearch-python
+tictoken
+readability-lxml
+EdgeGPT
+revChatGPT~=6.8.6
+baidu-aip~=4.16.9

util/cmd_config.py

@@ -0,0 +1,53 @@
+import os
+import json
+
+cpath = "cmd_config.json"
+
+def check_exist():
+    if not os.path.exists(cpath):
+        with open(cpath, "w", encoding="utf-8") as f:
+            json.dump({}, f, indent=4)
+            f.flush()
+
+class CmdConfig():
+
+    @staticmethod
+    def get(key, default=None):
+        check_exist()
+        with open(cpath, "r", encoding="utf-8") as f:
+            d = json.load(f)
+            if key in d:
+                return d[key]
+            else:
+                return default
+            
+    @staticmethod
+    def get_all():
+        check_exist()
+        with open(cpath, "r", encoding="utf-8") as f:
+            return json.load(f)
+        
+    @staticmethod
+    def put(key, value):
+        check_exist()
+        with open(cpath, "r", encoding="utf-8") as f:
+            d = json.load(f)
+            d[key] = value
+            with open(cpath, "w", encoding="utf-8") as f:
+                json.dump(d, f, indent=4)
+                f.flush()
+
+    @staticmethod
+    def init_attributes(keys: list, init_val = ""):
+        check_exist()
+        with open(cpath, "r", encoding="utf-8") as f:
+            d = json.load(f)
+            _tag = False
+            for k in keys:
+                if k not in d:
+                    d[k] = init_val
+                    _tag = True
+            if _tag:
+                with open(cpath, "w", encoding="utf-8") as f:
+                    json.dump(d, f, indent=4)
+                    f.flush()

