Turn an HTML file into a PDF

This takes the structure of an HTML file and lays it onto pages. Headings stay headings, lists stay lists, paragraphs wrap to a measured column and break across pages when they run out of room.

What it does not take is the appearance. No CSS is applied, nothing is fetched, and no script is run.

That limit is the whole design, so it is worth being clear about

If you want a page to come out looking like it looks in a browser, your browser already does that better than anything else could: Print, then Save as PDF. It has the stylesheet, the fonts, the images and the layout engine.

What a browser’s print does badly is everything else. It needs the page open, it cannot be automated over a folder of files, and it bakes in whatever the screen width happened to be.

So this is for the other job: you have HTML — an export, a saved article, a generated report, a mail archive — and you want its writing on pages you can file, send or read. Structure without decoration is the right output for that, and pretending to do full rendering with no layout engine would produce something that looks nearly right and is wrong in ways you would have to check every time.

Nothing is fetched and nothing is run

Two consequences worth stating.

There is no address bar. Giving this tool a URL would mean it making a request to a third party, and a tool whose promise is that nothing leaves your device does not get to make that request on your behalf. Save the page from your browser first.

And no script runs. A page whose words are drawn into the document by JavaScript has no text in the file itself, so there is nothing to convert and the tool says so rather than handing you an empty PDF. Saving the page from your browser after it has loaded puts the text in the file.

Stylesheets, scripts, templates and comments are dropped along with their contents, so a stylesheet cannot arrive as a page of CSS selectors — which is what happens when a converter strips tags rather than understanding them.

What it will get wrong

Tables come out as their text rather than as tables. Nested lists are flattened to one level. Anything positioned by CSS lands wherever it sits in the markup, because the markup is all there is to go on.

The parser is a walk over the tags rather than a full HTML5 parse, since a browser worker has no DOM and pulling in a complete parser would cost more than every other part of this tool. It handles unclosed paragraphs, inline tags inside a sentence, and entities. It does not build a tree.

Nothing leaves the device

The file is read and the PDF is written by a worker inside your browser. No upload, no account, no server copy, and no outbound request of any kind.