WeasyPrint: bring the web to PDF and paper

Bring the web to PDF and paper

Kozea

Software company, 10 persons, Lyon area
Builds custom web applications for businesses
- Industrialization
- HTML5/CSS3 e-learning
- Semi-automated reporting
Recent W3C member

Industrialization example: 2000 customized pharmacy websites on the same code base.
e-learning: stylesheets rather than PowerPoint-style slides.
Reporting through CSS and SVG, including vector graphs of data aggregated on the fly, producing PDF.

Open-source

We use lower-level open-source bricks a lot.
You may have had a similar experience: you just use it, then start reporting issues, start looking at the code, start fixing issues yourself and send pull requests …

Making open-source

community.kozea.org

And at some point you may start your own open-source project. Find ours at community.kozea.org and on GitHub.

PDF

Pages
Hi-Fi layout (set in stone)
For printing, archiving …
Tools? Automation?
Example: hundreds of invoices / month

Flexible web documents are great but sometimes you need PDF. We want to generate them automatically as much as possible.

LaTeX → PDF

Great typography, but …
Rigid stylesheets
Not (always) reproducible
Gloablly painful

Painful, in one word.

SVG → PDF

Viable for small, fixed form documents (eg. invoices)
No ”flow” (automatic line breaking…)

More pleasant but hack-ish.

CairoSVG

cairosvg.org
SVG 1.1 renderer
PDF, PS and PNG output
Has a non-standard hack for multi-page
Open-source

By the way, we made our own SVG implementation in Python.

HTML+CSS → PDF

We want this!
Reuse all server-side tools that generate HTML
Reuse our designer’s CSS skills
Profit from the web’s flexibility

This is what we really want. It makes sense however we think of it.

CSS for Paged Media

h1 { page-break-before: right }
h1, h2, h3 {
    page-break-after: avoid }
@page { size: A4 landscape;
        margin: 2cm }
@page :left { @bottom-right {
    content: counter(page) } }
@media print { nav { display: none } }

Actually, CSS does have a lot for Paged Media: page break control, page sizing, headers and footers, …

HTML+CSS → PDF

Browser engines: not great at pagination, little or no Paged Media CSS.
Print-specific engines: great document quality but not open-source

… but Gecko and Webkit do not really implement these parts of CSS.
Specialized engines do, but didn’t fit for various reasons.

WeasyPrint

1.5y ago: “Let’s make a new layout engine!”
weasyprint.org
HTML, CSS, SVG, … → PDF
Command-line API, Python API
Open-source

I said “No, it’s a bad I idea, we’re not gonna do that.” But we still did.
We’ve now been using WeasyPrint in production in multiple applications for 8+ months.
There are regular stable releases, you can use it.

W3C, CSS Working Group

Reading specs
Asking clarifications on the mailing list
Reporting issues
Invited Experts
Joining W3C as a Member company

Getting involved is a gradual process, even small contributions (like feedback on specs) are greatly appreciated.
But still, here I am today although one year ago I didn’t even consider this could ever happen!
My point is: you can do this too, at whatever scale you like.

The CSS 2.1 spec has some “dark corners”… but don’t worry about those, we figured them out for you :)
There is plenty more that is easier to approach.

Thank you.

Come and chat with us!