PdfPecker
Docs: From wkhtmltopdf
Download PDF Changelog

A wkhtmltopdf alternative for .NET

wkhtmltopdf was archived in January 2023 and gets no more fixes, security fixes included. It renders with a Qt WebKit engine from 2016, so flexbox, grid and most CSS written since then don't work, and it has to be deployed as a native binary of about 90 MB for each operating system. .NET wrappers such as DinkToPdf, Rotativa, NReco.PdfGenerator and WkHtmlToPdf-DotNet carry the same engine with them.

PdfPecker is a replacement written in plain C#. It is one NuGet package with no native files, it supports modern CSS, and it converts in your own process instead of starting a program for every PDF. It is free under the MIT license.

dotnet add package PdfPecker

Replace the wkhtmltopdf call

From the command line or a Process

// Before: one wkhtmltopdf process per document
var process = Process.Start("wkhtmltopdf", "--page-size A4 -T 20mm -B 20mm invoice.html invoice.pdf");
process.WaitForExit();

// After
using PdfPecker;

HtmlToPdf.ConvertFile("invoice.html", "invoice.pdf", new PdfOptions
{
    PageSize = PageSize.A4,
    Margins = PageMargins.FromMillimeters(20, 10, 20, 10),
});

From DinkToPdf or WkHtmlToPdf-DotNet

// Before
var converter = new SynchronizedConverter(new PdfTools());
var document = new HtmlToPdfDocument
{
    GlobalSettings = { PaperSize = PaperKind.A4, Orientation = Orientation.Landscape },
    Objects = { new ObjectSettings { HtmlContent = html } },
};
byte[] pdf = converter.Convert(document);

// After
byte[] pdf = HtmlToPdf.Convert(html, new PdfOptions { PageSize = PageSize.A4.Landscape });

There is no converter object to keep alive and no native library to load: HtmlToPdf is static and thread-safe, so many requests can convert at the same time without the SynchronizedConverter lock.

From Rotativa in ASP.NET Core

Rotativa renders a Razor view and passes the HTML to wkhtmltopdf. Render the view to a string the way you already do for emails, or build the HTML from a template, and return the PDF:

app.MapGet("/invoice/{id}", (int id) =>
    Results.File(HtmlToPdf.Convert(RenderInvoice(id)), "application/pdf", $"invoice-{id}.pdf"));

Options side by side

wkhtmltopdf PdfPecker
--page-size A4, -O Landscape PageSize = PageSize.A4.Landscape, or CSS @page { size: A4 landscape }
-T, -R, -B, -L Margins = PageMargins.FromMillimeters(top, right, bottom, left), or CSS @page { margin: ... }
--footer-center "[page] of [topage]" FooterText = "{page} of {pages}", or an @page margin box with counter(page) and counter(pages)
--header-html header.html An element with data-sticky-pdf-header in the page (see Paged media)
--title Title, or the document's <title>
--outline On by default: bookmarks come from h1 to h6
toc Write the table of contents in your HTML with target-counter() and leader() for page numbers
--print-media-type Always on: @media print rules apply
--enable-local-file-access Files under BaseDirectory are read; AllowFileAccess = true allows others
--zoom, smart shrinking Scale. PdfPecker does not shrink content to fit, so check page breaks after switching
--javascript-delay, --run-script Not supported: render the content into the HTML before converting
--enable-forms On by default: inputs become fillable fields (FormFields = false turns it off)

What changes in the output

  • Pages can count differently. wkhtmltopdf shrinks content to fit the page by default, so a report that was 24 pages can become 29 at its real size. Set Scale if you need the old size back, but the real size is usually what the CSS meant.
  • Flexbox, grid, calc(), CSS variables and other modern CSS now work, so workarounds written for the old WebKit, such as tables used for layout or -webkit-box, can be removed or left as they are.
  • JavaScript is not run. Charts or text that scripts add to the page have to be in the HTML already, for example as SVG made on the server.
  • Fonts come from files you register, not from the server's installed fonts, so the output is the same on every machine. See Fonts.

Speed and size

On the same 4-core Linux VM, a one-page invoice took 310 ms with wkhtmltopdf and 0.86 ms with PdfPecker, and a 30-page report 780 ms and 45 ms. Deployment drops from about 90 MB of native files to one 4 MB DLL. The full comparison is on the Performance page.

Next steps

Edit this page on GitHub