PDF Text Checker
See if your PDF's text is real, selectable and searchable.
How to use PDF Text Checker
- 01
Drop in your PDF
Add the PDF you're having trouble copying, searching, or extracting text from. Everything is analysed locally in your browser.
- 02
Read the page-by-page report
See which pages have real selectable text, which are scanned images, and which have garbled character encoding — plus whether the fonts are embedded.
- 03
Fix it if needed
If pages are scanned images, click through to OCR to add a searchable text layer, then re-check.
Tested and last verified on
About PDF Text Checker
Find out why a PDF won't let you copy text, search inside it, or extract cleanly. Page by page, this tool checks for a real text layer versus a scanned image, detects garbled or broken character encoding, and flags fonts that aren't embedded — all in your browser. If pages are scanned, jump straight to OCR to make them searchable.
Why some PDFs won't let you copy or search text
A PDF can look identical on screen whether its words are real, selectable text or just a flat picture of text. The difference only appears when you try to highlight a sentence, run a search, or paste the content somewhere else. If nothing selects, if the highlight covers the whole page as a single block, or if the text you paste comes out as nonsense, the document has a text-layer problem — and until you know which kind, you cannot fix it. PDF Text Checker opens your file with pdf.js, the same engine browsers use to display PDFs, and reads the underlying text stream of every page so you can see exactly what a computer sees rather than what your eyes see.
There are three common failure modes, and this tool tells them apart. A scanned page is an image with no text stream at all — typical when a document was photographed or run through a scanner without OCR. A garbled page does have a text stream, but the font is missing a correct ToUnicode map, so the characters that copy out are not the characters you read; you end up with random letters, empty boxes, or (cid:12)-style codes. A genuinely empty page has neither text nor images. Every page in the report is labelled with which of these applies, alongside the number of words and images that were found on it.
What the report tells you, page by page
For each page you get a status badge, a word count, and a short preview of the text that was actually extracted, so you can confirm at a glance whether the machine-readable content matches the visible page. The summary tiles count how many pages carry real selectable text, how many are scanned images, and how many are garbled — which matters when only part of a document is affected, such as a born-digital contract with a scanned signature page appended at the end.
The checker also lists the fonts used in the document and whether each one is embedded. Non-embedded fonts are substituted with whatever typeface the reader happens to have installed, which can quietly shift the layout and, in some cases, break copy-paste and search when the file is opened on a different device. Seeing this before you send a document out helps you avoid the awkward situation where a file looks perfect on your machine but falls apart on someone else's.
Fixing the problems it finds
If the checker reports scanned pages, the fastest fix is optical character recognition: use the "Make searchable with OCR" button to carry the same file straight into AsliPDF's OCR tool, which adds a real, selectable text layer behind the image so the document becomes searchable and copyable. If the problem is garbled encoding, the source document usually needs to be re-exported from its original application with proper font embedding, or rebuilt through OCR. And if the fonts are simply not embedded, re-saving the PDF with fonts embedded before you distribute it prevents surprises for other readers.
Because every check runs locally in your browser, you can safely diagnose confidential contracts, medical records, or financial statements without uploading them anywhere — the file never leaves your device. There is no account, no queue, and no watermark on anything you do here, so you can run as many documents through the checker as you need, as often as you need, and use the results to decide which of your PDFs are ready to share and which need repair first.
Questions
Everything runs in your browser. How we keep files private
Why can't I copy or search the text in my PDF?
Usually because the page is a scanned image (a picture of text, with no real text layer) or because the font has a broken character map, so the text extracts as gibberish. This tool tells you which of the two is happening, page by page.
How do I know if my PDF is scanned or has real text?
Drop it in above. Any page reported as a "scanned image" has no selectable text and needs OCR; pages reported as "selectable text" already copy and search normally.
What does 'garbled text' mean?
The page has a text layer, but the font is missing a correct ToUnicode map, so copied text comes out as wrong characters or (cid:NN) codes. This commonly breaks copy-paste, search, and text extraction even though the page looks fine on screen.
Does this upload my PDF anywhere?
No. The entire analysis runs in your browser with pdf.js — your document never leaves your device.
Why does it matter whether fonts are embedded?
Non-embedded fonts are substituted on other devices, which can shift layout and, in some cases, break text extraction and searchability when the document is opened elsewhere.
Part of: Sharing ID documents without leaking them
Mask, redact and strip your Aadhaar, PAN and other IDs before you send them.
- Find and black out Aadhaar and PAN numbers automaticallyChecksum-verified detection, every match shown for review, then true removal — or UIDAI-style masking that leaves the last four digits.
- Permanently remove anything sensitive by handDraw over names, addresses or figures and have the underlying content genuinely deleted, not covered.
- Strip hidden metadata before you share a fileDocuments carry author names, software details and edit history you never intended to send.