naptha/tesseract.js — reverse-engineered prompt

Build me a simple OCR tool using Tesseract.js. I want to drop or upload an image, choose a language if possible, then have the app pull the text out of the image and show it in a clean text box that I can copy from.

Please make it work in the browser, without needing a server for the actual recognition. Show a clear loading or progress state while the image is being processed, and let me run OCR on more than one image without restarting everything each time. If there’s an error or the file type isn’t supported, show a friendly message instead of crashing.

Keep the UI simple, like a practical demo page. It should have an image preview, extracted text, copy button, and maybe a reset button. Don’t add PDF support because this library doesn’t handle PDFs. Use the current Tesseract.js docs if you need to check the newest worker setup.