Loading Techie Utils…

PDF to Markdown Converter

Convert a PDF into Markdown with its headings, emphasis, lists, tables and pictures. Free, and runs entirely in your browser — nothing is uploaded.

About PDF to Markdown

Turns a PDF's text into Markdown, with headings, paragraphs and — if you want them — the tables it can find. Useful for getting a document into a wiki, a static site, a diff, or a language model's context window.

A PDF has no heading levels. It has text that happens to be larger. So the distinct heading sizes in the document are collected and ranked: the largest becomes #, the next ##, down to ####. That is a guess, and it is the same guess a reader makes looking at the page — which is why it usually agrees with what the author intended, and why it goes wrong on a document with an unusual design.

Paragraphs are reassembled from the gaps between lines, so prose comes out as prose rather than as one line per printed line. A two-column layout is the case where that inference is hardest, because the text runs are stored in drawing order and the columns can interleave.

Where it helps

  • Headings survive, ranked by text size the way a reader ranks them.
  • Paragraphs are reassembled rather than dumped one line at a time.
  • Tables found on the way can be appended as real Markdown tables.
  • Optional rules between pages, for checking against the original.
  • The document stays in your browser — which matters for anything being prepared for a model.

Limits

  • Heading levels are inferred from size, so a document that uses one size for everything comes out flat.
  • Two-column layouts can interleave, because PDF stores drawing order rather than reading order.
  • Lists are not detected: a bulleted list comes out as paragraphs starting with a bullet character.
  • Images are not extracted or referenced — the Extract Images tool does that separately.
  • A scan has no text at all; run OCR on it first.

Frequently asked questions

Why are all my headings the same level?

Because they are all the same size in the document. Heading levels here are ranked from the distinct sizes found, so a design that distinguishes headings by weight or colour rather than size gives no signal to rank.

Is this good enough for a language model?

It is considerably better than raw text, because the structure survives — a model can see where a section begins. Check a page or two against the original before feeding it a hundred: a layout that confuses the paragraph grouping will confuse the model too.

What happened to the images?

They are not referenced. Extract Images pulls them out as files, and you can add the links yourself where they belong.

Why is the text out of order?

Almost always a multi-column layout. PDF stores what was drawn, in the order it was drawn, which for two columns is often line by line across both.

Related tools

More Images & Files tools