Live preservation check / XLSX / DOCX / PPTX

What survives a save? Check it on your own file.

An agent that edits an Office file has to open it and save it again. Everything it did not mean to touch should come back unchanged: comments, tracked changes, charts, custom XML, document properties. This page saves your file with WolfSDK and with the tools agents usually reach for, then compares every part with the original.

Run a check Get the checker, verify.py

WolfSDK

Compared with

Edit baseline

openpyxl, python-docx, python-pptx

The library an agent usually picks for each format, used the careful way.

Your file

Deleted after the run

Results stay downloadable for a short time. Names and contents are not logged.

01 / Choose a file

Pick a sample or drop in your own.

Each run saves the file twice: once with no edit, once replacing one piece of text.

Drop an .xlsx, .docx or .pptx here or choose a file, 20 MB max

Find (literal text in the file)

Replace with

How to read this

Every part, compared byte for byte.

An Office file is a ZIP package of parts. Each part of a result is compared with the same part of the original after decompression.

expected

The operation had a reason to write it.

The part holding the replaced text, or, for a spreadsheet save, formula cells that recalculate.

rewritten, same content

Different bytes, same document.

XML that canonicalizes to the same document, relationship entries in another order, or a worksheet whose cell values are all unchanged.

unrelated change

Anything else.

A part removed, added, or altered that the operation had no reason to touch. This is the number at the top of each result.

Identical

Not listed.

Parts that come back byte-identical are counted, not shown. A whole file can come back identical.

Check it yourself

Download any result and run the same comparison on your machine with verify.py, which needs only the Python standard library. Recount what is inside a file with inventory.py, which sits next to it.

Each result lists the exact command behind it. The LibreOffice column reruns anywhere LibreOffice is installed, and the Python-library column reruns with baselines.py (pip install openpyxl==3.1.5 python-docx==1.2.0 python-pptx==1.0.2).

python3 verify.py roundtrip original.docx result.docx
python3 verify.py edit original.docx result.docx \
  --find "30 days" --replace "45 days"

Run it on your document set. WolfSDK is in private beta for agent builders.

For the binary, the Python SDK, or a run across your own files, email [email protected].

Get beta access Visit wolfsdk.com