util/errors/errors.py

@@ -1,3 +0,0 @@
-class PromptExceededError(Exception):
-    
-    pass

util/func_call.py

@@ -0,0 +1,237 @@
+
+import json
+import util.general_utils as gu
+
+import time
+class FuncCallJsonFormatError(Exception):
+    def __init__(self, msg):
+        self.msg = msg
+
+    def __str__(self):
+        return self.msg
+    
+class FuncNotFoundError(Exception):
+    def __init__(self, msg):
+        self.msg = msg
+
+    def __str__(self):
+        return self.msg
+    
+class FuncCall():
+    def __init__(self, provider) -> None:
+        self.func_list = []
+        self.provider = provider
+
+    def add_func(self, name: str = None, func_args: list = None, desc: str = None, func_obj = None) -> None:
+        if name == None or func_args == None or desc == None or func_obj == None:
+            raise FuncCallJsonFormatError("name, func_args, desc must be provided.")
+        params = {
+            "type": "object", # hardcore here
+            "properties": {}
+        }
+        for param in func_args:
+            params['properties'][param['name']] = {
+                "type": param['type'],
+                "description": param['description']
+            }
+        self._func = {
+            "name": name,
+            "parameters": params,
+            "description": desc,
+            "func_obj": func_obj,
+        }
+        self.func_list.append(self._func)
+
+    def func_dump(self, intent: int = 2) -> str:
+        _l = []
+        for f in self.func_list:
+            _l.append({
+                "name": f["name"],
+                "parameters": f["parameters"],
+                "description": f["description"],
+            })
+        return json.dumps(_l, indent=intent, ensur_ascii=False)
+    
+    def get_func(self) -> list:
+        _l = []
+        for f in self.func_list:
+            _l.append({
+                "type": "function",
+                "function": {
+                    "name": f["name"],
+                    "parameters": f["parameters"],
+                    "description": f["description"],
+                }
+            })
+        return _l
+    
+    def func_call(self, question, func_definition, is_task = False, tasks = None, taskindex = -1, is_summary = True, session_id = None):
+
+        funccall_prompt = """
+我正实现function call功能，该功能旨在让你变成给定的问题到给定的函数的解析器（意味着你不是创造函数）。
+下面会给你提供可能用到的函数相关信息和一个问题，你需要将其转换成给定的函数调用。
+- 你的返回信息只含json，请严格仿照以下内容（不含注释），必须含有`res`,`func_call`字段:
+```
+{
+    "res": string // 如果没有找到对应的函数，那么你可以在这里正常输出内容。如果有，这里是空字符串。
+    "func_call": [ // 这是一个数组，里面包含了所有的函数调用，如果没有函数调用，那么这个数组是空数组。
+        {
+            "res": string // 如果没有找到对应的函数，那么你可以在这里正常输出内容。如果有，这里是空字符串。
+            "name": str, // 函数的名字
+            "args_type": {
+                "arg1": str, // 函数的参数的类型
+                "arg2": str,
+                ...
+            },
+            "args": {
+                "arg1": any, // 函数的参数
+                "arg2": any,
+                ...
+            }
+        },
+        ... // 可能在这个问题中会有多个函数调用
+    ],
+}
+```
+- 如果用户的要求较复杂，允许返回多个函数调用，但需保证这些函数调用的顺序正确。
+- 当问题没有提到给定的函数时，相当于提问方不打算使用function call功能，这时你可以在res中正常输出这个问题的回答（以AI的身份正常回答该问题，并将答案输出在res字段中，回答不要涉及到任何函数调用的内容，就只是正常讨论这个问题。）
+
+提供的函数是：
+
+"""
+
+        prompt = f"{funccall_prompt}\n```\n{func_definition}\n```\n"
+        prompt += f"""
+用户的提问是：
+```
+{question}
+```
+"""
+
+        # if is_task:
+        #     # task_prompt = f"\n任务列表为{str(tasks)}\n你目前进行到了任务{str(taskindex)}, **你不需要重新进行已经进行过的任务, 不要生成已经进行过的**"
+        #     prompt += task_prompt
+
+        # provider.forget()
+
+        _c = 0
+        while _c < 3:
+            try:
+                res = self.provider.text_chat(prompt, session_id)
+                if res.find('```') != -1:
+                    res = res[res.find('```json') + 7: res.rfind('```')]
+                gu.log("REVGPT func_call json result", bg=gu.BG_COLORS["green"], fg=gu.FG_COLORS["white"])
+                print(res)
+                res = json.loads(res)
+                break
+            except Exception as e:
+                _c += 1
+                if _c == 3:
+                    raise e
+                if "The message you submitted was too long" in str(e):
+                    raise e
+
+        invoke_func_res = ""
+
+        if "func_call" in res and len(res["func_call"]) > 0:
+            task_list = res["func_call"]
+
+            invoke_func_res_list = []
+
+            for res in task_list:
+                # 说明有函数调用
+                func_name = res["name"]
+                # args_type = res["args_type"]
+                args = res["args"]
+                # 调用函数
+                # func = eval(func_name)
+                func_target = None
+                for func in self.func_list:
+                    if func["name"] == func_name:
+                        func_target = func["func_obj"]
+                        break
+                if func_target == None:
+                    raise FuncNotFoundError(f"Request function {func_name} not found.")
+                t_res = str(func_target(**args))
+                invoke_func_res += f"{func_name} 调用结果：\n```\n{t_res}\n```\n"
+                invoke_func_res_list.append(invoke_func_res)
+                gu.log(f"[FUNC| {func_name} invoked]", bg=gu.BG_COLORS["green"], fg=gu.FG_COLORS["white"])
+                # print(str(t_res))
+
+            if is_summary:
+
+                # 生成返回结果
+                after_prompt = """
+有以下内容："""+invoke_func_res+"""
+请以AI助手的身份结合返回的内容对用户提问做详细全面的回答。
+用户的提问是：
+```""" + question + """```
+- 在res字段中，不要输出函数的返回值，也不要针对返回值的字段进行分析，也不要输出用户的提问，而是理解这一段返回的结果，并以AI助手的身份回答问题，只需要输出回答的内容，不需要在回答的前面加上身份词。
+- 你的返回信息必须只能是json，且需严格遵循以下内容（不含注释）:
+```json
+{
+    "res": string, // 回答的内容
+    "func_call_again": bool // 如果函数返回的结果有错误或者问题，可将其设置为true，否则为false
+}
+```
+- 如果func_call_again为true，res请你设为空值，否则请你填写回答的内容。"""
+
+                _c = 0
+                while _c < 5:
+                    try:
+                        res = self.provider.text_chat(after_prompt, session_id)
+                        # 截取```之间的内容
+                        gu.log("DEBUG BEGIN", bg=gu.BG_COLORS["yellow"], fg=gu.FG_COLORS["white"])
+                        print(res)
+                        gu.log("DEBUG END", bg=gu.BG_COLORS["yellow"], fg=gu.FG_COLORS["white"])
+                        if res.find('```') != -1:
+                            res = res[res.find('```json') + 7: res.rfind('```')]
+                        gu.log("REVGPT after_func_call json result", bg=gu.BG_COLORS["green"], fg=gu.FG_COLORS["white"])
+                        after_prompt_res = res
+                        after_prompt_res = json.loads(after_prompt_res)
+                        break
+                    except Exception as e:
+                        _c += 1
+                        if _c == 5:
+                            raise e
+                        if "The message you submitted was too long" in str(e):
+                            # 如果返回的内容太长了，那么就截取一部分
+                            time.sleep(3)
+                            invoke_func_res = invoke_func_res[:int(len(invoke_func_res) / 2)]
+                            after_prompt = """
+函数返回以下内容："""+invoke_func_res+"""
+请以AI助手的身份结合返回的内容对用户提问做详细全面的回答。
+用户的提问是：
+```""" + question + """```
+- 在res字段中，不要输出函数的返回值，也不要针对返回值的字段进行分析，也不要输出用户的提问，而是理解这一段返回的结果，并以AI助手的身份回答问题，只需要输出回答的内容，不需要在回答的前面加上身份词。
+- 你的返回信息必须只能是json，且需严格遵循以下内容（不含注释）:
+```json
+{
+    "res": string, // 回答的内容
+    "func_call_again": bool // 如果函数返回的结果有错误或者问题，可将其设置为true，否则为false
+}
+```
+- 如果func_call_again为true，res请你设为空值，否则请你填写回答的内容。"""
+                        else:
+                            raise e
+
+                if "func_call_again" in after_prompt_res and after_prompt_res["func_call_again"]:
+                    # 如果需要重新调用函数
+                    # 重新调用函数
+                    gu.log("REVGPT func_call_again", bg=gu.BG_COLORS["purple"], fg=gu.FG_COLORS["white"])
+                    res = self.func_call(question, func_definition)
+                    return res, True
+
+                gu.log("REVGPT func callback:", bg=gu.BG_COLORS["green"], fg=gu.FG_COLORS["white"])
+                # print(after_prompt_res["res"])
+                return after_prompt_res["res"], True
+            else:
+                return str(invoke_func_res_list), True
+        else:
+            # print(res["res"])
+            return res["res"], False
+
+
+
+
+

