LlmFiles

Is this PDF a scan or a real text PDF?

Drop a PDF and get a verdict for every page: does it carry extractable text, or is the page a picture of text? This is the same question behind “no text could be extracted from this file” — answered page by page, with the one command that fixes it.

Drop a PDF here or pick a file — up to 25 MB, nothing is forwarded anywhere Choose a PDF
Read on the LlmFiles server, no third-party OCR or cloud Does not use your daily conversion limit
—
PageVerdictTextWordsImages

What the verdicts mean

textCharacters are stored on the page, so any converter reads them. If your tool still fails, the problem is the file’s format, not a missing text layer.
scanThe page is an image. A scanner, a phone photo saved as PDF or “print to PDF” from a picture produces this — there is nothing to extract until OCR runs.
partialSome pages have text, others are scans. OCR only the missing pages with ocrmypdf --skip-text.
encryptedThe file is password-protected. Remove protection on a copy with qpdf --decrypt, then check it again.
brokenThe structure cannot be parsed — a truncated download or a damaged cross-reference table. Re-save it from the app that produced it.
emptyPages carry neither text nor images: a placeholder or blank PDF, nothing to convert.

Same check from the command line

The local CLI runs the identical check offline — no upload, no quota:

llmfiles doctor report.pdf
llmfiles doctor report.pdf --json   # machine-readable, same fields as the API

Exit code is 0 when every page has text, 1 when something needs OCR or unpacking, 2 for a bad argument. After OCR, convert the file to Markdown — one page, one ZIP of many files, or the whole folder from the CLI.

Questions

Why does a PDF converter say “no text could be extracted”?

The PDF has no text layer on those pages. A scanner, a phone photo saved as PDF, or “print to PDF” from an image writes a picture of the page, not characters — so every extractor returns empty text. This page reads the file and tells you which pages are affected.

Does checking a PDF upload it anywhere else?

No. The file is read by the LlmFiles server that serves this page and is not forwarded to any third-party service, OCR API or cloud storage. The check does not count against the daily conversion limit.

What is the fastest fix for a scanned PDF?

Run OCR once on a copy: ocrmypdf --redo-ocr in.pdf out.pdf. That writes an invisible text layer under the images, keeps the look of the document, and the result converts to Markdown like any digital PDF. A deep dive into what a PDF really contains is in the PDF guide.