FastAPI application for processing videos and images with LLM integration using OpenRouter.

aidit/
├── app/                          # Main application package
│   ├── __init__.py
│   ├── main.py                  # FastAPI application
│   ├── database.py              # SQLAlchemy models
│   ├── video_index.py           # VideoIndex class (LLM API calls)
│   ├── video_processor.py       # VideoProcessor class (video splitting)
│   └── video_index_processor.py # Queue management
├── docs/                         # Documentation
│   ├── README_API.md            # Original API documentation
│   └── README_VIDEO_API.md      # Video API documentation
├── examples/                     # Example scripts and old code
│   ├── wow.py                   # Original example script
│   ├── api.py                   # Original API implementation
│   └── chat_interface.html      # HTML interface
├── run.py                        # Application entry point
├── requirements.txt              # Python dependencies
├── .gitignore                    # Git ignore rules
└── README.md                     # This file

1. Install dependencies:

pip install -r requirements.txt

2. Set environment variables: Create a .env file:

OPENROUTER_KEY=your-api-key-here
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_KEY=your-supabase-service-role-key
VIDEODB_API_KEY=your-videodb-api-key  # Optional, for scene indexing

3. Initialize database tables:

   • Option 1: Run the SQL in schema.sql in your Supabase SQL Editor
   • Option 2: Use the init script to check table status:

   uv run python init_db.py check

4. Run the application:

python run.py

Or with uvicorn directly:

uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

• POST /process/image - Process a single image from URL

• POST /process/video - Process a video (downloads, splits, processes frames)
• GET /status/{video_id} - Get processing status
• GET /status/{video_id}/frames - Get all frames with results

• GET /detect/errors/{video_id} - Detect and analyze errors
• GET /detect/conflicts/{video_id} - Detect conflicts and provide resolution

• ✅ Background processing with queue system
• ✅ Video splitting by granularity (seconds) using OpenCV
• ✅ Base64 conversion for efficient image transmission
• ✅ SQLAlchemy database tracking
• ✅ Configurable prompts per request
• ✅ Comprehensive logging
• ✅ Error detection and conflict resolution

See docs/README_VIDEO_API.md for detailed API documentation.

The application uses a modular architecture:

• VideoIndex: Handles LLM API calls (OpenRouter)
• VideoProcessor: Downloads and splits videos using OpenCV
• VideoIndexProcessor: Manages background queue and processing pipeline
• Database: SQLAlchemy models for tracking videos, frames, and logs

The application is organized as a Python package in the app/ directory. All core functionality is modular and can be imported:

from app.video_index import VideoIndex
from app.video_processor import VideoProcessor
from app.video_index_processor import VideoIndexProcessor