util/general_utils.py

@@ -0,0 +1,514 @@
+import datetime
+import time
+import socket
+from PIL import Image, ImageDraw, ImageFont
+import os
+import re
+import requests
+
+PLATFORM_GOCQ = 'gocq'
+PLATFORM_QQCHAN = 'qqchan'
+
+FG_COLORS = {
+    "black": "30",
+    "red": "31",
+    "green": "32",
+    "yellow": "33",
+    "blue": "34",
+    "purple": "35",
+    "cyan": "36",
+    "white": "37",
+    "default": "39",
+}
+
+BG_COLORS = {
+    "black": "40",
+    "red": "41",
+    "green": "42",
+    "yellow": "43",
+    "blue": "44",
+    "purple": "45",
+    "cyan": "46",
+    "white": "47",
+    "default": "49",
+}
+
+LEVEL_DEBUG = "DEBUG"
+LEVEL_INFO = "INFO"
+LEVEL_WARNING = "WARNING"
+LEVEL_ERROR = "ERROR"
+LEVEL_CRITICAL = "CRITICAL"
+
+level_codes = {
+    LEVEL_DEBUG: 0,
+    LEVEL_INFO: 1,
+    LEVEL_WARNING: 2,
+    LEVEL_ERROR: 3,
+    LEVEL_CRITICAL: 4
+}
+
+level_colors = {
+    "INFO": "green",
+    "WARNING": "yellow",
+    "ERROR": "red",
+    "CRITICAL": "purple",
+}
+
+def log(
+        msg: str,
+        level: str = "INFO",
+        tag: str = "System",
+        fg: str = None,
+        bg: str = None,
+        max_len: int = 500,
+        err: Exception = None,):
+    """
+    日志打印函数
+    """
+    _set_level_code = level_codes[LEVEL_INFO]
+    if 'LOG_LEVEL' in os.environ and os.environ['LOG_LEVEL'] in level_codes:
+        _set_level_code = level_codes[os.environ['LOG_LEVEL']]
+
+    if level in level_codes and level_codes[level] < _set_level_code:
+        return
+    
+    if err is not None:
+        msg += "\n异常原因: " + str(err)
+        level = LEVEL_ERROR
+
+    if len(msg) > max_len:
+        msg = msg[:max_len] + "..."
+    now = datetime.datetime.now().strftime("%m-%d %H:%M:%S")
+    pre = f"[{now}] [{level}] [{tag}]: {msg}"
+    if level == "INFO":
+        if fg is None:
+            fg = FG_COLORS["green"]
+        if bg is None:
+            bg = BG_COLORS["default"]
+    elif level == "WARNING":
+        if fg is None:
+            fg = FG_COLORS["yellow"]
+        if bg is None:
+            bg = BG_COLORS["default"]
+    elif level == "ERROR":
+        if fg is None:
+            fg = FG_COLORS["red"]
+        if bg is None:
+            bg = BG_COLORS["default"]
+    elif level == "CRITICAL":
+        if fg is None:
+            fg = FG_COLORS["purple"]
+        if bg is None:
+            bg = BG_COLORS["default"]
+    
+    print(f"\033[{fg};{bg}m{pre}\033[0m")
+
+
+def port_checker(port: int, host: str = "localhost"):
+    sk = socket.socket(socket.AF_INET,socket.SOCK_STREAM)
+    sk.settimeout(1)
+    try:
+        sk.connect((host, port))
+        sk.close()
+        return True
+    except Exception:
+        sk.close()
+        return False
+    
+def word2img(title: str, text: str, max_width=30, font_size=20):
+    if os.path.exists("resources/fonts/syst.otf"):
+        font_path = "resources/fonts/syst.otf"
+    elif os.path.exists("QQChannelChatGPT/resources/fonts/syst.otf"):
+        font_path = "QQChannelChatGPT/resources/fonts/syst.otf"
+    elif os.path.exists("C:/Windows/Fonts/simhei.ttf"):
+        font_path = "C:/Windows/Fonts/simhei.ttf"
+    elif os.path.exists("/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc"):
+        font_path = "/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc"
+    else:
+        raise Exception("找不到字体文件")
+    
+    width_factor = 1.0
+    height_factor = 1.5
+    # 格式化文本宽度最大为30
+    lines = text.split('\n')
+    i = 0
+    length = len(lines)
+    for l in lines:
+        if len(l) > max_width:
+            # lines[i] = l[:max_width] + '\n' + l[max_width:]
+            # for
+            cp = l
+            for ii in range(len(l)):
+                if ii % max_width == 0:
+                    cp = cp[:ii] + '\n' + cp[ii:]
+                    length += 1
+            lines[i] = cp
+        i += 1
+    text = '\n'.join(lines)
+    width = int(max_width * font_size * width_factor)
+    height = int(length * font_size * height_factor)
+    image = Image.new('RGB', (width, height), (255, 255, 255))
+    draw = ImageDraw.Draw(image)
+
+    text_font = ImageFont.truetype(font_path, font_size)
+    title_font = ImageFont.truetype(font_path, font_size + 5)
+    # 标题居中
+    title_width, title_height = title_font.getsize(title)
+    draw.text(((width - title_width) / 2, 10), title, fill=(0, 0, 0), font=title_font)
+    # 文本不居中
+    draw.text((10, title_height+20), text, fill=(0, 0, 0), font=text_font)
+
+    return image
+
+
+def render_markdown(markdown_text, image_width=800, image_height=600, font_size=26, font_color=(0, 0, 0), bg_color=(255, 255, 255)):
+
+    HEADER_MARGIN = 20
+    HEADER_FONT_STANDARD_SIZE = 42
+
+    QUOTE_LEFT_LINE_MARGIN = 10
+    QUOTE_FONT_LINE_MARGIN = 6 # 引用文字距离左边线的距离和上下的距离
+    QUOTE_LEFT_LINE_HEIGHT = font_size + QUOTE_FONT_LINE_MARGIN * 2
+    QUOTE_LEFT_LINE_WIDTH = 5
+    QUOTE_LEFT_LINE_COLOR = (180, 180, 180)
+    QUOTE_FONT_SIZE = font_size
+    QUOTE_FONT_COLOR = (180, 180, 180)
+    # QUOTE_BG_COLOR = (255, 255, 255)
+
+    CODE_BLOCK_MARGIN = 10
+    CODE_BLOCK_FONT_SIZE = font_size
+    CODE_BLOCK_FONT_COLOR = (255, 255, 255)
+    CODE_BLOCK_BG_COLOR = (240, 240, 240)
+    CODE_BLOCK_CODES_MARGIN_VERTICAL = 5 # 代码块和代码之间的距离
+    CODE_BLOCK_CODES_MARGIN_HORIZONTAL = 5 # 代码块和代码之间的距离
+    CODE_BLOCK_TEXT_MARGIN = 4 # 代码和代码之间的距离
+
+    INLINE_CODE_MARGIN = 8
+    INLINE_CODE_FONT_SIZE = font_size
+    INLINE_CODE_FONT_COLOR = font_color
+    INLINE_CODE_FONT_MARGIN = 4
+    INLINE_CODE_BG_COLOR = (230, 230, 230)
+    INLINE_CODE_BG_HEIGHT = INLINE_CODE_FONT_SIZE + INLINE_CODE_FONT_MARGIN * 2
+
+    LIST_MARGIN = 8
+    LIST_FONT_SIZE = font_size
+    LIST_FONT_COLOR = font_color
+
+    TEXT_LINE_MARGIN = 8
+
+    IMAGE_MARGIN = 15
+    # 用于匹配图片的正则表达式
+    IMAGE_REGEX = r"!\s*\[.*?\]\s*\((.*?)\)"
+
+    
+    if os.path.exists("resources/fonts/syst.otf"):
+        font_path = "resources/fonts/syst.otf"
