Drop a PDF and get its text back as Markdown, with headings picked out by font size. It runs in your browser, so the file is never uploaded anywhere — including to us.
Drop a PDF or choose it below
It is converted here, in this tab. The file is not uploaded.
What it does
The file is read in the page by JavaScript. Nothing is sent to a server.
Every text run is extracted with its font size and position, page by page.
Lines noticeably larger than the body text become headings, and the two largest distinct sizes become `#` and `##`. Bullet and numbered lines are recognised by their markers.
You get Markdown to clean up by hand — which, for a PDF, is the honest end state.
What it will not do
A PDF describes where glyphs sit on a page, not what they mean. There is no heading, no list and no table in the file — only text at coordinates. Everything structural here is inferred, and inference is sometimes wrong.
Tables come out as loose lines. Reconstructing a grid from coordinates reliably is a much larger problem than this tool solves.
A scanned PDF is an image of text and produces nothing. It needs OCR first, which this does not do.
Multi-column layouts are read in the order the file stores the text, which is often not reading order.
Images, figures and equations are dropped.
Wanted a documentation site, not a text file?
Uploading the same file to Docszero runs this conversion and keeps going: images stay, the document splits into pages by heading, and you get a searchable site on your own subdomain in under a minute. Free for one site, no card, and you can export everything whenever you like.
No. The conversion happens in your browser using JavaScript on this page. The file is read from disk into the tab and never sent over the network. You can confirm it by opening your browser's network tab, or by disconnecting from the internet after the page loads — the converter still works.
Why does the output need editing?
Because a PDF does not contain the structure you are asking for. It records glyphs at coordinates, so headings, lists and paragraphs have to be guessed from font size and spacing. Expect to fix headings and rejoin the odd paragraph. Any tool claiming a clean PDF conversion with no edits is either using a model to rewrite the text or is not telling you the whole story.
Nothing came out. What happened?
The PDF is almost certainly a scan — an image of a page rather than text. There are no characters in the file to extract. You need OCR to turn the image into text first; this tool does not do that.
My tables look wrong.
They will. A table in a PDF is text positioned to look like a grid, with nothing marking the rows and columns. The lines come out in roughly the right order but not as a Markdown table, and rebuilding the grid is manual.
Can I turn a manual into a documentation site instead?
Yes. Uploading the PDF to Docszero runs the same extraction, splits it into pages, adds search and navigation, and publishes it — which is usually what someone converting a manual actually wants. The free plan covers one site and needs no card.