Redact a PDF, So the Words Are Actually Gone
A black box drawn over a number is not a redaction. On most free tools the number is still in the file, and anybody can select it, copy it or pull it out with a search. Here the page you mark is flattened, so what was under the bar is not covered, it is gone. And you do not have to find every copy of a price yourself: type it in once and it is barred wherever it appears.
How it works
- Choose the PDF. It is opened on your device and never uploaded.
- Drag across anything that has to come off, or type a price, a name or a phone number and have every copy of it on every page barred at once.
- Download. Every page you marked comes back as a flattened image with the text removed, and every page you did not is untouched.
What you get
- Drag to draw a bar anywhere on the page, as many as you want
- Type a figure in once and it is barred everywhere it appears, on every page
- Undo the last bar, clear a page, or remove one bar with its own X
- Works with a mouse and with a finger, so it can be done on a phone
- Marks stay put when you page around, resize, or turn the phone sideways
- Only the pages you marked are flattened, so the rest keep their searchable text
- 150 or 300 dpi for the flattened pages, and sideways pages come out the right way round
What a black rectangle actually does
In a PDF the words and the drawings are two separate things stacked on top of each other. Drawing a filled rectangle adds one more thing to the stack. It hides the words on screen and it changes nothing about the words, which are still sitting in the file underneath.
Anybody can get them back, and it does not take a specialist. Select across the black bar and copy. Open the file in a reader and use the find box. Save it as text. Drag it into a word processor. The number you thought you covered comes straight out, because covering and removing are not the same operation and almost every free redaction tool only does the first one.
This is not a theoretical problem. Court filings, government reports and corporate disclosures have all been published with the covered text still readable, and the reason is always the same: somebody drew a box and assumed a box was enough.
The four things on contractor paperwork that have to come off
Subcontractor pricing on a pay application. The owner is entitled to see what they are being billed. They are not entitled to see what you pay the framer, and a schedule of values that still carries your sub's line items is a negotiating position you handed over for free.
Social security and tax numbers. A W-9, a certified payroll report and most insurance paperwork carry a full SSN or EIN. Those sheets travel with the packet to a general contractor, an owner and sometimes a lender, and they end up in email threads that get forwarded.
Bank details. A joint check agreement, a lien waiver that names a payment account, and any document produced by your accounting software can carry a routing and account number that nobody in the chain needs.
Home addresses and personal phone numbers. A residential contract reused as a reference in a bid package still has your past customer's address on it. Take it off before the package goes out, not after somebody asks.
Why flattening is the only honest answer in a browser
Genuinely removing a run of text from a PDF means rewriting the content stream of that page, re-encoding the fonts that were in it, and hoping nothing else on the page referenced what was cut. Desktop software that does this well costs money and still gets it wrong on files it did not produce.
Turning the page into a picture avoids the whole problem. The page is drawn once, the bars are painted onto the picture, and the picture becomes the page. There is no text layer left to recover from, because there is no text layer at all.
That is a real cost and it is worth being plain about it. A flattened page cannot be searched, selected or copied from, and it is a larger file than the page it replaced. So only the pages you actually marked get flattened. The rest of the document is copied through untouched and keeps everything it had.
What this does not do
It removes what is under your bars on the pages you marked, and only that. Search will find every copy of a figure that sits in the file as text, which on a document your own software produced is all of them. A page that came off a scanner is a picture of words rather than words, so there is nothing on it to match against and it has to be marked by hand.
So page through the whole document before you download rather than trusting the count. The search tells you how many times it found something, and that number is only ever a count of what was findable.
It is also not a tool for hiding something from a determined opponent in a file you are legally required to produce complete. It is a tool for not handing out information nobody asked for.
One thing you get for free: the file that comes back is built new rather than edited, so the original's document properties, the author and title fields that quietly carry a company name or a username, do not come with it.
Do this last, and keep the original
Redact at the end, after you merge the parts, after you put them in order and after you turn the sideways scans. A flattened page is harder to work with, and there is no reason to make the rest of the job harder than it has to be.
Keep the clean original somewhere you can find it. You cannot undo this, which is the entire point, and the version with the prices in it is the version you will need when the sub disputes their invoice.
If the document is not final, put a DRAFT stamp on it as well. A redacted copy circulating without a label is the kind of thing that gets signed by accident.
Questions people ask
Is the file uploaded?
No. The page is drawn, blacked out and rebuilt in your browser, on your own device. There is no server here that could receive a document.
Can it black out the same number on every page at once?
Yes. Type it into the search box and press the button, and every copy of it in the document gets a bar. That is what stops a price you took off the schedule of values going out in a summary nine pages later. The bars it puts on are ordinary bars, so you can take any one of them off again with the x in its corner.
The search covered more than I typed. Why?
A bar covers the whole run of text your match landed in, so searching for a figure inside a line can take the label with it. That is deliberate. A PDF gives up its text in runs rather than letters, and working out where inside a run a character sits means counting letters and assuming they are all the same width, which is wrong on almost every font. A bar wider than it needed to be is still a redaction. A bar one character too narrow is a leak.
Can the text under the bar be recovered?
Not from the file you download. On every page you marked, the text is not covered, it is gone, because the page is replaced by a picture of itself with the bars already painted on.
Will the whole document stop being searchable?
Only the pages you marked. Pages with no bars on them are copied straight through and keep their text, so a forty page packet with two redacted sheets is still thirty eight searchable pages.
Why is the file bigger than it was?
A page of text is a few kilobytes and a picture of that same page is a few hundred. That is the trade for the text being genuinely removed. Choosing 150 dpi instead of 300 roughly quarters it, and is enough for anything that will be read on screen.
Does it work on a scan?
Yes, and a scan is the easy case since it was already a picture. The bar is painted on and the page comes back the same size it went in.
Can I do this on my phone?
Yes. Drag with a finger the same way you would with a mouse. Pinch to zoom the page first if the thing you are covering is small, because a bar that is slightly short is worse than no bar at all.