Reality Unbound

What is this?

A text-based adventure powered by OpenAI's GPT 3.5-Turbo model. Every adventure is new and different; even the same choices entail a different story each time. Immerse and lose yourself in an endless narrative that could be curated by your dreams. Very lightweight with a clean, modern, and user-friendly interface.

Setup 📦

Local Machine Installation (Recommended)

Setup is easy, just follow the steps:

1. Download Python (v3.11.4+) from here. To check if you have it installed, run the following command

   $python --version
   

2. Fork and clone the repository
3. Run the following command to install all libraries

   $pip install openai customtkinter tk pillow
   

4. (OPTIONAL) Create a tokens.py with the following function:

   def openAIToken():
       return "sk-..." # Enter your token in the string

   Grab your token from here.
   If this step is omitted, it will prompt for a token once the app is launched. The entered token is stored locally only for the duration of the session
5. Open cmd and change the directory to the repository folder, then run main.py using:

   $python main.py
   

Replit Installation (Not recommended)

Currently, due to an issue with the Nix environment, there is unexpected behaviour on the Replit installation; there are some known issues with the Replit version:

1. The token request window does not appear if tokens.py is omitted
2. Popup windows not capturing priority
3. Can not clone a chat, you must create a new adventure instead

It is highly advised to use the Local Machine installation instead! Following are the steps to run it on Replit:

1. Fork the repository and ensure it is public
2. Go to replit, log in, and press Create Repl
3. Import from GitHub
4. Run the following command in the shell to install all dependencies

   $pip install openai customtkinter tk pillow
   

5. Create a tokens.py with the following function:

   def openAIToken():
       return "sk-..." # Enter your token in the string

   Replit will automatically ask you to turn that into a secret. Press yes, name the secret 'OpenAI Token', and refactor the code:

   import os
   
    def OpenAIToken():
    	return os.environ['OpenAI Token']
   

   Grab your token from here.
   If this step is omitted, the program will refuse to open.
6. Press on Run to test out the program :)

Customization 🎨

There are a few things that can not be changed from the app and must be changed from within the code (found in main.py).

    mock_data: boolean

Whether the application should make API calls or use mock data.
False (default) means it uses API calls.

    story_length_choices: int

This is the number of choices the GPT model is instructed to contain in the story.
This value is only a suggestion to the language model, and should not be used as a hard cap by any means.
Default is 10

    story_length_minutes: int

This is the number of minutes the GPT model is instructed to make the story.
This value is only a suggestion to the language model, and should not be used as a hard cap by any means.
Default is 2

    MODEL: str

This value determines the model that will be queried for a response. A list of models can be found here
Default is "gpt-3.5-turbo"

Known Bugs 🪰

These are some bugs I encountered during development and will look into

• Replit does not output Token Request window. Presumably because the App does not get loaded until after a token has been received
• Window freezes when making API calls
• If selected chat is deleted, it stays on screen
• Duplicate chats clone new messages
• Chats do not right align properly sometimes - keeping window axis X locked for now
• TypeError with NoneType is sometimes raised when working with mock data