Verifying that the text is really gone

Any tool can claim it redacted your document. The claim is worth exactly as much as the check behind it, and the check has to run on the file that was produced, not on the editor's idea of what it produced. Here is what a real verification looks like and how to do it yourself.

Check one: extract the text

Reopen the finished PDF and pull the text out of it the way a copy-paste does, decoding each font's own encoding. This is the authoritative check on page content: whatever this returns is what a recipient gets when they select the page and copy. If a redacted term appears here, the redaction did not happen — the text is still in the page. The app runs this automatically on its own output, comparing against the terms you removed, with whitespace ignored and case folded, because extraction routinely splits a name into fragments like "Jo hn" that a naive string search would miss.

Check two: scan the raw streams

Text extraction only reads page content. It does not read the document's metadata, its XMP packet, comments and annotations, attached files, or anything sitting in a stream that is not part of a page. A name can survive in all of those places while the page itself is genuinely clean. So the second check decompresses every stream in the file and searches the bytes for the same terms. It is cruder and it catches what extraction structurally cannot. Neither check substitutes for the other, so the app runs both, and does not present the output as clean unless both come back clean.

Doing it by hand

You do not need a tool to sanity-check a file. Open it in a normal viewer, select the whole page including the black areas, copy, and paste into a plain text editor — if the redacted words appear, stop. Then open the document properties and read title, author, subject and keywords. For a deeper look, any text-extraction utility will dump a PDF's text to the terminal, and searching that dump for the removed term takes one command. It is worth doing once on a file you produced with whatever tool you normally use, just to find out what it actually does.

What verification cannot tell you

It can prove a string is absent. It cannot prove a document is safe. If the redacted name still appears as an image of a signature, or the removed figure can be derived from a subtotal you left in, or the person is identifiable from a job title plus a date plus a location, the file passes every automated check and still discloses. The two searches described here answer one narrow question — is this text still in this file — and answering it is necessary rather than sufficient. The judgement about what to remove stays yours.