How I Built a Free Invoice OCR That Runs 100% in Your Browser (Open Source)

#opensource #react #javascript #ocr

Every expense report tool I've tried either required uploading sensitive invoices to someone else's server or couldn't handle Chinese invoices properly. So I built a browser-native solution.

Ticket-Check-Bro is an open-source, client-side invoice parser that extracts structured data from PDFs and images — all within the browser, no server upload.

What's New in v1.4.1

Multi-page PDF: reads ALL pages, not just the first 3
Multi-format export: Excel (.xlsx) / CSV / JSON — amount fields exported as raw numbers
Search & filter: find files by name, invoice number, or amount
Summary dashboard: total amount + document type breakdown
Confidence indicators: per-field confidence scores, low-confidence fields highlighted

Key Features

8 document types: Chinese VAT invoices, train tickets, flight itineraries, taxi receipts, fixed-amount receipts, toll invoices, vehicle invoices, and English invoices
Dual engine: PDF text extraction (pdfjs-dist) + image OCR (Tesseract.js WASM)
PDF merge with per-page selection
Batch rename with template variables → ZIP download
Drag-and-drop sort for file ordering
i18n: English & Chinese UI
Dark mode with system preference detection
PWA: installable, works offline
Keyboard shortcuts: Ctrl+O / Delete / Ctrl+E
Session persistence: auto-save/restore via localStorage