ChatGPT

Reverse Engineered ChatGPT by OpenAI. Extensible for chatbots etc.

Disclaimer

This is not an official OpenAI product. This is a personal project and is not affiliated with OpenAI in any way. Don't sue me

This is a library and not intended for direct CLI use

The CLI functionality is for demo and testing only.

@rawandahmad698 has a much better CLI tool at

PyChatGPT

Urgent help needed

• Writing tests
• More verbose error handling
• Decrecate bs4 in favor of pure requests

Wiki is now open to all contributors

Features

• No moderation
• Programmable

Setup

Install

pip3 install revChatGPT --upgrade

Email and password authentication

{
"email": "<YOUR_EMAIL>",
"password": "<YOUR_PASSWORD>"
}

Save this in config.json in current working directory

Legacy authentication or if using Google Auth

Go to https://chat.openai.com/chat and log in or sign up

1. Open console with F12
2. Open Application tab > Cookies
3. Copy the value for __Secure-next-auth.session-token and paste it into config.json.example under session_token. You do not need to fill out Authorization
4. Save the modified file to config.json (In the current working directory)

Running

 $ python3 -m revChatGPT ChatGPT - A command-line interface to OpenAI's ChatGPT (https://chat.openai.com/chat)
Repo: github.com/acheong08/ChatGPT
Arguments: Use --stream to enable data streaming. (Doesn't work on MacOS)
Type '!help' to show commands
Press enter twice to submit your question.
You: !help
!help - Show this message
!reset - Forget the current conversation
!refresh - Refresh the session authentication
!exit - Exit the program

Refresh every so often in case the token expires.

Development:

pip3 install revChatGPT --upgrade

fromrevChatGPT.revChatGPTimportChatbotimportjson# Get your config in JSONconfig= {
"email": "<YOUR_EMAIL>",
"password": "<YOUR_PASSWORD>"
}
chatbot=Chatbot(config, conversation_id=None)
chatbot.reset_chat() # Forgets conversationchatbot.refresh_session() # Uses the session_token to get a new bearer tokenresp=chatbot.get_chat_response(prompt, output="text") # Sends a request to the API and returns the response by OpenAIresp['message'] # The message sent by the responseresp['conversation_id'] # The current conversation idresp['parent_id'] # The ID of the response# This returns a stream of text (live update)resp=chatbot.get_chat_response(prompt, output="stream") formessageinresp: # You have to loop through the response streamprint(line['message']) # Same format as text return type
...

This can be imported to projects for bots and much more. You can have multiple independent conversations by keeping track of the conversation_id.

Awesome ChatGPT

My list

If you have a cool project you want added to the list, open an issue.

Star History