Table of Contents
The symptom is always the same. You open a PDF, drag your cursor across a paragraph, and… nothing. No highlight, no copy. Or the whole page selects as one big block. Then you run it through a PDF-to-EPUB converter anyway, and out comes an EPUB with blank pages, or one where every “page” is a picture you can’t resize or search.
These are the same problem wearing two outfits: your PDF has no usable text layer. The converter isn’t broken — there’s simply no text in the file for it to convert.
Here’s how to confirm the diagnosis, figure out which of the four causes you have, and fix it.
Confirm the Diagnosis (30 Seconds)
Two quick checks:
- Ctrl+F / Cmd+F and search for a word you can see on the page. No results for a word that’s clearly there → no text layer.
- Select All (Ctrl+A / Cmd+A) and copy into a text editor. If you get nothing, or a scattering of whitespace, or gibberish like
□□□ ⌘⌥→ no usable text layer.
If the search works and the copied text is clean, your text is fine and your conversion problem lies elsewhere — probably formatting rather than text.
The Four Causes (and How to Tell Which One You Have)
Cause 1: It’s a scanned document
The most common case by far. Someone photographed or scanned paper pages; the PDF is a stack of images. Zoom in to 400% — if letter edges get fuzzy or pixelated, it’s a scan.
Fix: OCR. There is no shortcut — the text must be recognized from the images. Your options range from free-but-manual to automatic; we compared all four routes in our scanned-PDF guide.
Cause 2: The PDF was flattened or exported as images
Some tools rasterize pages on export — deliberately (to lock content) or as a side effect (“print to PDF” from an image viewer, some slide-export tools, some CAD/design apps). The document was born digital, but the text arrived as pixels anyway. Zoom in: same fuzzy edges as a scan, though usually cleaner and higher-resolution.
Fix: Same as a scan — OCR. The good news: these files are usually crisp, so recognition accuracy is excellent.
Cause 3: The text is there, but converted to outlines
Designers sometimes export PDFs with fonts “converted to outlines” — every letter becomes a vector shape. Zoom to 1000% and the letters stay perfectly sharp (unlike a scan), but nothing selects.
Fix: Also OCR, surprisingly. The vector shapes can’t be turned back into characters directly, but because they render razor-sharp at any resolution, OCR reads them nearly perfectly.
Cause 4: The PDF is copy-protected
The text layer exists, but the file’s permissions flag forbids copying. Telltale sign: you can search the document (search reads the text layer), you just can’t select or copy.
Fix: This is the one case that isn’t an OCR problem. If you own the document or have the right to convert it, the restriction can often be lifted by the document owner re-exporting it. For DRM-protected purchased ebooks, conversion is a legal gray-to-black area that depends on your jurisdiction — we don’t advise on stripping DRM, and PDF2EPUB doesn’t process DRM-protected files.
Why “Just Convert It Anyway” Produces Garbage
A converter given a text-free PDF has two choices, and both are bad:
- Give up on text and wrap each page image in EPUB markup. Result: a huge file that looks like an ebook but behaves like a photo album — no reflow, no font-size changes, no search, terrible on e-ink. (Details: why conversions come out as images.)
- Run a bundled quick-and-dirty OCR pass. Some converters do this silently, using dated engines with default settings. Result: garbled text, scrambled reading order on multi-column pages, and footnotes fused into the body.
Either way, the failure was decided before conversion started. The OCR step is the conversion — treat it as the thing you’re choosing a tool for.
The Fix, Step by Step
For causes 1–3 (which cover ~95% of “can’t select text” cases):
- Check your source quality. If you’re the one scanning: 300 DPI minimum, flatten the page, avoid shadows in the gutter. OCR can’t recover what the pixels don’t contain.
- Pick your OCR route. Free and local: OCRmyPDF + Calibre — workable for clean, single-column text in a major language. Paid desktop: ABBYY FineReader — better recognition, still structure-blind (comparison here). AI-native: PDF2EPUB — recognition plus structure reconstruction (headings, footnotes, tables, reading order) in one pass, free to try on a sample.
- Verify the output before converting nine more books. Open the EPUB: search for a rare word, change the font size, open the table of contents. If all three work, the text layer is real and structured.
FAQ
Adobe Reader says “This document has been secured.” Is that cause 4? Probably, but check causes 1–2 as well — scanned PDFs are sometimes also flagged as secured. If search finds words inside the document, it’s pure cause 4; if search finds nothing, it’s a scan wearing a padlock.
My PDF selects text fine, but the copied text is gibberish. Same problem? Close cousin. The text layer exists but its encoding is broken (common in PDFs from older non-Latin typesetting systems). Practically it’s treated the same way: ignore the broken layer and re-OCR. Full write-up here.
Can I just add a text layer to the PDF and keep it as PDF? Sure — that’s exactly what OCRmyPDF does, and it makes the PDF searchable. But it doesn’t make it reflowable: the pages are still fixed images with invisible text on top. For actual ebook reading on a Kindle, Kobo, or phone, you still want a real EPUB conversion.
Will this work for non-English books? Yes, but engine choice matters more. Tesseract needs the right language pack; mixed-language pages (an English book quoting Chinese, say) break traditional OCR far more often than AI-based recognition. Test one page before committing.
Got a PDF that won’t select? Run a free sample through PDF2EPUB — if the text can be recognized, you’ll see it in the output within minutes, on your own document.