CatGPT is a Telegram bot that integrates with OpenAI's api for people who like to use OpenAI in Telegram.

• Multiple topics

• Multiple endpoints

• Powerful and easy to use

• Install from source code

  git clone https://github.com/xingty/CatGPT
  cd CatGPT

  pip:

  pip3 install -r requirements.txt
  
  python3 start.py --config ${config_file}

  pdm:

  ```shell
  pdm install
  pdm run start.py --config ${config_file}
  ```
  

• Install from release files

  Download the latest build at Release

  pip3 install mqgpt-${version}-py3-none-any-whl
  mqgpt --config ${config_file}

• /key {access key}

  The access key is the key configured in the config file config.json

  You will be enrolled into the system if a correct access key is supplied

• /list

  Display all topics in the current chat window. This indicates that the chat histories for Private Chats, Channels, and Groups are kept separate.

  

  You can click a button to view the next step you can take.

  

  As you see, you can share, download, or delete the topic on which you clicked

• /new [title]

  Create a new topic with the given title.

  For example, using /new english learning will create a topic with the title english learning. A default title will be used if you do not supply a specific title.

  

• /topic [share | download | iv | inner | title]

  There are three optional operations that you can use to perform some task quickly.

  • share

    Use /topic share to share the topic directly without a confirmation operation

  • download

    Use /topic download to download the topic's content directly without a confirmation operation. The content will be encoded as Markdown file

  • iv

    Display the chat history of this conversation in the form of Instant View even if the text length does not overflow.

    

    • inner

      Display the chat history of this conversation in the form of Telegram Message.

    • other characters

      Any other characters will be treated as a title, and the topic's title will be updated."

​

​ if none of the operations is supplied, it displays the chat history.

​

• /profile

  Display user's profile (including the current endpoint, model, and topic)

• /endpoint [endpoint name]

  Display all endpoints.

​ You can also swtich to an endpoint directly with the command /endpoint [your endpoint name]

• /model [model name]

  Display all the support models of the endpoint.

  Use /model [model name] to switch to it directly. Otherwise, the list of models will be displayed in the chat window.

  There are some short names that map to the full names of those models

  {
      "gpt4": "gpt-4",
      "gpt4_32k": "gpt-4-32k",
      "gpt4_16k": "gpt-4-16k",
      "0314": "gpt-4-0314",
      "0613": "gpt-4-0613",
      "1106": "gpt-4-1106-preview",
      "0125": "gpt-4-0125-preview",
      "0409": "gpt-4-turbo-2024-04-09",
      "gpt4o": "gpt-4o",
    	"4o": "gpt-4o"
  }

  For example, you can use /model 0409 to switch to the model gpt-4-turbo-2024-04-09 directly.

  

  It displays all the models if a invalid short name is provided.

• /revoke

  Removes the latest messages from the chat window and chat histories so that you can have another attempt.

• /clear [history | all]

  Clears all the messages of the topic

  • history: Only clear the chat history, messages are remained in the chat window.
  • all: Clears both chat history and messages.

• /respond [y|yes|n|no]

  if enable, the bot will respond to all the group messages.

  This command is only granted to the group owner or adminstrators.

usage: mqgpt [-h] --config CONFIG [--preset PRESET] [--db-file DB_FILE]

options:
  -h, --help       show this help message and exit
  --config CONFIG  Config file
  --preset PRESET  preset file
  --db-file DB_FILE  database file

• tg_token: the token of your telegram bot

• access_key: access key

• proxy: a http or https proxy server. the chatbot will run behind the proxy server if it's supplied

• respond_group_message: If true, the bot will respond to group messages even if it is not mentioned. default: false, can be changed at runtime using the command /respond in groups.

• topic_preview_type: TELEGRAPH or INTERNAL

• endpoinds: your endpoints

  endpint:

  • name: endpoint name
  • provider: openai or gemini
  • api_url: api url. e.g. https://api.openai.com/v1
  • secret_key: secret_key
  • models: list of supporting models of this endpoint
  • generate_title: If true, the endpoint will be used to automatically generate titles for topics that lack one, based on their chat history."

• share: a share provider(only supports github currently)

  • name: require, an identification of your provider
  • type: github
  • repo: the repo name of your github accout
  • owner: github username
  • token: your github personal token

{
  "tg_token": "YOUR_TELEGRAM_BOT_TOKEN",
  "access_key": "Specify Access Key to use this bot",
  "proxy": "http://proxy:port",
  "respond_group_message": false,
  "topic_preview_type": "INTERNAL or TELEGRAPH",
  "share": [
    {
      "name": "notes",
      "type": "github",
      "repo": "name of your repo",
      "owner": "github username",
      "token": "your github personal token"
    }
  ],
  "endpoints": [
    {
      "name": "endpoint_1",
      "api_url": "https://api.openai.com",
      "secret_key": "YOUR_API_KEY",
      "provider": "openai",
      "models": [
        "gpt-4o",
        "gpt-4-turbo-2024-04-09",
        "gpt-4-0125-preview",
        "gpt-4-1106-preview",
        "gpt-4"
      ]
    }
  ]
}

You can specific your own preset file.

[
  {
    "role": "ai",
    "prompt": "you are a helpful assistant."
  },
  {
    "role": "translator",
    "prompt": "you are a translator."
  }
]

Each topic in a group has its own profile, it means that all configuration in topics are isolated.

1. Talk to BotFather
2. Use the command /newbot to create a new bot
3. Follows the guideline of BotFather and then get the access token of your bot

The idea of this project was inspired by YiHong0618's tg_bot_collections, thanks a lot for YiHong0618.

爱发电：

MIT