Why a PDF fails to upload or process

Why a PDF will not upload: reading the refusals

A refusal looks the same every time, the file is not accepted, but the reasons behind it differ and so do the cures. Sort them by the moment the refusal happened: before the upload, during it, or already during processing. Each group holds one or two things that go against intuition.

The file extension settles nothing

The first thing worth knowing: the file type is read from the content, not from the name. Every format carries a signature of a few bytes at the very start: a PDF begins with `%PDF`, JPEG with its own sequence, PNG with another.

Two consequences follow, both against habit.

Renaming does not work. A picture called `document.pdf` is refused with a message saying a PDF is required. The tool sees a JPEG inside and is not fooled by the extension.

The reverse holds too. A genuine PDF with no extension, or with somebody else's extension, is accepted normally, because the content is what gets checked. A file named `scan` with no dot in the name goes through if there really is a PDF inside.

Every tool accepts its own set of types, and the refusal names the one it wants: a PDF is required, an image is required, a Word document is required. A message about the type is always about mismatched content, never about the name.

"The file is damaged" almost always means "it did not finish downloading"

A correct PDF carries an end-of-file marker. It is checked at upload separately from the signature at the front, and for a good reason: a file cut off mid-transfer starts out perfectly correct. The header is there, part of the content is there, the ending is not.

Such a file is refused immediately, before it ever reaches the queue. That beats accepting it and having processing die a minute later on an opaque error.

The most common source of those cut-offs is cloud folders. iCloud Drive, OneDrive and Google Drive keep a placeholder on disk rather than the file, pulling the content down on first access. A browser reading such a file sometimes gets only part of the data. On the surface everything looks fine: the file is visible, the size is shown, the icon is in place.

The cure is dull: open the file locally in any program, wait until it has genuinely downloaded, and upload again.

There is one deliberate exception, repair-pdf. The truncation check is switched off for it on purpose: fixing broken files is its whole job, and refusing them at the door would make the tool useless. If the file really is damaged rather than half-downloaded, that is the only door.

There are three limits, and they are different

They get confused because all three sound like "too much". They are about different things, and their advice points in opposite directions.

The third row used to be lumped in with the first, and the user was advised to compress a PDF in a situation where they simply had one file too many. It is now a separate cause with its own message.

The current numbers are shown on each tool's page, next to the upload area. That is where to look, rather than in articles: the values are configurable and they move.

For one heavy file the order is: compress-pdf first, and only if compression did not get you there, split-pdf. The reverse order leaves you with more files and more work.

An archive instead of separate files

A little-known option: you can upload a ZIP instead of separate files. For tools that accept only PDFs or only images, the archive is unpacked automatically and the files inside become the inputs.

That speeds up scan work noticeably, because the folder your scanner produced does not need unpacking by hand.

There is one limitation, and it is logical. If the tool takes exactly one file and the archive holds several, you get a message saying exactly that, telling you to unpack the archive and pick the file you need. Not a generic refusal, a specific explanation.

Passwords: which one matters

A PDF can carry two different passwords, and confusing them costs time.

The open password encrypts the content. Without it neither a program nor a service can read the file, and the job stops with a direct message asking for the password. It comes off through unlock-pdf, where you supply the password yourself.

The owner password restricts actions: no printing, no copying, no changes. It does not encrypt the text and does not interfere with processing at all. A document that forbids printing compresses, splits and merges quite happily.

A practical conclusion: if the file opens on your machine with a double click and no password prompt, but the job complains about a password, it is almost certainly your reader remembering the password and supplying it silently.

There is a button under the error message

For the four most common causes the interface does not stop at text; it offers a jump straight into the right tool with the file already loaded.

That last row belongs to refusals that happen during processing rather than at upload. Translation, for instance, works from text, and a scan is useless to it because the pages are pictures. Such a file uploads fine and is refused later, because reading the content only happens once processing starts.

Refusals that arrive later

A job can stop partway rather than at the door. There are not many causes and they are recognisable.

The document turned out to be a scan where text was needed: recognition through ocr-pdf fixes it, after which the original operation goes through.

Processing did not fit into the time allowed: usually a very long document, and splitting into parts helps.

The daily free operation allowance ran out: the counter is visible on the tool page, and this is not a problem with your file.

Too many requests in a row: a temporary restriction that lifts after a minute's pause.

What to do when a refusal makes no sense

1. Read the whole message: it almost always names both the cause and the offending file. 2. If the cause is one from the table above, press the offered button and come back. 3. If it is about the file type, open the document and see what it actually is. 4. If it is about damage, download the file locally and upload it again. 5. If it is about a limit, check the numbers on the tool page; those are the current ones.