Clean up a PDF before you send it

What a PDF carries with it

A PDF is not a photograph of the document: it is a container, and things travel inside it that never show on screen. The author and the company get in by themselves, taken from the installation of Word or of whatever program printed the file; the internal title is very often the file name as it used to be, which can be «offer-client-B-discounted» when the file you are sending is called «offer.pdf».

Then there are the things somebody added: comments and annotations, which in many readers hide in a closed side panel; files attached inside the PDF, which are whole documents nobody remembers; filled-in form fields, with their value inside; and earlier revisions, which stay stuck at the end of the file every time somebody saves over it. The page shows them to you one by one with their value, because «there is something» is no use at all and «Author: Mario Rossi» is understood in a second.

How it gets removed, and why like this

The entries are not deleted from the document: a new one is built and only the pages get copied into it. The reason is technical but worth knowing, because it is the difference between a job done and a job that looks done: in the library that handles PDFs, removing the reference to an object does not remove the object from the file. The comment no longer shows in the reader, but it is still written in there, and anyone opening the file with the right tool finds it.

By building a new document, whatever is not referenced by a page never gets in: away go the metadata, the XMP block, the attachments, the form fields and, with the same move, the earlier revisions stuck at the end too. The text and the layout stay as they were, and the test for this tool goes looking for the secret strings in the raw bytes of the file produced, not in what the reader says.

Two things worth knowing first

Cleaned up does not mean anonymous. The layout, the fonts and above all the content stay, and the content is nearly always the most identifiable thing of the lot: a quote with your prices is still a quote with your prices. This tool removes what travels hidden, not what can be read.

And removing metadata does not remove the text under a black rectangle: if you covered an account number by drawing over it, that text is still selectable and copyable. For that you need something else, which rasterises the page with the rectangle already burnt in: it is Redact a PDF.

Signed PDFs do not get touched

If the document has a digital signature, the page recognises it and stops. Cleaning up means rewriting the file, and rewriting a signed file invalidates the signature: the document would still be perfectly readable but would no longer count as signed, which is exactly the damage that whoever sends a signed PDF does not expect.

Same for PDFs protected by a password: they do not open here. If you know the password, the step to take first is Remove the password from a PDF. And if what you need instead is a lighter file rather than a cleaner one, that is a different job and it lives in Compress a PDF.