+    elif os.path.exists("QQChannelChatGPT/resources/fonts/syst.otf"):
+        font_path = "QQChannelChatGPT/resources/fonts/syst.otf"
+    elif os.path.exists("C:/Windows/Fonts/simhei.ttf"):
+        font_path = "C:/Windows/Fonts/simhei.ttf"
+    elif os.path.exists("/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc"):
+        font_path = "/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc"
+    else:
+        raise Exception("找不到字体文件")
+    # backup
+    if os.path.exists("resources/fonts/simhei.ttf"):
+        font_path1 = "resources/fonts/simhei.ttf"
+    elif os.path.exists("QQChannelChatGPT/resources/fonts/simhei.ttf"):
+        font_path1 = "QQChannelChatGPT/resources/fonts/simhei.ttf"
+    else:
+        font_path1 = font_path
+    # 加载字体
+    font = ImageFont.truetype(font_path, font_size)
+    
+    images: Image = {}
+    
+    # pre_process, get height of each line
+    pre_lines = markdown_text.split('\n')
+    height = 0
+    pre_in_code = False
+    i = -1
+    _pre_lines = []
+    for line in pre_lines:
+        i += 1
+        # 处理图片
+        if re.search(IMAGE_REGEX, line):
+            try:
+                image_url = re.findall(IMAGE_REGEX, line)[0]
+                print(image_url)
+                image_res = Image.open(requests.get(image_url, stream=True, timeout=5).raw)
+                images[i] = image_res
+                # 最大不得超过image_width的50%
+                img_height = image_res.size[1]
+
+                if image_res.size[0] > image_width*0.5:
+                    image_res = image_res.resize((int(image_width*0.5), int(image_res.size[1]*image_width*0.5/image_res.size[0])))
+                    img_height = image_res.size[1]
+
+                height += img_height + IMAGE_MARGIN*2
+            
+                line = re.sub(IMAGE_REGEX, "", line)
+            except Exception as e:
+                print(e)
+                line = re.sub(IMAGE_REGEX, "\n[加载失败的图片]\n", line)
+                continue
+            
+        line.replace("\t", "    ")
+        if font.getsize(line)[0] > image_width:
+            cp = line
+            _width = 0
+            _word_cnt = 0
+            for ii in range(len(line)):
+                # 检测是否是中文
+                _width += font.getsize(line[ii])[0]
+                _word_cnt+=1
+                if _width > image_width:
+                    _pre_lines.append(cp[:_word_cnt])
+                    cp = cp[_word_cnt:]
+                    _word_cnt=0
+                    _width=0
+            _pre_lines.append(cp)
+        else:
+            _pre_lines.append(line)
+    pre_lines = _pre_lines
+
+    i=-1
+    for line in pre_lines:
+        if line == "":
+            height += TEXT_LINE_MARGIN
+            continue
+        i += 1
+        line = line.strip()
+        if pre_in_code and not line.startswith("```"):
+            height += font_size + CODE_BLOCK_TEXT_MARGIN
+            # pre_codes.append(line)
+            continue
+        if line.startswith("#"):
+            header_level = line.count("#")
+            height += HEADER_FONT_STANDARD_SIZE + HEADER_MARGIN*2 - header_level * 4
+        elif line.startswith("-"):
+            height += font_size+LIST_MARGIN*2
+        elif line.startswith(">"):
+            height += font_size+QUOTE_LEFT_LINE_MARGIN*2
+        elif line.startswith("```"):
+            if pre_in_code:
+                pre_in_code = False
+                # pre_codes = []
+                height += CODE_BLOCK_MARGIN
+            else:
+                pre_in_code = True
+                height += CODE_BLOCK_MARGIN
+        elif re.search(r"`(.*?)`", line):
+            height += font_size+INLINE_CODE_FONT_MARGIN*2+INLINE_CODE_MARGIN*2
+        else:
+            height += font_size + TEXT_LINE_MARGIN*2
+
+    markdown_text = '\n'.join(pre_lines)
+    print("Pre process done, height: ", height)
+    image_height = height
+    if image_height < 100:
+        image_height = 100
+    image_width += 20
+    
+    # 创建空白图像
+    image = Image.new('RGB', (image_width, image_height), bg_color)
+    draw = ImageDraw.Draw(image)
+
+
+    # # get all the emojis unicode in the markdown text
+    # unicode_text = markdown_text.encode('unicode_escape').decode()
+    # # print(unicode_text)
+    # unicode_emojis = re.findall(r'\\U\w{8}', unicode_text)
+    # emoji_base_url = "https://abs.twimg.com/emoji/v1/72x72/{unicode_emoji}.png"
+
+    # 设置初始位置
+    x, y = 10, 10
+
+    # 解析Markdown文本
+    lines = markdown_text.split("\n")
+    # lines = pre_lines
+
+    in_code_block = False
+    code_block_start_y = 0
+    code_block_codes = []
+
+    index = -1
+    for line in lines:
+        index += 1
+        if in_code_block and not line.startswith("```"):
+            code_block_codes.append(line)
+            y += font_size + CODE_BLOCK_TEXT_MARGIN
+            continue
+        line = line.strip()
+
+        if line.startswith("#"):
+            # unicode_emojis = re.findall(r'\\U0001\w{4}', line)
+            # for unicode_emoji in unicode_emojis:
+            #     line = line.replace(unicode_emoji, "")
+            # unicode_emoji = ""
+            # if len(unicode_emojis) > 0:
+            #     unicode_emoji = unicode_emojis[0]
+            
+            # 处理标题
+            header_level = line.count("#")
+            line = line.strip("#").strip()
+            font_size_header = HEADER_FONT_STANDARD_SIZE - header_level * 4
+
+            # if unicode_emoji != "":
+            #     emoji_url = emoji_base_url.format(unicode_emoji=unicode_emoji[-5:])
+            #     emoji = Image.open(requests.get(emoji_url, stream=True).raw)
+            #     emoji = emoji.resize((font_size, font_size))
+            #     image.paste(emoji, (x, y))
+            #     x += font_size
+
+            font = ImageFont.truetype(font_path, font_size_header)
+            y += HEADER_MARGIN # 上边距
+            # 字间距
+            draw.text((x, y), line, font=font, fill=font_color)
