VeilPDF

The problem nobody talks about

Copy text from an academic PDF. Paste it. Get $%&@#^*~.

That's not a bug. It's intentional — called font obfuscation.
Publishers and teachers deliberately break Unicode mappings so
copy-paste returns garbage, while the PDF looks visually perfect.

No clean open-source tool existed to fix this. So I built one.

What VeilPDF does

recover — detects broken ToUnicode tables, reconstructs text,
falls back to OCR automatically.

protect — lets YOU obfuscate your own PDFs, inject invisible
ownership watermarks, encrypt with AES-256.

forensics — detects fake redactions (black boxes with text
still underneath), font fingerprints, hidden metadata.

ai — summarize and extract structured data via Ollama.
Fully offline. No API keys. No cloud.

The stack

Rust CLI → orchestrates everything, single binary

Python core → pypdf, pdfminer, tesseract

Zig crypto layer → AES-256, byte-level watermark injection

One command setup:

[github](https://github.com/Gaurav-x111/veilpdf)
cd veilpdf
./veilpdf   # auto-installs everything

Zero cloud. Zero cost. Runs entirely on your machine.

GitHub: https://github.com/Gaurav-x111/veilpdf

Feedback welcome — especially on the Zig crypto layer
and OCR fallback heuristics.