Convert a PDF to Markdown, without uploading it

What this tool does

• The structure is recovered from the layout: text larger than the body becomes a heading, a gap taller than a line space starts a paragraph, a bullet starts a list.
• Bold and italics are read from the font name, links from the clickable areas of the document.
• Repeated headers, footers and page numbers are removed, and the screen says how many lines went.
• A table is only written as a table when its columns are clear all the way down. At the slightest doubt, it stays as text.
• The result is shown before it is saved, and the screen flags pages that seem to be in columns, where the reading order is probably wrong.

What it does not do: The structure is guessed from the layout, which is why it is shown to you. It works well on a document written in a word processor (a report, a contract, a manual) and badly on free layouts. A page in several columns is read line by line, from one edge to the other, so out of order: the screen flags it. Formulas, footnotes and images are not carried over. A word split at the end of a line is joined back when the next line starts in lower case, which can fuse a hyphenated compound. A scanned document has no text to read: run it through text recognition first.

Nothing is uploaded. Cut your network connection once this page has loaded: the tool keeps working. How to check it yourself.

The other tools

• Merge PDFs: Several files, one document, in the order you decide.
• Organize pages: Reorder, rotate, delete, duplicate: the whole plan.
• Extract pages: Pick pages, take them out into a document of their own.
• Delete pages: Remove pages while looking at them; they are not left under the file you get.