bhimrazy/receipt-ocr ? reverse-engineered prompt

Build me a receipt scanner app that can take a photo of a receipt and turn it into useful data I can actually use.

I want one simple command or small web app where I upload a receipt image and get back clean structured JSON with things like merchant name, address, date, time, total, and line items. It should work with an LLM based OCR flow for structured extraction, and also have a basic Tesseract option for plain text extraction if I want the raw OCR output instead. If it helps, include a small FastAPI service and a simple CLI so I can run it locally or call it from Python.

Make it easy to configure with an API key in a .env file, and please add a couple of example receipts plus tests so I can see it working right away. If you need to check current docs for the OCR or API setup, go ahead and look them up online.