Word → PDF

Convert a .docx to PDF right in your browser. Free, no upload. 1 GB max · Up to 1 GB · Processed locally, never sent to a server.

You have to hand in a document as a PDF and Word is nowhere to be found: a workstation with no licence, a Chromebook, a borrowed computer, a phone. The .docx file is there, the upload form takes PDF only, and Word’s built-in export is out of reach. That is the situation this tool addresses, and only that one.

The conversion happens in three stages. The mammoth library opens the .docx and rebuilds an HTML page from the document’s styles: paragraphs carrying the Heading 1 style become first-level headings, lists become lists again, tables stay tables. That page is then painted off-screen by html2canvas at a width of 794 pixels, which is the width of an A4 sheet, and finally sliced into pages that jsPDF assembles into a single file.

It is worth being clear about what that produces, because the word conversion suggests a photocopy. Word knows its own pagination to the point; here the document is rebuilt from its structure, then rendered with the site’s typography. The PDF you get is faithful to the content and to the hierarchy of the text, not to the original layout.

How to use it

  1. Drop a .docx file Only the .docx format of Word 2007 and later is read. An older .doc is refused, with a message inviting you to save it again in the right format.
  2. Start the conversion Clicking “Run” triggers the download of mammoth, html2canvas and jsPDF, then the reading of the document. Each phase is announced in the progress bar.
  3. Let the layout happen The rebuilt document is rendered off-screen in Georgia on a column of 794 pixels, with a margin of about 1.7 cm on each side. Nothing is adjustable at this stage.
  4. Check the page count reported The result states how many pages were produced. Past 200 rendered pages the conversion stops and invites you to split the document into several files.
  5. Open the PDF before sending it The file takes the name of the original document. Check above all where the page breaks fall, since they do not follow the ones Word showed you.

What mammoth keeps from your document

mammoth does not copy formatting across, it reads semantics. Word’s six heading levels are carried over, bulleted and numbered lists are rebuilt including nested ones, and bold, italic, underline, strikethrough, superscript and subscript are preserved, as are links and images. Tables keep their structure but lose their dressing: borders, shading and column widths are ignored. Footnotes, for their part, are gathered into a numbered list at the end of the document.

The consequence is direct: formatting applied by hand does not survive. A heading set in 18-point bold red Arial, without the Heading 1 style ever being applied to it, arrives as an ordinary paragraph. A document structured with Word’s styles therefore comes out markedly better than one formatted by eye.

What is lost, and why

Headers, footers and page numbers are not extracted. The same goes for columns, with the text falling back into a single column, for page breaks inserted by hand, for watermarks, and for the original margins.

Fonts go as well: mammoth deliberately ignores the typeface, size and colour of body text, so a report set in Calibri comes back in Georgia. Text boxes are a case of their own: their content is recovered, but handled as a separate paragraph placed after the one that contained them. The information is there, its position is not.

A PDF made of images, not of text

This is the most important point on the page. The final PDF is made of JPEG slices cut out of one single rendering. The text in it is not selectable, not copyable and not findable by a search, a screen reader has nothing to announce, and links are not clickable. Since the rendering is done at twice screen resolution, close to 190 dots per inch on an A4 sheet, reading stays comfortable and ordinary printing comes out correctly.

Page breaks fall at a fixed height, with no regard for the text: a line can end up cut in two across two pages, a table split down the middle. On running prose it is barely noticeable; on a heavily structured document it shows.

When another route is the better one

If you have access to Word, LibreOffice Writer or Google Docs, their PDF export remains superior: selectable and indexable text, pagination respected, headers where they belong, live links. For a contract heading to an electronic signature, a dissertation with imposed pagination, or a document that will have to be searched full text, go through those programs.

This tool serves a different purpose: freezing a piece of content, sending a readable note to someone who will not open a .docx, producing a PDF when no word processor is installed. The document never leaves the tab, the rendering taking place in a hidden area of the page; only the three libraries are fetched on first use, from files the site hosts itself.

Frequently asked questions

Why is the text in my PDF not selectable?

Because the page is painted and then cut into JPEG images before being placed in the document: the PDF holds pixels, not characters. If you need text that can be copied and searched, use the PDF export in Word or LibreOffice.

My .doc file is refused. What should I do?

Only .docx is read, because it is an open format that mammoth knows how to unpack. Open the document in Word or LibreOffice, save it as .docx, then drop it in again.

Where did my headers and page numbers go?

They are not extracted from the document. If those lines matter, copy them into the body of the text before converting, or produce the PDF from software that handles sections.

Why do my headings look like ordinary text?

They were most likely never given the Heading 1 to Heading 6 styles, but enlarged and set bold by hand instead. The conversion relies on styles and does not guess at intentions.

Why is a document of several hundred pages refused?

The whole document is rendered as one image before being sliced. Past 200 produced pages the operation stops, so as not to exhaust the browser’s memory. Split the file into several parts.

Are the images in my document kept?

Yes, they are built into the rendering and brought back to the width of the text if they overflow it. An image anchored in the margin or wrapped by text, on the other hand, simply takes its place back in the flow.

Similar tools

Discover other tools

Categories