ChatGPT Python SDK

Library that allows developers to easily integrate the ChatGPT into their Python projects.

Install or update

pip install -U chatgpt

Config

Create a file with your credentials

Create the file config.json in your working directory:

{
    "email": "[email protected]",
    "password": "xxx"
}

With proxy

{
    "email": "[email protected]",
    "password": "xxx",
    "proxy": "socks5://user:pass@host:port"
}

With other parameters

{
    "email": "[email protected]",
    "password": "xxx",
    "timeout":300,
    "cache_file_path":"/path/filename",
    "access_token_seconds_to_expire":1800
}

Environment variable

You can specify the default configuration folder for chatgpt by setting the CHATGPT_HOME environment variable to the desired directory path.

export CHATGPT_HOME="/home/$USER/.config/chatgpt"

Usage

CLI

You can launch the CLI with:

chatgpt

or

python -m chatgpt

These are the available commands:

• reset: forget the context of the current conversation.
• clear: clear the terminal.
• exit: exit the CLI.

SDK

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import sys
from chatgpt import Conversation

conversation = Conversation()

# Stream the message as it arrives.
for chunk in conversation.stream("We are going to start a conversation. I will speak English and you will speak Portuguese."):
    print(chunk, end="")
    sys.stdout.flush()

# Wait until the message is fully received.
print(conversation.chat("What's the color of the sky?"))

# The AI will forget it was speaking Portuguese
conversation.reset()
print(conversation.chat("What's the color of the sun?"))

it is recommended to use stream instead of chat.

Exceptions:

from chatgpt import ChatgptError, ChatgptErrorCodes

try:

    for chunk in conversation.stream("Hello, world!"):
        print(chunk, end="")
        sys.stdout.flush()

except ChatgptError as chatgpt_error:

    message = chatgpt_error.message
    code = chatgpt_error.code

    if code == ChatgptErrorCodes.INVALID_ACCESS_TOKEN:
        print("Invalid token")

Chatgpt Error codes:

• INVALID_ACCESS_TOKEN: This error code indicates that the access token provided to the chatbot's API is not valid or has expired.
• CHATGPT_API_ERROR: This error code indicates that an error has occurred while making a request to the chatbot's API.
• CONFIG_FILE_ERROR: This error code indicates that there is a problem with the configuration file for the chatbot.
• UNKNOWN_ERROR: This error code is used when the cause of the error is unknown or cannot be determined.
• LOGIN_ERROR: This error code indicates that there was a problem with the login process, such as an incorrect username or password.
• TIMEOUT_ERROR: This error code indicates that a request to the chatbot's API has timed out.
• CONNECTION_ERROR: This error code indicates that there is a problem with the connection to the chatbot's API.