You drag across a paragraph, press Ctrl+C, paste — and nothing arrives. Or something arrives and it reads like a ransom note: boxes, wrong letters, words glued together. Three very different problems produce this one symptom: the page has no text at all (it’s a picture), the document forbids copying (a permissions flag), or the text exists but is wired to the wrong characters (a broken font map). Each has its own fix, and telling them apart takes about twenty seconds.
The twenty-second diagnosis
Open the file in any reader and drag the cursor across a line of body text. If nothing highlights, there is no text in the file — you are looking at an image, and the fix is recognition, not permissions. If it highlights but Copy is greyed out (or pasting produces nothing), the document carries a restriction flag; most readers will also show “Content Copying: Not Allowed” in the document properties under Security. If it copies but the paste is nonsense, the text is real and unrestricted — it is the translation table between the font and actual characters that is broken. The three sections below take the cases in that order.
Case 1 — there is no text, only a picture of it
Scans, phone photos and faxes all produce pages that show text without containing any. A PDF viewer can only hand your clipboard characters that exist in the file, and in a raw scan there are none — just pixels arranged into letter shapes. The fix is OCR (optical character recognition): software reads the pixels, recognises the words, and writes them into the file as an invisible text layer positioned exactly over the printed ones. The page looks identical afterwards, but selection, copying and search all work.
Our OCR PDF tool does exactly this, in English, German or both at once, and straightens skewed pages while it works. On a clean 300 dpi scan of printed text, modern OCR is close to flawless; on faint faxes, low-resolution photos and handwriting, expect errors — recognition is a reconstruction, not a recovery. The full walkthrough lives in our guide to making a scanned PDF searchable.
Case 2 — copying is switched off on purpose
PDF has two different locks, and they behave nothing alike. A user password encrypts the document: nobody sees a single page without typing it. An owner password is subtler — the file opens normally for everyone, but carries flags that tell readers what to refuse: no copying, no printing, no editing. Your viewer isn’t broken when Copy is greyed out; it is politely obeying instructions embedded in the file.
The practical consequence: if the file already opens for you, Unlock PDF can usually lift the restrictions — leave the password box empty and run it. We verified this exact case: a document with copying and editing disallowed came out with all flags removed and copying working everywhere. If the tool reports a wrong password instead, the file uses a stronger scheme and you genuinely need the owner password. And if the document demands a password just to open, that is the encrypting kind — you must know it; nothing reputable gets around real encryption.
Use this on your own documents. Restrictions often exist for a reason — licensed reports, purchased e-books, filings. Removing them is for files you own or have permission to use, and rules differ by country. This is practical guidance, not legal advice; our guide to securing a PDF covers the other direction.
Case 3 — it copies, but pastes as garbage
The strangest case: selection works, copying works, and the paste reads □□□ or Wkh txlfn eurzq ira. Here’s why. A PDF doesn’t store letters — it stores glyphs, numbered drawing instructions for a particular font, plus a translation table (the ToUnicode map) that says which character each glyph means. Copying reads that table. When a generator embeds a subsetted or symbol font and writes the table wrong — or skips it — the page still renders perfectly, but the copied glyph numbers translate to nothing sensible. Old design tools, CAD exports and some print drivers are the usual culprits.
The broken table can’t be repaired from outside, so the reliable fix rebuilds the text from the one thing that is definitely correct: the pixels. Two steps on Ream. First, Flatten PDF in raster mode turns each page into a clean image — this runs in your browser and deliberately discards the faulty text layer. Then run the result through OCR PDF, which writes a fresh, correct layer from what it reads. The two steps matter because OCR, by design, never overwrites text that already exists in a file — it would otherwise skip exactly the pages you need fixed.
Getting the text out cleanly once it exists
Even from a healthy PDF, the clipboard is the clumsiest exit. PDF to Text extracts the entire document to a plain .txt in one step, right in your browser — and if it detects a scan with no text layer, it will tell you and point you to OCR instead of handing you an empty file. When you need the formatting too — headings, tables, layout — PDF to Word produces an editable document; our honest guide to what survives that conversion sets expectations before you rely on it. Either way, expect extracted text to carry line breaks where the layout had them — that isn’t damage, it’s how PDFs store position rather than flow.
| Symptom | Cause | Route |
|---|---|---|
| Nothing highlights | Scanned image — no text in the file | OCR PDF |
| Highlights, but Copy is blocked | Owner-password restriction flags | Unlock PDF, password box empty |
| Asks for a password to open | User-password encryption | You need the password — then Unlock removes it for good |
| Pastes as □□□ or wrong letters | Broken font-to-character map | Flatten (raster), then OCR |
| Pastes with odd breaks or lost formatting | Normal extraction behaviour | PDF to Text or PDF to Word |
Is it private, and is it free?
All of it is free, with no sign-up and no watermark. Flatten and PDF to Text run entirely in your browser — those files never leave your device. OCR, Unlock and PDF to Word need real server-side engines, so they run in our hardened sandbox: HTTPS in, an isolated container with no network access, and everything deleted the moment your download is ready — never stored, never logged. The specifics are on our privacy overview.
Frequently asked questions
Is it legal to remove copy protection from a PDF? For documents you created, own, or have permission to use — generally yes; that is what the owner-password mechanism is for. For licensed or purchased content, restrictions may be backed by law, and rules differ by country. When in doubt, ask the document’s owner. This is not legal advice.
Why does copied text lose all its formatting? The clipboard’s plain-text lane carries characters, not layout. Bold, columns and tables live in the PDF’s drawing instructions, which Copy never touches. If formatting matters, convert with PDF to Word instead of copying.
Why is there a line break after every line? A PDF stores each line’s position on the page, not the paragraph it belongs to. Extraction reads the lines in order and honours their ends. Pasting into a word processor and using find & replace to remove single line breaks fixes it in seconds.
Why are the spaces missing (or doubled) in what I paste? Some PDFs place words by coordinates without storing space characters at all — extractors must guess the gaps from the geometry, and narrow or stretched typesetting fools the guess. OCR-rebuilt text (flatten, then OCR) usually spaces more sensibly than a broken original layer.
Can OCR read handwriting? Print, yes; handwriting, mostly no. Standard OCR engines are trained on typefaces, and cursive defeats them. For a handwritten page, expect fragments at best — transcription is still a human job.
Why do I get empty boxes (□) when pasting? The copied characters are real but your target app’s font can’t display them — or the PDF’s map translated glyphs to nothing meaningful. Paste into a plain-text editor first: if boxes remain there, it’s the map (case 3 above); if they become letters, it was just the font.
Why can I copy from some pages but not others? Mixed documents: born-digital pages carry text, while inserted scans (a signed page, an appendix) carry none. Run OCR over the file — it adds text only where it’s missing and leaves the good pages untouched.
Does “printing to PDF” fix garbled copying? Sometimes — the re-render can rebuild a usable map — but just as often it carries the fault straight across, because the print path reuses the same font data. The flatten-then-OCR route works from pixels, so it cannot inherit the broken table. That reliability is why we recommend it.
