Dangerzone — Convert Risky Documents to Safe PDFs

Introduction

Dangerzone takes potentially dangerous documents — PDFs, Office files, images — and converts them into safe PDFs by rendering each page as pixels inside a sandboxed container, then reassembling the pixel data into a clean PDF. Developed by Freedom of the Press Foundation, it protects journalists and activists who handle untrusted documents.

What Dangerzone Does

• Converts PDFs, DOCX, XLSX, PPTX, ODP, and images into safe PDF output
• Renders each page as raw pixels inside an isolated container to strip all active content
• Reassembles sanitized pixel data into a clean, safe PDF document
• Optionally applies OCR to make the output PDF searchable
• Runs without network access in the sandbox to prevent data exfiltration

Architecture Overview

Dangerzone uses a two-stage container pipeline. The first container converts the input document to raw RGB pixel data using LibreOffice or other renderers, stripping macros, scripts, and embedded objects. The second container takes the pixel data and reconstructs a clean PDF, optionally running Tesseract OCR. Containers run via Docker or Podman with no network access.

Self-Hosting & Configuration

• Available as native packages for Windows, macOS, and Linux (Debian, Fedora)
• Requires Docker Desktop (Windows/macOS) or Podman (Linux) for container isolation
• No server component needed; runs entirely on the local machine
• Configure OCR language and output quality through the GUI or CLI flags
• Supports batch conversion of multiple files via the command-line interface

Key Features

• Container-based isolation ensures malicious payloads never reach the host system
• No network access in sandbox prevents phone-home or exfiltration attempts
• Optional OCR preserves text searchability in the sanitized output
• Cross-platform support for Windows, macOS, and Linux
• Handles diverse input formats including Office documents and images

Comparison with Similar Tools

• Qubes TrustedPDF — Requires Qubes OS; Dangerzone runs on any OS with Docker or Podman
• DKIM/PDF sanitizers — Strip metadata but do not re-render content; Dangerzone converts to pixels
• ClamAV — Scans for known malware signatures; Dangerzone neutralizes unknown threats by re-rendering
• LibreOffice headless — Can convert formats but does not sandbox or strip active content

FAQ

Q: Does Dangerzone modify the original file? A: No. It creates a new safe PDF; the original file is left untouched.

Q: What types of threats does it protect against? A: Embedded macros, JavaScript, exploits targeting PDF/Office parsers, and tracking pixels are all neutralized by the pixel-rendering approach.

Q: Can I use Dangerzone without a GUI? A: Yes. The CLI supports single-file and batch conversion for scripted workflows.

Q: Does OCR work for non-English documents? A: Yes. Tesseract supports 100+ languages; select the appropriate language in settings.

Sources

• https://github.com/freedomofpress/dangerzone
• https://dangerzone.rocks/