FOP.Sharp 26.8.4329

Apache FOP for .NET

A modern, cross-platform C# port of Apache FOP (Formatting Objects Processor): read an XSL-FO document and render it to PDF, plain text, Markdown or HTML. It targets .NET 10, runs anywhere .NET runs (Windows, Linux, macOS), and replaces FOP's Java AWT/Java2D/Batik stack with managed libraries (PdfSharp, SixLabors.ImageSharp, SixLabors.Fonts).

This is the C# port. The original Java sources live alongside it in this repository (fop-core/, fop-util/, fop/, …) and remain the source of truth for the port; see the upstream Apache FOP README for the Java project. The .NET solution lives under dotnet/.

Packages

The port is split into many small component projects internally, but ships as two NuGet packages:

What's inside the FOP.Sharp package

The FOP.Sharp library package is self-contained: it bundles every component assembly of the port (you only ever reference FOP.Sharp). The bundled assemblies are:

Its third-party NuGet dependencies (PdfSharp, SixLabors.ImageSharp, SixLabors.Fonts, SixLabors.ImageSharp.Drawing) are declared on the package and restored automatically.

Installation

# Library
dotnet add package FOP.Sharp

# Command-line tool (global)
dotnet tool install --global FOP.Sharp.Cli

Usage

Render XSL-FO to PDF (library)

using Fop.Render.Pdf;

var processor = new FopProcessor();

// From a file
processor.ConvertFile("input.fo", "output.pdf");

// From a string, into a byte[]
byte[] pdf = processor.Convert(File.ReadAllText("input.fo"));

// From streams
using var fo  = File.OpenRead("input.fo");
using var pdf2 = File.Create("output.pdf");
processor.Convert(fo, pdf2);

Register custom fonts before converting (Liberation faces are bundled as the fallback):

var processor = new FopProcessor();
processor.RegisterFont("MyFont-Regular.ttf", "My Font");
processor.RegisterFontsDirectory("./fonts");
processor.ConvertFile("input.fo", "output.pdf");

Use the native, PdfSharp-free renderer (supports font subsetting and encryption):

byte[] pdf = processor.ConvertNative(File.ReadAllText("input.fo"));

Resolve images that are not files (library)

An fo:external-graphic or a background-image normally names a file on the local disk. When a document's illustrations live somewhere else -- a content management system, an object store, a zip, a database -- set a ResourceResolver and the engine asks it for the bytes instead of writing every image out to a temporary file first:

using Fop.Layout;

var processor = new FopProcessor
{
    ResourceResolver = ResourceResolvers.FromDelegate(
        uri => uri.StartsWith("icn:") ? store.Open(uri[4..]) : null),
};

processor.ConvertFile("input.fo", "output.pdf");

The resolver is consulted for every image URI before the file system is; returning null for a URI leaves it to the normal handling, so a resolver covering one scheme need not know about any other. A URI nothing resolves is passed to the renderer as a path exactly as before, and an image that cannot be read leaves its area empty rather than failing the document.

ResourceResolvers also has Directory(...) (files under fixed base directories, looked up by the URI's last segment), Compose(...) (first resolver that has the URI wins) and None. RFC 2397 data: URIs are decoded whether or not a resolver is set.

Render to plain text, Markdown or HTML (library)

using Fop.Render.Text;

string text     = new PlainTextRenderer().Convert(foXml);
string markdown = new MarkdownRenderer().Convert(foXml);
string html     = new HtmlRenderer().Convert(foXml);

Command-line tool

# Output format inferred from the extension
fop input.fo output.pdf
fop input.fo output.md

# Explicit flags
fop -fo input.fo -pdf output.pdf
fop -xml data.xml -xsl style.xsl -html output.html

# Register fonts / use the native renderer
fop -fo input.fo -pdf out.pdf -fontdir ./fonts -native

Run fop -help for the full option list.

Building from source

cd dotnet
dotnet build Fop.slnx
dotnet test  Fop.slnx

Requires the .NET 10 SDK (pinned in dotnet/global.json).

To produce the NuGet packages locally:

cd dotnet
dotnet pack src/Fop/Fop.csproj         -c Release   # the Fop library
dotnet pack src/Fop.Cli/Fop.Cli.csproj -c Release   # the fop .NET tool

Status

The port is an incremental, bottom-up effort. A working end-to-end FO → PDF pipeline already covers a substantial XSL-FO subset (block/inline text, fonts, colour, the box model, tables, lists, static content, keeps & breaks, footnotes, floats, markers, links, leaders, hyphenation, Knuth–Plass line breaking, embedded SVG, bookmarks and more). See dotnet/CLAUDE.md for the architecture and dotnet/TODO.md for the phased plan and current progress.

License

Licensed under the Apache License, Version 2.0, the same license as upstream Apache FOP.