+            draw.line((x, y + font_size_header + 8, image_width - 10, y + font_size_header + 8), fill=(230, 230, 230), width=3)
+            y += font_size_header + HEADER_MARGIN
+
+        elif line.startswith(">"):
+            # 处理引用
+            quote_text = line.strip(">")
+            # quote_width = image_width - 20  # 引用框的宽度为图像宽度减去左右边距
+            # quote_height = font_size + 10  # 引用框的高度为字体大小加上上下边距
+            # quote_box = (x, y, x + quote_width, y + quote_height)
+            # draw.rounded_rectangle(quote_box, radius=5, fill=(230, 230, 230), width=2)  # 使用灰色填充矩形框作为引用背景
+            y+=QUOTE_LEFT_LINE_MARGIN
+            draw.line((x, y, x, y + QUOTE_LEFT_LINE_HEIGHT), fill=QUOTE_LEFT_LINE_COLOR, width=QUOTE_LEFT_LINE_WIDTH)
+            font = ImageFont.truetype(font_path, QUOTE_FONT_SIZE)
+            draw.text((x + QUOTE_FONT_LINE_MARGIN, y + QUOTE_FONT_LINE_MARGIN), quote_text, font=font, fill=QUOTE_FONT_COLOR)
+            y += font_size + QUOTE_LEFT_LINE_HEIGHT + QUOTE_LEFT_LINE_MARGIN
+        
+        elif line.startswith("-"):
+            # 处理列表
+            list_text = line.strip("-").strip()
+            font = ImageFont.truetype(font_path, LIST_FONT_SIZE)
+            y += LIST_MARGIN
+            draw.text((x, y), "  ·  " + list_text, font=font, fill=LIST_FONT_COLOR)
+            y += font_size + LIST_MARGIN
+
+        elif line.startswith("```"):
+            if not in_code_block:
+                code_block_start_y = y+CODE_BLOCK_MARGIN
+                in_code_block = True
+            else:
+                # print(code_block_codes)
+                in_code_block = False
+                codes = "\n".join(code_block_codes)
+                code_block_codes = []
+                draw.rounded_rectangle((x, code_block_start_y, image_width - 10, y+CODE_BLOCK_CODES_MARGIN_VERTICAL + CODE_BLOCK_TEXT_MARGIN), radius=5, fill=CODE_BLOCK_BG_COLOR, width=2)
+                font = ImageFont.truetype(font_path1, CODE_BLOCK_FONT_SIZE)
+                draw.text((x + CODE_BLOCK_CODES_MARGIN_HORIZONTAL, code_block_start_y + CODE_BLOCK_CODES_MARGIN_VERTICAL), codes, font=font, fill=font_color)
+                y += CODE_BLOCK_CODES_MARGIN_VERTICAL + CODE_BLOCK_MARGIN
+        # y += font_size+10
+        elif re.search(r"`(.*?)`", line):
+            y += INLINE_CODE_MARGIN # 上边距
+            # 处理行内代码
+            code_regex = r"`(.*?)`"
+            parts_inline = re.findall(code_regex, line)
+            # print(parts_inline)
+            parts = re.split(code_regex, line)
+            # print(parts)
+            for part in parts:
+                # the judge has a tiny bug.
+                # when line is like "hi`hi`". all the parts will be in parts_inline.
+                if part in parts_inline:
+                    font = ImageFont.truetype(font_path, INLINE_CODE_FONT_SIZE)
+                    code_text = part.strip("`")
+                    code_width = font.getsize(code_text)[0] + INLINE_CODE_FONT_MARGIN*2
+                    x += INLINE_CODE_MARGIN
+                    code_box = (x, y, x + code_width, y + INLINE_CODE_BG_HEIGHT)
+                    draw.rounded_rectangle(code_box, radius=5, fill=INLINE_CODE_BG_COLOR, width=2)  # 使用灰色填充矩形框作为引用背景
+                    draw.text((x+INLINE_CODE_FONT_MARGIN, y), code_text, font=font, fill=font_color)
+                    x += code_width+INLINE_CODE_MARGIN-INLINE_CODE_FONT_MARGIN
+                else:
+                    font = ImageFont.truetype(font_path, font_size)
+                    draw.text((x, y), part, font=font, fill=font_color)
+                    x += font.getsize(part)[0]
+            y += font_size + INLINE_CODE_MARGIN
+            x = 10
+
+        else:
+            # 处理普通文本
+            if line == "":
+                y += TEXT_LINE_MARGIN
+            else:
+                font = ImageFont.truetype(font_path, font_size)
+                
+                draw.text((x, y), line, font=font, fill=font_color)
+                y += font_size + TEXT_LINE_MARGIN*2
+
+        # 图片特殊处理
+        if index in images:
+            image_res = images[index]
+            # 最大不得超过image_width的50%
+            if image_res.size[0] > image_width*0.5:
+                image_res = image_res.resize((int(image_width*0.5), int(image_res.size[1]*image_width*0.5/image_res.size[0])))
+            image.paste(image_res, (IMAGE_MARGIN, y))
+            y += image_res.size[1] + IMAGE_MARGIN*2
+    return image
+
+
+def save_temp_img(img: Image) -> str:
+    if not os.path.exists("temp"):
+        os.makedirs("temp")
+
+    # 获得文件创建时间，清除超过1小时的
+    try:
+        for f in os.listdir("temp"):
+            path = os.path.join("temp", f)
+            if os.path.isfile(path):
+                ctime = os.path.getctime(path)
+                if time.time() - ctime > 3600:
+                    os.remove(path)
+    except Exception as e:
+        log(f"清除临时文件失败: {e}", level=LEVEL_WARNING, tag="GeneralUtils")
+
+    # 获得时间戳
+    timestamp = int(time.time())
+    p = f"temp/{timestamp}.png"
+    img.save(p)
+    return p
+
+
+def create_text_image(title: str, text: str, max_width=30, font_size=20):
+    '''
+    文本转图片。
+    title: 标题
+    text: 文本内容
+    max_width: 文本宽度最大值（默认30）
+    font_size: 字体大小（默认20）
+
+    返回：文件路径
+    '''
+    try:
+        img = word2img(title, text, max_width, font_size)
+        p = save_temp_img(img)
+        return p
+    except Exception as e:
+        raise e
+    
+def create_markdown_image(text: str):
+    '''
+    markdown文本转图片。
+    返回：文件路径
+    '''
+    try:
+        img = render_markdown(text)
+        p = save_temp_img(img)
+        return p
+    except Exception as e:
+        raise e
+    
+def test_markdown():
+    # 示例使用
+    markdown_text = """# Help Center
+    ! [] (https://soulter.top/helpme.jpg)
+    """
+    image = render_markdown(markdown_text)
+    image.show()
+
+# test_markdown()
+

