Receipt OCR Parser

Browser-based OCR tool for extracting prices and dates from receipts. Built with TypeScript and Tesseract.js.

What it does

• Drag & drop receipt images or upload from filesystem
• Extracts final price and receipt date
• Returns structured JSON output
• Everything runs client-side (no server needed)

Quick start

npm install
npm run build
npm run serve

Then open http://localhost:8000

Or just open index.html directly in a browser.

How to use

1. Drop a receipt image on the page (or click to upload)
2. Click "Process Receipt"
3. Wait for OCR to finish (takes ~5-15 seconds)
4. Get JSON with extracted data

Docker

npm run docker:compose:up

See DOCKER.md for details.

Tech

• TypeScript (vanilla, no frameworks)
• Tesseract.js v5
• Tesseract loaded from CDN

Development

Auto-rebuild on changes:

npm run watch

Or build manually:

npm run build

Testing

Run automated tests on all 7 demo receipts:

npm test

Then open http://localhost:8000/test-receipts.html

See TESTING.md for ground truth values.

Known issues

• OCR is slow on big images (10-20 seconds sometimes)
• Accuracy varies wildly depending on photo quality
• Sometimes grabs the wrong number as the total
• Works better with Finnish receipts than Swedish (more training data in Tesseract)
• Blurry or dark photos usually fail

Privacy

Everything runs in your browser. No data sent anywhere.