vs Dangerzone

Dangerzone is an excellent free tool. It solves a different problem than pdf-defang.

The difference

	Dangerzone	pdf-defang
Approach	Render PDF to images, reassemble	Strip dangerous structures from existing PDF
Output	Visually identical, but flattened to images	Original PDF, with active content removed
Searchable text	Lost (becomes images), or OCR'd back imperfectly	Preserved
Form fields	Lost	Preserved
Bookmarks / TOC	Lost	Preserved
Per-file time	Minutes (container + render + OCR)	Milliseconds
Setup	Docker required, ~1GB image	pip install, ~250KB
CPU/RAM	Significant (full container + Tesseract)	Minimal
GUI	Yes	No (CLI + library)
Library API	No (CLI/GUI only)	Yes (Python)

When to use Dangerzone

Use Dangerzone when:

• The PDF source is highly untrusted (random email attachments, leaks)
• You can tolerate visible content becoming images
• You don't need to preserve form interactivity
• Per-file processing time isn't critical
• You want defense against parser-level vulnerabilities too

When to use pdf-defang

Use pdf-defang when:

• You operate a service that processes user-uploaded PDFs (web app, SaaS)
• You need to preserve visible content (text searchability, forms)
• You're handling thousands of files per day - throughput matters
• You're integrating into a Python codebase (Flask, FastAPI, etc.)
• The threat model is "users uploading PDFs without realising they contain active content" rather than "nation-state APT delivering a targeted exploit"

Use both?

You can. Pattern:

1. Run pdf-defang scan cheaply on every file
2. For files where risk_level == "high", route to Dangerzone for full re-imaging
3. For others, pdf-defang sanitize and ship

This gets you near-zero latency on the 99% of files that are mostly fine and full-paranoia treatment on the 1% that look suspicious.