util/gplugin.py

@@ -0,0 +1,279 @@
+import requests
+import util.general_utils as gu
+from bs4 import BeautifulSoup
+import time
+from util.func_call import (
+    FuncCall, 
+    FuncCallJsonFormatError, 
+    FuncNotFoundError
+)
+from openai.types.chat.chat_completion_message_tool_call import Function
+import traceback
+from googlesearch import search, SearchResult
+from model.provider.provider import Provider
+import json
+from readability import Document
+
+
+def tidy_text(text: str) -> str:
+    '''
+    清理文本，去除空格、换行符等
+    '''
+    return text.strip().replace("\n", " ").replace("\r", " ").replace("  ", " ")
+
+def special_fetch_zhihu(link: str) -> str:
+    '''
+    function-calling 函数, 用于获取知乎文章的内容
+    '''
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) \
+            AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
+    }
+    response = requests.get(link, headers=headers)
+    response.encoding = "utf-8"
+    soup = BeautifulSoup(response.text, "html.parser")
+
+    if "zhuanlan.zhihu.com" in link:
+        r = soup.find(class_="Post-RichTextContainer")
+    else:
+        r = soup.find(class_="List-item").find(class_="RichContent-inner")
+    if r is None:
+        print("debug: zhihu none")
+        raise Exception("zhihu none")
+    return tidy_text(r.text)
+
+def google_web_search(keyword) -> str:
+    '''
+    获取 google 搜索结果, 得到 title、desc、link
+    '''
+    ret = ""
+    index = 1
+    try:
+        ls = search(keyword, advanced=True, num_results=4)
+        for i in ls:
+            desc = i.description
+            try:
+                desc = fetch_website_content(i.url)
+            except BaseException as e:
+                print(f"(google) fetch_website_content err: {str(e)}")
+            gu.log(f"# No.{str(index)}\ntitle: {i.title}\nurl: {i.url}\ncontent: {desc}\n\n", level=gu.LEVEL_DEBUG, max_len=9999)
+            ret += f"# No.{str(index)}\ntitle: {i.title}\nurl: {i.url}\ncontent: {desc}\n\n"
+            index += 1
+    except Exception as e:
+        print(f"google search err: {str(e)}")
+        return web_keyword_search_via_bing(keyword)
+    return ret
+
+def web_keyword_search_via_bing(keyword) -> str:
+    '''
+    获取bing搜索结果, 得到 title、desc、link
+    '''
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) \
+            AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
+    }
+    url = "https://www.bing.com/search?q="+keyword
+    _cnt = 0
+    _detail_store = []
+    while _cnt < 5:
+        try:
+            response = requests.get(url, headers=headers)
+            response.encoding = "utf-8"
+            gu.log(f"bing response: {response.text}", tag="bing", level=gu.LEVEL_DEBUG, max_len=9999)
+            soup = BeautifulSoup(response.text, "html.parser")
+            res = []
+            ols = soup.find(id="b_results")
+            for i in ols.find_all("li", class_="b_algo"):
+                try:
+                    title = i.find("h2").text
+                    desc = i.find("p").text
+                    link = i.find("h2").find("a").get("href")
+                    res.append({
+                        "title": title,
+                        "desc": desc,
+                        "link": link,
+                    })
+                    if len(res) >= 5: # 限制5条
+                        break
+                    if len(_detail_store) >= 3:
+                        continue
+
+                    # 爬取前两条的网页内容
+                    if "zhihu.com" in link:
+                        try:
+                            _detail_store.append(special_fetch_zhihu(link))
+                        except BaseException as e:
+                            print(f"zhihu parse err: {str(e)}")
+                    else:
+                        try:
+                            _detail_store.append(fetch_website_content(link))
+                        except BaseException as e:
+                            print(f"fetch_website_content err: {str(e)}")
+
+                except Exception as e:
+                    print(f"bing parse err: {str(e)}")
+            if len(res) == 0:
+                break
+            if len(_detail_store) > 0:
+                ret = f"{str(res)} \n具体网页内容: {str(_detail_store)}"
+            else:
+                ret = f"{str(res)}"
+            return str(ret)
+        except Exception as e:
+            gu.log(f"bing fetch err: {str(e)}")
+            _cnt += 1
+            time.sleep(1)
+            
+    gu.log("fail to fetch bing info, using sougou.")
+    return google_web_search(keyword)
+
+def web_keyword_search_via_sougou(keyword) -> str:
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) \
+            AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36",
+    }
+    url = f"https://sogou.com/web?query={keyword}"
+    response = requests.get(url, headers=headers)
+    response.encoding = "utf-8"
+    soup = BeautifulSoup(response.text, "html.parser")
+    
+    res = []
+    results = soup.find("div", class_="results")
+    for i in results.find_all("div", class_="vrwrap"):
+        try:
+            title = tidy_text(i.find("h3").text)
+            link = tidy_text(i.find("h3").find("a").get("href"))
+            if link.startswith("/link?url="):
+                link = "https://www.sogou.com" + link
+            res.append({
