Copy and paste is the least glamorous feature a document can support, which makes it all the more annoying when a PDF refuses. You drag across a paragraph and nothing selects - or everything selects, and pastes as □□□▯▯. There are exactly three common causes, each with a different fix, and you can diagnose yours in about ten seconds.
The ten-second diagnosis
Try to select a sentence, then try Ctrl+F for a word you can see:
| What happens | Cause |
|---|---|
| Nothing selects, search finds nothing | Scanned image - no text layer |
| Text selects, paste comes out as gibberish | Broken font mapping |
| Selection or copy is blocked outright | Copy restrictions |
Cause 1 - The PDF is a scanned image
The most common case by far. A PDF made by a scanner, a phone camera, or a fax machine contains pictures of pages, not characters. There is literally nothing to copy - the "text" you see is pixels.
The fix is OCR, which recognises the characters in the image and adds them as a real text layer. Your options, in rough order of effort:
- Read it in place: Sidenote OCRs a scanned PDF automatically on ingest (paid feature, included in the trial), after which you can copy, search, summarise, and question it with cited answers.
- Convert the file once: see how to make a PDF searchable for free and paid routes.
- Just extract the raw text: how to extract text from a scanned PDF.
Cause 2 - The text copies, but pastes as gibberish
Sometimes selection works fine and the paste is garbage: wrong letters, symbol soup, or empty boxes. This is a font problem, not a scanning problem. PDFs store text as internal character codes plus instructions for drawing them; a separate table (ToUnicode) maps those codes back to real characters for copy and search. When a PDF generator embeds a font without that table - common with subsetted fonts, old design tools, and some print drivers - the page renders perfectly but copies nonsense.
The workaround is to stop trusting the broken text layer and read the rendered page instead: OCR the document as if it were a scan. An OCR pass reads what the page looks like, which is exactly the thing that's still correct. Any of the routes from cause 1 work here too.
Cause 3 - The PDF restricts copying
PDFs can carry a permissions flag that disallows copying while still letting you open and read the file. Some viewers enforce it strictly, others don't, which is why the same file behaves differently in different apps.
Two honest points about this case:
- The restriction was set deliberately. If it's a colleague's or vendor's document, the right move is asking for an unrestricted copy - there's usually no drama, the flag is often a default someone never chose.
- Restricted is not the same as encrypted. A file you can't open without a password is a different situation from one you can read but not copy from.
When what you actually want is the answer, not the clipboard
A lot of copy-paste frustration is a means to an end: you were going to paste the text into a chatbot, a summary, or an email to ask about it. If that's the real goal, you can skip the clipboard entirely. Sidenote reads the PDF you have open in the browser - scanned, broken-font, or born-digital - and answers questions about it directly, with every claim carrying a citation that scrolls to the exact passage. The text you couldn't copy becomes text you can quote, with the source one click away.