NetPdf

Pure C# / .NET 10 HTML + CSS → PDF. No browser, no native Chromium, no subprocess at render time, no AGPL or revenue-capped dependencies. Apache-2.0. A real HTML/CSS layout engine built for print and paged media — rendered straight to PDF, not driven through a headless browser.

NetPdf ships its own HTML/CSS layout engine — block / inline / flex / grid / table layout, fragmentation across pages, international text shaping (HarfBuzz), a bounded "least-ugly page split" pagination cost model, and its own PDF byte writer — rendered natively to PDF, with a Skia raster fallback only where a feature can't be expressed natively.

using NetPdf;

byte[] pdf = HtmlPdf.Convert("""
    <!DOCTYPE html>
    <html lang="en"><body>
      <h1>Hello, PDF</h1>
      <p>Rendered by NetPdf — no browser required.</p>
    </body></html>
    """);

File.WriteAllBytes("hello.pdf", pdf);

Highlights

  • Deterministic — the same input produces byte-identical PDF output (opt-in frozen /CreationDate).
  • AOT-clean — no reflection in core paths; a Native AOT smoke gate runs in CI.
  • Paged media — @page, the 16 margin boxes, running headers/footers via string() / element(), counter(page), orphans/widows, and break controls.
  • International text — HarfBuzz shaping, UAX #9 bidi, UAX #14 line breaking, and language-aware hyphens: auto via the optional NetPdf.Languages.* packs.
  • Diagnostics, not silent corruption — unsupported features emit a stable, documented code.

Next steps