/ –
Upload a PDF to preview it here.

Help

This app converts PDF, Word and PowerPoint files to Markdown with Docling , using this project's doc2md command.

  1. Upload one or more documents in the sidebar.
  2. Check them in the Input tab (use the dropdown when there are several).
  3. Choose the options (hover over each for an explanation).
  4. Click Convert and wait for the progress bar.
  5. Read the result in the Output tab, then click Download .
Setting Meaning
Documents Choose one or more PDF (.pdf), Word (.docx) or PowerPoint (.pptx) files, up to 200 MB each. Other file types are refused. A new selection replaces the current files. Two files that differ only by extension (report.pdf and report.docx) cannot be converted together, because both would become report.md.
Extract images Save the pictures of each document and show them in the Markdown. They are stored in a '<name>_artifacts' folder next to the Markdown file, so the download becomes a ZIP file. Pictures with a figure caption are named after it, the others image_1.png, image_2.png, ... (see Figure file names). The caption is looked up automatically and can be missed. Without this option, pictures are left out.
Image resolution Resolution of the pictures taken from PDF files. 72 dpi is enough for the screen, 300 dpi for printing. Higher values make the pictures and the download much bigger, and the conversion slower: 600 and 1200 dpi need a lot of memory for documents with many pages. Pictures in Word and PowerPoint files are always kept at their original resolution.
Figure file names How pictures that belong to a figure are named. 'Keep as in the article' uses the label written in the article: Fig_1.png, Extended_Data_Fig_3.png, Supplementary_Figure_S2.png. 'Normalised' always uses Figure_1.png, and Figure_S3.png for supplementary figures. In both cases the parts of one figure get a letter (Fig_3a.png, Fig_3b.png).
Recognise formulas Recognise mathematical formulas in PDF files and write them as LaTeX ($...$ and $$...$$). Slower, and downloads an extra model the first time it is used. Equations in Word files are always converted, with or without this option. Equations in PowerPoint files are not supported by Docling.
Convert / Cancel Convert all loaded files with the options above. The first conversion can take several minutes because Docling loads (and, the very first time, downloads) its models. Cancel stops the conversion; files finished before that are kept.
Download One file converted without images: a .md file. Several files, or a file with images: a ZIP containing each <name>.md, its <name>_artifacts folder, and conversion_log.txt if some files failed. The button is greyed out until a conversion has finished, and again when you change the files or options.
File (dropdown) Choose which of the loaded files to show. The Input and Output tabs always show the same file.
Rendered / Source 'Rendered' shows the Markdown formatted, with formulas typeset and pictures shown. 'Source' shows the raw Markdown text, which you can copy.
  • Preview: PDF files are shown page by page. Word and PowerPoint files show an information card (pages or slides, title, author) instead; their Markdown is shown in the Output tab.
  • Formulas: PDF formulas need 'Recognise formulas'. Word equations are always converted. PowerPoint equations are not supported by Docling.
  • Scanned PDFs: text in scanned pages is recognised automatically (OCR), which is slower than reading a text PDF.
  • Chemical structures are not converted to text; with 'Extract images' they are kept as pictures.
  • Upload limit: 200 MB per file.
  • / buttons: previous / next page.
  • Type a page number and press Enter to jump to it.
  • Click on the page, then use the ← / → keys (or Page Up / Page Down, Home / End).

Files are converted on the machine running this app, in a temporary folder that is deleted when you close the app window. Pictures are never sent anywhere.