PDF to Markdown screenshot

PDF to Markdown

Extract a PDF's text to Markdown, inferring headings, lists and tables.

Privacy First
0 Runs
Updated 2026-11-07
Runs in your browser
No account needed
Nothing uploaded
No OCR, stated plainly

Extract a PDF's text to Markdown, inferring headings, lists and tables. Paste your content and get an instant result.

Drop a PDF and this page extracts its text as Markdown, regrouping glyphs into lines by baseline, lines into paragraphs by vertical gap, and paragraphs into headings by relative text size.

The file is parsed in the browser with a vendored PDF library, which is loaded the first time you drop a document rather than on every page. Nothing is uploaded and the text never leaves the tab.

Only a text-based PDF can be read. A scan is a picture of text and has nothing to extract, so the page refuses it with the sentence that names OCR as the step it cannot perform.

How the structure is recovered

Text placement is read from the PDF's glyph coordinates and grouped into lines that share a baseline. A vertical gap larger than about one and a half line heights starts a new paragraph.

Relative text height decides headings, with three thresholds for three levels. The pipeline is ordered deliberately: getting the reading order right matters more than reproducing every font difference.

  • Lines grouped by baseline
  • Paragraphs split at a vertical gap
  • Three heading levels from relative text height

Where the extraction is approximate

A two-column article is the hard case: the reader has to decide which column a line belongs to, and it will sometimes interleave them. A table has no grid in the text layer, so its columns may be flattened into running text.

Bold and italic are inferred from font information where it is available. A PDF with no text layer at all is refused rather than returned as an empty file.

How to use the drop zone

Drop the PDF onto the Converter input workspace or the dashed drop zone. The page parses it in the browser with a vendored PDF library loaded on first use, regrouping glyphs into lines by shared baseline, lines into paragraphs by vertical gap, and paragraphs into headings by relative text size; nothing is uploaded and the text never leaves the tab.

Only a text-based PDF can be read: a scan is refused with the sentence naming OCR as the missing step. A two-column article can interleave columns and a table can flatten into running text, because the text layer carries no grid; those are the known approximate cases.

What to Expect

Best for

  • Getting editable text out of a digital PDF
  • Turning a report into Markdown notes
  • Keeping paragraph and heading structure approximately intact

Not ideal for

  • Scanned PDFs, which have no text layer
  • Two-column papers whose reading order matters
  • Tables that must keep their columns

What this tool keeps

  • The document's reading order in the common single-column case
  • Paragraph boundaries inferred from vertical spacing
  • Heading levels inferred from relative size

What may need cleanup

  • Interleaved lines from a two-column layout
  • A table flattened into running text
  • A heading whose size was close to body text

Common errors

  • Dropping a scan and expecting text
  • Reading an approximate heading as the document's real outline
  • Assuming a table survives as a Markdown table

Example use cases

  • Extracting a report's text for editing
  • Turning a digital book chapter into notes
  • Quoting a paragraph from a PDF

Sample input

Drop a two-page digital report with a heading and three paragraphs.

Sample output

A Markdown file with one heading line, three paragraphs separated by blank lines and the text in reading order, downloaded as document.md.

Who this is for

  • Writers extracting text from a PDF
  • Students turning a reading into notes
  • Anyone who needs editable text without an upload

Frequently asked questions

How is the PDF structure recovered?

Glyphs are regrouped into lines by baseline, lines into paragraphs by vertical gap, and paragraphs into headings by relative text size.

Can it convert a scanned PDF?

No. Without a text layer there is nothing to extract.

How does a two-column paper come out?

Interleaved lines from a two-column layout can be read in the wrong order.

Developer program

Build custom tools for your store

Join the SDK program to create useful ecommerce utilities for your store.