html2pdf

HTML to a single continuous PDF page. (中文说明)

html2pdf report.html

The PDF is one page as wide as you choose and exactly as tall as the content, so it reads the way the page scrolls in a browser. No paper-size breaks, no blank band at the bottom of every sheet, no card or table cut in half. Output is vector by default: text stays selectable and searchable, links stay clickable, and fonts are embedded.

Why not Save as PDF

A browser's print dialog always cuts the document into sheets. For anything laid out as a web page rather than a paper document, that leaves you choosing between white space and split content. Client-side libraries that turn the page into a canvas first produce images, so the text is gone.

html2pdf drives headless Chromium through Playwright and declares the page size with a CSS @page rule sized to the rendered content, with prefer_css_page_size set. Very tall pages render correctly this way; passing the size through the PDF API's width and height parameters did not in testing.

Install

uv tool install git+https://github.com/openwhale-labs/html2pdf
html2pdf --install-browser

pipx install git+https://github.com/openwhale-labs/html2pdf works the same way. The second command downloads the headless Chromium build matching the installed Playwright. Run it again after upgrading html2pdf if it reports that Chromium is missing.

Usage

html2pdf input.html [output.pdf]

html2pdf input.html --width 1200          # page width in CSS px (default 1000)
html2pdf input.html --hide ".lang-switch" # hide elements before rendering
html2pdf input.html --image               # full-page screenshot, for pages with
                                          # blur, backdrop-filter or soft gradients
html2pdf input.html --image --scale 3     # sharper screenshot (default 2)
html2pdf input.html --paged               # multi-page output following the
                                          # document's own @page and page-break rules
html2pdf input.html --open                # open the PDF when done

The output path defaults to the input name with a .pdf extension. Elements with the class toolbar are hidden automatically, which is a common name for a fixed on-screen control bar that has no place in a document. --open uses the platform's default PDF viewer.

Screen styles are used, not print styles, and rendering waits for web fonts to load. The document must be self-contained: inline CSS, local images, or data URIs. To export a language or theme variant of a page that switches at runtime, write a copy of the HTML with the initial state set in the markup and convert that.

--paged is for documents that are designed as pages, such as slide decks with an @page { size: 1280px 720px } rule and explicit page breaks. It renders with print media and lets Chromium paginate. The layout viewport is 1280 px wide (or --width if larger) and 1400 px tall, so size slides in px rather than vw and vh. samples/slides.html is a three-slide example.

As a library

from html2pdf.cli import html_to_pdf

width, height = html_to_pdf("report.html", "report.pdf", width=1000)

Development

git clone https://github.com/openwhale-labs/html2pdf
cd html2pdf
uv tool install -e .
uv run --with pytest pytest

Rendering tests are skipped when Playwright's Chromium is not installed (html2pdf --install-browser). samples/sample.html is taller than any sheet of paper and exercises the fixed toolbar, tables, cards and a tall block; samples/slides.html covers --paged.

License

MIT. Copyright (c) 2026 OpenWhale Labs.