Ghostscript PDF/A conversion fails: "unable to convert document" on a signed contract

TL;DR: "Unable to convert document" is the wrapper's generic message - Ghostscript's real error is on stderr. Re-run it with stderr visible to see the actual failure, and expect it to be the signature: signed PDFs use incremental updates that the PDF/A writer cannot faithfully rewrite. Convert an unsigned working copy instead, and keep the signed original as the legal record.

unable to convert document

1. Re-run the conversion with stderr unhidden: gs -dPDFA -dBATCH -dNOPAUSE -sDEVICE=pdfwrite -sOutputFile=out-a.pdf signed.pdf Expected: stderr shows the real PostScript error (often around incremental updates or permissions).

2. Check whether the file is signed or encrypted: qpdf --check signed.pdf Expected: signature dictionaries or encryption flags - the usual blockers.

3. Render an unsigned working copy: gs -dBATCH -dNOPAUSE -sDEVICE=pdfwrite -sOutputFile=working-copy.pdf signed.pdf Expected: a visually identical PDF with the signature structures flattened away.

4. Run the PDF/A conversion on the working copy. Expected: conversion succeeds and a validator confirms PDF/A compliance.

5. Archive both: the signed original as the legal record, the PDF/A copy as the preservation copy, linked in your metadata.

Use this when

• Ghostscript (or a tool wrapping it) fails PDF/A conversion only on signed contracts
• the error message is generic with no PostScript details
• unsigned contracts convert fine with the same command

Not for this skill when

• Ghostscript is not installed or not found (fix the install first)
• the real stderr error is about missing fonts (embed the fonts instead)
• every PDF fails conversion (check the Ghostscript version and command syntax)

Variant phrasings

ghostscript pdf/a conversion signed pdf fails

ghostscript unable to convert document pdfa

convert signed pdf to pdf/a workaround

Why it happens

Digital signatures live in incremental updates appended to the PDF. The pdfwrite device rebuilds the file from scratch for PDF/A, and it cannot preserve signature structures through that rewrite - so it fails (or the wrapper gives up and prints the generic message). The content is fine; the signature container is what breaks.

Edge cases

• Never present the unsigned PDF/A copy as the signed original - the signature is gone.
• Missing-font errors look similar at first glance; always read stderr before assuming it is the signature.
• PDF/A-1 forbids transparency and layers that A-2 and A-3 allow - pick the right conformance level for the source.
• Some signers lock the file against any processing; the print-to-PDF path through a viewer is the fallback.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst__jyMVAU3XfwXlJgAbkAyoA