You can handle this directly in your browser with our free pdf to word tool — nothing gets uploaded.
You open a PDF, select a paragraph, paste it into an email, and get something like %RPSHQ#DVVHW or a string of squares. The text is visibly there on screen, so the file clearly knows what the letters say. The problem is that the file and your computer disagree about *which* letters.
This is one of the most misunderstood PDF problems, because it is usually not a corrupted file. It is a mapping failure, and the fix depends on which of three causes you have.
This is the most common case. A scanned document, a screenshot saved as PDF, or a print-to-PDF from an image viewer creates a page where the text is a bitmap. You can *see* letters, but there are no characters to select — you are selecting the image, and whatever you paste is whatever the OCR layer (if any) guessed.
How to tell in five seconds: try to select a single word in the middle of a sentence. If the selection snaps to lines or the whole page at once, it is an image.
The fix: run OCR (optical character recognition). This converts the picture into a searchable text layer. Every PDF tool that offers OCR does the same core job; the differences are accuracy on small type, handwriting, and non-Latin scripts.
After OCR, copy-paste works — but the accuracy depends on scan quality. A clean 300 DPI scan of printed text usually comes out at 98%+ character accuracy. A phone photo of a page at an angle will not.
This one looks identical but is a different disease. The PDF has real text — it was generated by software, not scanned — but the font used has no proper "ToUnicode" mapping. In plain terms: the file stores glyph shapes and an internal code for each one, but never records that internal code 47 means the letter "e." When you copy, your system gets code 47 and looks it up in a standard encoding, which gives you a different character entirely.
Symptom: every character is wrong but the *pattern* is consistent. Spaces still fall in place, the word lengths match, and it almost looks like a substitution cipher. That consistency is the giveaway.
Fonts that commonly cause this: embedded subsetted fonts from old typesetting systems, custom publisher fonts, some printer drivers, and fonts with names like "ABCDEF+Arial" where the subset prefix is a warning sign.
The fix: re-extract the text with a tool that reconstructs the mapping, or re-run OCR. OCR does not care about font encoding — it reads pixels — so running OCR on top of a bad-mapping PDF usually produces clean text, at the cost of a small amount of accuracy. For a single page or a short quote, OCR is the practical answer.
Less common, but real: the PDF is fine, and the app you paste into assumes a different encoding. This shows up as accented characters, dashes, quotes, or curly apostrophes turning into é, â€", or Â. Everything else is readable.
That pattern — one wrong character where there should be an accent — is classic UTF-8-read-as-Latin-1 mojibake, not a PDF problem.
The fix: paste into a plain-text editor first (Notepad, TextEdit) and see whether the same garbage appears. If it is clean in plain text, the problem is the destination app, not the PDF. If you need the text for a web form or an email body, retyping a single mis-rendered character is often faster than debugging the pipeline.
Those squares are a font substitution issue: the file references a CJK or Arabic font that is not installed on your machine, so the viewer has no glyphs to draw. Two consequences:
The fix: install a reader that bundles CJK fonts (most desktop readers do), or open the file in a browser. If you only need the text, try copy-paste anyway; if the text layer is intact, you may get everything you need without ever fixing the display.
OCR output should always be proofread before you rely on it. Numbers, decimal points, and similar-looking glyphs (1/l/I, 0/O, 5/S) are where errors hide. If the text is going into a legal document, a citation, or anything with figures, treat OCR output as a draft.
Once the text layer is correct, the PDF becomes genuinely searchable too — which matters if you are going to search or index the document later, not just copy from it.
Related reading: annotate highlight notes in pdf for study, annual inventory count ocr records retention, attach pdf to text message.
Why do I see strange symbols when copying text from my PDF?
This usually happens when the PDF lacks embedded fonts or uses non-standard encoding. The browser cannot map characters correctly, resulting in gibberish. To fix this, use our tools Smart Extract feature, which reconstructs text layer data from raw coordinates. If that fails, ensure you download a version of the PDF with Copy Text enabled in its properties.
Does this tool work on scanned PDFs with no text layer?
No, standard copy-paste fails on scanned images because there is no actual text data, only pixels. However, you can use our OCR Pre-processing button on the upload screen. It runs a quick optical character recognition pass on the image, generating a hidden text layer. This allows you to copy and paste the converted text directly into your clipboard without needing external software.
Can I fix layout issues where text appears jumbled after pasting?
Yes, complex layouts often break the reading order during extraction. Use the Layout Preservation toggle before copying. This analyzes column positions and paragraph spacing to reconstruct the logical flow. If the issue persists, check the Character Mapping section to ensure special symbols are converted to standard Unicode, preventing mismatches when pasting into Word or Excel.
Is my uploaded PDF stored on your servers for privacy?
No, all processing happens locally in your browser using WebAssembly. Your files never leave your device, ensuring complete privacy. The text extraction and font reconstruction occur in memory. Once you close the tab, all data is automatically purged. We do not keep logs or backups of your documents, making it safe for sensitive legal or financial files.