Discord Bot

This repository contains a Discord bot, developed using Python, Pycord, and the OpenAI API. The bot showcases an array of AI capabilities through its interaction with OpenAI's advanced models.

Overview

The bot connects to Discord servers and channels, and listens for user messages. Once triggered, it initiates a conversation.

Key features include:

• Natural language conversations enabled by OpenAI's ChatGPT API
• Image captioning through computer vision APIs
• Verification of user permission levels
• Token counting to prevent exceeding API limits
• Integration with Google search
• Web scraping capabilities
• Integration with WolframAlpha
• Execution of Python code within a Docker container
• Generation of images from text prompts
• Moderation of messages to filter inappropriate content
• Management of user timezones

Directory Structure

├── bot.py - Main bot script
├── chat_functions.py - Defines OpenAI chat completion functions
├── prompt.py - Contains helper functions for initiating conversations  
├── strings.py - Contains bot responses and status messages
├── .env - Stores environment variables and API keys (not committed)
├── code_interpreter/ - Docker container for running Python code
│   ├── docker.py - Manages execution of code within Docker
│   ├── Dockerfile - Specifies the Docker container configuration
│   └── requirements.txt - Lists Docker Python dependencies
└── utils/ - Contains utility functions
    ├── check_user_level.py - Verifies Discord user permissions
    ├── counter_tokens_in_conversation.py - Counts tokens in a chat
    ├── exponential_backoff.py - Manages ChatGPT retry logic
    ├── generate_image.py - Generates images with DALL-E 
    ├── get_and_set_timezone.py - Handles user timezone settings
    ├── get_conversation.py - Retrieves conversation context
    ├── google_search.py - Queries the Google search API  
    ├── handle_send_to_discord.py - Sends incremental responses
    ├── image_processing.py - Analyses images with CV APIs    
    ├── moderate_message.py - Filters out inappropriate content
    ├── scrape_web_page.py - Extracts text/data from web pages    
    ├── store_conversation.py - Stores conversation history 
    ├── wolfram_alpha.py - Queries the WolframAlpha API
    └── would_exceed_limit.py - Checks for token count limits

Requirements

Python Packages

The bot requires the following key packages:

pip install openai tiktoken pycord

To install requirements for a specific file:

# For image_processing.py
pip install transformers google-cloud-vision azure-cognitiveservices-vision-computervision 

System Requirements

• Python 3.10
• Docker

The BLIP image captioning model demands substantial resources:

• RAM: 32GB+
• GPU VRAM: 8GB+

For optimal performance, GPU acceleration is highly recommended.

Setup

Installation

1. Clone the repository
2. Execute pip install asyncio jsonlib os-sys random2 regex sqlite3 threading aiofiles pytz aiohttp cachetools discord.py openai tiktoken python-dotenv google-api-python-client
3. Add your API keys to .env
4. Start the bot by running python bot.py

Trigger the bot by typing "byte".

Usage

The bot initiates a conversation when it hears "byte" or receives DMs.

Key commands include:

• !clear - Clears conversation history (in servers, only users with the "Bot Manager" role can run this)
• !timezone - Sets your timezone

Commands exclusive to admins:

• !unblock - Allows the bot to respond

image_processing.py

This script manages the generation of image captions using various computer vision APIs:

• Google Cloud Vision
• Azure Computer Vision
• BLIP

Each API operates in a separate thread to generate multiple captions that are then merged into the final result.

The get_detailed_caption function orchestrates the API calls and combines the results.

Key functionalities include:

• Image download from URL
• Generation of multiple captions by the BLIP thread
• Extraction of labels, text, etc. by Google Vision
• Provision of alternate tags and descriptions by Azure Vision
• Compilation of results into a single object

Customization

• Modify strings.py to add new responses
• Add API integrations in the utils/ directory
• Alter moderation rules in moderate_message.py
• Substitute image generation models