I'm building a self-hosted workflow for sensitive documents, what PDF API or library makes sense?

For sensitive documents and a self-hosted setup, the best choice usually depends on whether you need rendering/viewing, text extraction, editing/merging, or full conversion.

Good self-hosted options

1) Apache PDFBox (Java)

Best if you want a solid, mature open-source library for:

• reading PDFs
• extracting text
• merging/splitting
• form handling
• basic PDF creation

Pros

• Fully self-hosted
• No external service
• Mature and widely used
• Good for document workflows

Cons

• Java-based
• Not the best for high-fidelity rendering or complex conversions

---

2) iText 7 (commercial/open-core)

Best if you need advanced PDF generation/manipulation.

Pros

• Very powerful
• Excellent for creation, stamping, signatures, forms, encryption
• Good documentation

Cons

• License matters a lot
• Open-source use is limited; commercial license often required for proprietary use
• Not ideal if you want “free and simple”

---

3) Apache Tika + PDFBox

Best if your workflow is about extracting and indexing document content.

Pros

• Great for text extraction and metadata
• Self-hosted
• Useful in document pipelines

Cons

• Not for editing/rendering PDFs
• More of a processing stack than a single PDF API

---

4) qpdf (CLI/library)

Best for low-level PDF operations like:

• linearization
• encryption/decryption
• splitting/merging
• structural fixes

Pros

• Very reliable
• Self-hosted
• Great for backend automation

Cons

• Not a high-level document API
• Limited creation/editing features compared to PDFBox/iText

---

5) Ghostscript

Best if you need PDF/PostScript conversion or rendering pipelines.

Pros

• Powerful
• Common in document-processing systems

Cons

• Can be tricky to secure and sandbox
• More conversion-oriented than application-level PDF handling

---

6) LibreOffice headless

Best for Office-to-PDF conversions self-hosted.

Pros

• Excellent for DOCX/XLSX/PPTX → PDF
• Fully self-hosted

Cons

• Heavy
• Not a PDF API
• Needs sandboxing and process isolation

---

If you need a web viewer/editor

If your “workflow” includes end users viewing PDFs in browser:

PDF.js

• Best open-source browser PDF viewer
• Can be self-hosted
• Great for secure document viewing in a web app

---

My practical recommendation

If you’re building a backend workflow:

• Java stack: PDFBox + qpdf
• Python stack: pypdf / PyMuPDF for operations, plus LibreOffice for conversion
• Need advanced PDF creation/signing: iText 7 if license is acceptable
• Need document text indexing: Apache Tika + PDFBox
• Need browser viewing: PDF.js

---

For sensitive documents specifically

A self-hosted workflow should also include:

• store files encrypted at rest
• isolate conversion/rendering in containers or VMs
• disable outbound network access for PDF-processing workers
• use temporary file cleanup
• validate untrusted PDFs carefully
• log access, not content

---

Short answer

If you want the safest, most general self-hosted choice:

• PDFBox if you’re in Java
• PyMuPDF or pypdf if you’re in Python
• PDF.js for viewing
• qpdf for backend PDF hygiene

If you tell me your language/runtime and what you need to do with the PDFs (view, extract, sign, redact, convert, generate), I can recommend a tighter stack.