toolsmith

No upload · everything runs in your browser

Why won't my PDF open?

"This file cannot be opened" covers at least five completely different problems. Telling them apart takes about ten seconds and saves you from trying the wrong fix.

Updated

First: does it fail everywhere, or only in one place?

Open it in a browser — drag the file onto a browser tab. Browsers have their own PDF engine, so this tells you whether the file is broken or your reader is.

If it opens in the browser but not in your desktop reader, the file is fine. Update the reader, or just use the browser. If it fails in both, keep reading.

It asks for a password — and there are two kinds

PDFs can carry two separate passwords and people rarely realise it. A user password is required to open the document at all. An owner password leaves it readable but restricts printing, copying or editing.

This matters because tools behave differently. A file with only an owner password opens fine in most readers, so it feels unprotected — but many tools will still refuse to modify it, which looks like a broken file when it is really a permission.

There is no honest way around a user password: the content is genuinely encrypted. Our PDF tools refuse encrypted files and say so rather than producing something empty. We tested that with real AES-encrypted files and with ones written by LibreOffice; all four tools declined correctly.

The text is there but shows as empty boxes

This is a font problem, and it is very common with Chinese, Japanese and Korean documents. The characters exist in the file, but the font that draws them was not embedded — so the reader substitutes something that has no glyphs for those characters and you get rows of hollow rectangles.

It is worth knowing that this is a *rendering* failure, not a data failure. Copy the text out and it is intact. If you need it to look right, the fix is on the machine that made the PDF: embed the fonts when exporting.

Pages are blank, or a scan shows nothing

Some PDFs store their images in formats that not every engine implements — JBIG2 and JPEG 2000 are the usual culprits, and both are common in scanned documents from office equipment. An engine without those decoders draws a blank page rather than an error.

The tell is that the page is blank but the file size is large. Blank pages that are genuinely blank are tiny.

It is truncated or corrupt

A PDF keeps its index of objects at the end of the file. That is why a partially downloaded PDF fails completely rather than showing the first few pages — the reader looks for the index, does not find it, and gives up.

If a file arrived by email or download, check its size against the original. A file that stops early is not repairable in any meaningful sense; get it again.

What to do once you know which one it is

Matching the fix to the cause saves a lot of time:

  • Opens in a browser but not your reader → the reader. Use the browser or update it.
  • Asks for a password → you need the password. No tool can honestly bypass it.
  • Empty boxes instead of characters → fonts were not embedded. The text is fine; re-export from the source.
  • Blank pages, large file → an image format your reader cannot decode. Try a different reader.
  • Fails immediately, file looks small → truncated. Download it again.

Do it now, without uploading anything

Can a tool remove a PDF password?

Not the one that encrypts the content. If a document needs a password to open, the bytes are encrypted and there is nothing to work with. Tools that claim otherwise are either guessing passwords or handling the permissions-only case.

Why does the same PDF look different in two programs?

Because every reader has its own engine, and they differ in which fonts they substitute and which image formats they decode. A PDF describes a page; it does not guarantee two engines draw it identically.

My PDF opens but I cannot select the text.

Then there is no text layer — the pages are pictures. That happens with scans, and also with files that some compressors produce by flattening every page to an image. Text recognition can put a layer back.

Worth reading