+                "title": title,
+                "link": link,
+            })
+            if len(res) >= 5: # 限制5条
+                break
+        except Exception as e:
+            gu.log(f"sougou parse err: {str(e)}", tag="web_keyword_search_via_sougou", level=gu.LEVEL_ERROR)
+    # 爬取网页内容
+    _detail_store = []
+    for i in res:
+        if _detail_store >= 3:
+            break
+        try:
+            _detail_store.append(fetch_website_content(i["link"]))
+        except BaseException as e:
+            print(f"fetch_website_content err: {str(e)}")
+    ret = f"{str(res)}"
+    if len(_detail_store) > 0:
+        ret += f"\n网页内容: {str(_detail_store)}"
+    return ret
+
+def fetch_website_content(url):
+    gu.log(f"fetch_website_content: {url}", tag="fetch_website_content", level=gu.LEVEL_DEBUG)
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) \
+            AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
+    }
+    response = requests.get(url, headers=headers, timeout=3)
+    response.encoding = "utf-8"
+    # soup = BeautifulSoup(response.text, "html.parser")
+    # # 如果有container / content / main等的话，就只取这些部分
+    # has = False
+    # beleive_ls = ["container", "content", "main"]
+    # res = ""
+    # for cls in beleive_ls:
+    #     for i in soup.find_all(class_=cls):
+    #         has = True
+    #         res += i.text
+    # if not has:
+    #     res = soup.text
+    # res = res.replace("\n", "").replace("  ", " ").replace("\r", "").replace("\t", "")
+    # if not has:
+    #     res = res[300:1100]
+    # else:
+    #     res = res[100:800]
+    # # with open(f"temp_{time.time()}.html", "w", encoding="utf-8") as f:
+    # #     f.write(res)
+    # gu.log(f"fetch_website_content: end", tag="fetch_website_content", level=gu.LEVEL_DEBUG)
+    # return res
+    doc = Document(response.content)
+    # print('title:', doc.title())
+    ret = doc.summary(html_partial=True)
+    soup = BeautifulSoup(ret, 'html.parser')
+    ret = tidy_text(soup.get_text())
+    return ret
+
+def web_search(question, provider: Provider, session_id, official_fc=False):
+    '''
+    official_fc: 使用官方 function-calling
+    '''
+    new_func_call = FuncCall(provider)
+    new_func_call.add_func("google_web_search", [{
+        "type": "string",
+        "name": "keyword",
+        "description": "google search query (分词，尽量保留所有信息)"
+        }],
+    "通过搜索引擎搜索。如果问题需要在网页上搜索(如天气、新闻或任何需要通过网页获取信息的问题)，则调用此函数；如果没有，不要调用此函数。",
+    google_web_search
+    )
+    new_func_call.add_func("fetch_website_content", [{
+        "type": "string",
+        "name": "url",
+        "description": "网址"
+        }],
+    "获取网页的内容。如果问题带有合法的网页链接(例如: `帮我总结一下https://github.com的内容`), 就调用此函数。如果没有，不要调用此函数。",
+    fetch_website_content
+    )
+    question1 = f"{question} \n> hint: 最多只能调用1个function, 并且存在不会调用任何function的可能性。"
+    has_func = False
+    function_invoked_ret = ""
+    if official_fc:
+        func = provider.text_chat(question1, session_id, function_call=new_func_call.get_func())
+        if isinstance(func, Function):
+            # arguments='{\n  "keyword": "北京今天的天气"\n}', name='google_web_search'
+            # 执行对应的结果：
+            func_obj = None
+            for i in new_func_call.func_list:
+                if i["name"] == func.name:
+                    func_obj = i["func_obj"]
+                    break
+            if not func_obj:
+                gu.log("找不到返回的 func name " + func.name, level=gu.LEVEL_ERROR)
+                return provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
+            try:
+                args = json.loads(func.arguments)
+                function_invoked_ret = func_obj(**args)
+                has_func = True
+            except BaseException as e:
+                traceback.print_exc()
+                return provider.text_chat(question1, session_id) + "\n(网页搜索失败, 此为默认回复)"
+        else:
+            # now func is a string
+            return func
+    else:
+        try:
+            function_invoked_ret, has_func = new_func_call.func_call(question1, new_func_call.func_dump(), is_task=False, is_summary=False)
+        except BaseException as e:
+            res = provider.text_chat(question) + "\n(网页搜索失败, 此为默认回复)"
+            return res
+        has_func = True
+
+    if has_func:
+        provider.forget(session_id)
+        question3 = f"""请你用可爱的语气回答`{question}`问题。\n以下是相关材料，请直接拿此材料针对问题进行总结回答，再给参考链接, 参考链接首末有空格。不要提到任何函数调用的信息。在总结的末尾加上1-2个相关的emoji。```\n{function_invoked_ret}\n```\n"""
+        gu.log(f"web_search: {question3}", tag="web_search", level=gu.LEVEL_DEBUG, max_len=99999)
+        _c = 0
+        while _c < 3:
+            try:
+                print('text chat')
+                final_ret = provider.text_chat(question3)
+                return final_ret
+            except Exception as e:
+                print(e)
+                _c += 1
+                if _c == 3: raise e
+                if "The message you submitted was too long" in str(e):
+                    provider.forget(session_id)
+                    function_invoked_ret = function_invoked_ret[:int(len(function_invoked_ret) / 2)]
+                    time.sleep(3)
+                    question3 = f"""请回答`{question}`问题。\n以下是相关材料，请直接拿此材料针对问题进行回答，再给参考链接, 参考链接首末有空格。```\n{function_invoked_ret}\n```\n"""
+    return function_invoked_ret

