[TOC]
Using ChatGPT-weBot based on ChatGPT(API key call), Stable Diffusion AI drawing and official WeChat hook interface. 中文文档 | English
cover created from DALL·E3 (openai.com)
- Support conversation
- Support context-aware question answering
- Support multithreaded
Stable Diffusion
AI drawing (English Only, Support (Negative) Prompt) - Never get banned by using official WeChat execution
- Support API calls for
gpt-3.5-turbo
and newer models - Support
WebChatGPT
function - Support bot's character setting
- Set the keywords to wake up the WeChat robot in private
- Set the keywords to wake up the WeChat robot in the group
- Support replying at-message when mentioning your bot in the group
- Get help doc inline
- Regenerate conversation
- Rollback conversation
- Conclusion (save
token
consumption) - Reset the whole conversation
- Support multithreaded conversation in one account
- No need to manually reboot service after error exists
- Other
---> Configurable options detailed guide
-
Environment: Windows 7+, python 3.7+
-
Install all packages listed in
requirements.txt
, use the command like:pip install -r ./requirements.txt
Note that v1.2 requires more packages to be installed and upgraded, so please execute this command once after upgrading.
-
Download package from Github Releases. (You can download it step by step when they are mentioned)
-
Install
WeChat-3.9.5.81.exe
on your computer, if your version is higher than 3.9.5.81, you can downgrade instantly, or install seperately in other directory. Afterwards, please start it as an administrator and log in. **If you want to dual-open WeChat, you need to install two different versions and modify./dual-start.bat
according to the comments **, the subsequent steps are slightly different, please continue to read [here](. /doc/Dual_Start.md). -
Monitoring WeChat message by running a server. It has been modified to 1 solution after version V1.20:
> cd .\wxinject\bin\ > .\injector.exe -n WeChat.exe -i .\wxinject.dll
-
The last step is fill json files listed in
.config/
.-
In
api_config.json
, you need to fill in your own parameter settings for API calls. If you don’t know the specific parameters, you only need to fill in the "api_key" and optional "proxy" items. -
In
server_config.json
, you can customize the listening address and port. If you don’t know it exactly, no modification needed by default. -
In
config.json
, you need to configure your custom options based on your preferences. -
In
sys_character.json
, you can customize the character the bot needs to play, and use the command to activate when chatting.
-
-
Run
main.py
by using command:python main.py
Everything is ready, feel free to go online with your ChatGPT-weBot !
No limitation, but since switching to OpenAI API, there are usage counts and payment requirements.
-
How to get all response? You can say "continue" in your language.
-
Have problems? Feel free to create an issue.
-
How to trace problems in multithreaded program? Print or using debug with information of thread-stack.
-
Have any preview images related to functionality? Yes, go to -> Preview
-
Wanna buy me coffee? Thank you, qrcode is as follows.
- 2023.10.23 Version v1.20.2 is released, fix some memory leak and change XML parser.
- 2023.10.18 Version v1.2 is released, with new SDK and new WeChat version.
- 2023.10.17 Complete the Python ws_client, update the source code to support the new SDK, and prepare for main architecture upgrades
- 2023.10.15 A large number of APIs are added and type-testing is conducted
- 2023.10.10 Build and improve the SDKs and support the ws client
- 2023.3.24 Exception handler updates and other pull requests from rogue-shadowdancer and wbbeyourself
- 2023.3.23 Fix bugs for Stable Diffusion and others, released v1.01 version
- 2023.3.21 Add plenty of new features, fixed bugs, released v1.00 version
- 2023.3.4 Add Stable Diffusion into function (English Only)
- 2023.3.3 Add multithread and rewrite the whole program structure
- 2023.2.27 Add zip version of WeChat and
dual-start.bat
, fix the bug that prevents other operations when the response keyword is empty - 2023.2.25 Add the option in
config.json
to quote the original question before answering - 2023.2.25 Complete all API function on features and Debugs for errors
- 2023.2.23 Accomplish some API listed on features
- 2023.2.23 Fix streaming issue when connecting to reverse server
- 2023.2.21 Report issue on ChatGPT API
- 2023.2.20 v0.90-dev released, for basic ChatGPT API usage on WeChat
- 2023.2.17 Start to develop the whole process