Excise PDF

Redacting a PDF without a full editing suite.

Most people arriving at this question are not shopping for software. They have one document going out the door, and they want to know whether the black boxes on it are real — without buying a suite to find out. That is answerable, and you can answer it yourself.

Two operations that look identical on screen

A PDF page is a program: a content stream that says draw this glyph here, this line there. Every tool that puts a black rectangle on a page is doing one of two very different things.

• Drawing over the text. A rectangle is added — as a shape on the page, or as an annotation floating above it. The characters underneath are untouched. They are still selectable, still extractable, still one copy-paste from the person they name. If the rectangle is an annotation, it can also simply be deleted.
• Editing the page's contents. The content stream is rebuilt without those characters, so there is nothing underneath to find. Anything drawn afterwards is decoration on an already-empty spot.

On screen these produce the same picture. That is the entire problem, and it is why the question “is this redacted?” cannot be answered by looking at the page.

The check, in about a minute, in whatever tool you have

This works on any PDF from any program, including ones we have never seen. Do it on the output file, not the original.

1. Select across the black box — drag from just before it to just after it — then copy, and paste into a plain text editor. If the words come with it, they are still in the file.
2. Search the document for a name or number you thought you removed. A hit inside a redacted region is the same finding.
3. Open the document properties and read the title, author, subject and keywords. Names get left there routinely, and nothing you do to a page changes them.
4. Open the bookmarks or outline panel. Bookmark titles are written by whoever built the document — “Smith deposition” is a real bookmark title on a real file.
5. Click any form field that is still interactive and read what is in it. A field's value is stored separately from the page.
6. Check for attachments. A PDF can carry whole files inside it, and they travel with the document.

Steps 3 to 6 are the ones people skip, and they are where the surprises live: the page can be spotless while the file still names the person.

What Excise PDF does about each of those

This site's whole argument is that the check above is the product, not a marketing claim — so the app runs it for you, on the file it wrote, and prints the result.

• The text is removed from the page's content stream rather than covered, and the mark drawn afterwards is a page object, never an annotation.
• Twelve categories of hidden data — document info, XMP metadata, JavaScript, open and additional actions, hidden layers, embedded files, form fields, annotation actions, annotations, bookmarks, and page thumbnails — are removed by default. Anything you keep is reported as kept.
• The written file is reopened and read back through two channels — every page's text, and every object in the document graph — and the receipt states what was inspected before what was found.
• Your original is never touched. Output is always a separate copy, and the app refuses to write over its input.

Before you decide it fits

• macOS, iPhone and iPad only. There is no Windows, Linux or web version.
• No command-line tool, no API, no server, no watched folders. Mac batch mode is a folder you pick and press a button on.
• Scanned pages stop the document by name unless you ask for OCR, which runs on device and is reported as the weaker check it is.
• It certifies nothing. It reports what it inspected in one file, under the settings of that run — and it gives no legal advice.

Read next