util/log.py

@@ -1,11 +0,0 @@
-import logging
-from logging.handlers import RotatingFileHandler
-import colorlog
-
-logger = logging.getLogger("QQChannelChatGPT")
-logger.setLevel(logging.DEBUG)
-handler = colorlog.StreamHandler()
-fmt = "%(log_color)s[%(name)s] %(message)s"
-handler.setFormatter(colorlog.ColoredFormatter(
- fmt))
-logger.addHandler(handler)

util/plugin_util.py

@@ -0,0 +1,34 @@
+'''
+插件工具函数
+'''
+import os
+import inspect
+
+# 找出模块里所有的类名
+def get_classes(p_name, arg):
+    classes = []
+    clsmembers = inspect.getmembers(arg, inspect.isclass)
+    for (name, _) in clsmembers:
+        # print(name, p_name)
+        if p_name.lower() == name.lower()[:-6] or name.lower() == "main":
+            classes.append(name)
+            break
+    return classes
+
+# 获取一个文件夹下所有的模块, 文件名和文件夹名相同
+def get_modules(path):
+    modules = []
+    for root, dirs, files in os.walk(path):
+        # 获得所在目录名
+        p_name = os.path.basename(root)
+        for file in files:
+            """
+            与文件夹名（不计大小写）相同或者是main.py的，都算启动模块
+            """
+            if file.endswith(".py") and not file.startswith("__") and (p_name.lower() == file[:-3].lower() or file[:-3].lower() == "main"):
+                modules.append({
+                    "pname": p_name,
+                    "module": file[:-3],
+                })
+    return modules
+    

webapp_replit.py

@@ -0,0 +1,25 @@
+from flask import Flask
+from threading import Thread
+import datetime
+
+app = Flask(__name__)
+
+
+@app.route('/')
+def main_func():
+  content = "<h1>QQChannelChatGPT Web APP</h1>"
+
+  content += "<p>" + "Online @ " + str(datetime.datetime.now()) + "</p>"
+  content += "<p>欢迎Star本项目！！！</p>"
+  return content
+
+
+def run():
+  app.run(host="0.0.0.0", port=8080)
+  
+
+
+
+def keep_alive():
+  server = Thread(target=run)
+  server.start()