Introducing Melville.PDF

The purpose of Melville.PDF is simple: make PDF documents show up in .NET apps for free. Free means free! Free as in speech? Free as in beer? Who cares? Free does not mean:

• There is an old version for free but the modern version is locked behind a paywall.
• The version is free if you agree to give your software away.
• It's free but it only works on Windows, or only on the desktop,
• The free version works for short files, or watermarks your files or anything else.
• The free lite version is so feature strapped that you will eventually have to buy the Pro version.

In short free means free.

I have recently become a fan of Bob Martin and his dogma surrounding Clean Code. You will have to decide for yourself if my code is clean. Jumping into any codebase is difficult. I added some introductory material under Architecture to orient you to the codebase.

Enough Gabbing -- Show Me Some Code!

1. The Wpf Way -- Show PDF in a Control

Using Melville.Pdf.Wpf from Nuget.

<UserControl x:Class="Melville.Pdf.WpfViewer.Home.HomeView"
             ...
             xmlns:controls="clr-namespace:Melville.Pdf.Wpf.Controls;assembly=Melville.Pdf.Wpf"
              >
    <controls:PdfViewer Source="C:\File.pdf"/>
</UserControl>

2. Using Skia Sharp -- save PDF page to a png file

Using Melville.Pdf.SkiaSharp from Nuget.

using System.IO;
using System.Threading.Tasks;
using Melville.Pdf.Model;
using Melville.Pdf.SkiaSharp; // Need to reference project Melville.Pdf.SkiaSharp

namespace Test;

var document = await new PdfReader().ReadFromFile("File.Pdf");
var output = new MemoryStream();
await RenderWithSkia.ToPngStreamAsync(document, 1, output);

For more examples and explanations please got to QuickStart

Who else contributed?

I made an effort to contain the dependencies as much as I can. The goal is to be free, and that means I need to control the number of people who can make an ownership claim to the source code. So far, I have taken, 3 dependencies, as discussed below.

• Freetype via SharpFont is used for font file parsing. This is the only native code dependency in the project, and much as I know it will make portability a headache, this was a necessity. PDF allows a lot of different font formats and writing parsers for them all would have added another year to the project. I forked SharpFont into the repository so I could fix some bugs arround finding the native DLLs. Freetype is licensed under GPL and a BSD-style FreeType License. SharpFont is licensed under the MIT license.
• JpegLibrary is the source of my JPEG parsing code. I forked this library into my own codebase in Melville.JpegLibrary. I have added some bugfixes and other features to this code to make it parse some of the unusual JPEG images this project had introduced me to. JpegLibrary is licensed under the MIT License.
• My JPEG2000 parser is another unofficial fork of csj2k. I made some superficial modifications to make it library play nicer with the stream based filter paradigm which is fairly fundamental to PDF. CSJ2k is licensed under th BSD license.

I made significant efforts to keep Melville.Pdf agnostic with regard to rendering or display technologies. To help me toward this end I developed the initial library with two initial targets, WPF and SkiaSharp. Melville.Pdf.SkiaSharp is a nice and contained example of the limited interface needed to connect Melville.Pdf to a display technology. The WPF and SkiaSharp bindings clearly depend on Wpf and SkiaSharp, respectively.

All of the dependences listed above have nonrestrictive open source style licenses which allow source and binary royalty-free distribution, including usage in closed source and commercial projects.

What does Melville.Pdf not do well?

This first release of Melville.Pdf is really focused on one goal, to render PDF documents in .NET applications. Years ago, someone whom I cannot remember told me that "To Ship is to Choose." Here are the things I chose not to include so I could ship this version.

• Public Key Encryption
• Types 4-7 Shader patterns
• Document navigation features like document outlines or page thumbnails.
• Pdf Generation.
  • Melville.Pdf actually has fairly robust PDF generation capabitites, but they are part of the test suite. I generate lots of different PDF files in my integration testing. The problem is I do not need to do much PDF Generation, and I do not a "customer" to drive a rational design. The result, I am sure, is a PDF generation framework much more focused on ability to produce unusual variants of PDF syntax over anything that might be useful in creating real documents. Most notably, I do not implement JPEG and JPEG2000 encoders.
• Forms of any kind.
• Annotations
• Transparency (Partial support)

What's next on the roadmap?

I have tried to keep the code quality high, but you will have to judge that. I hope to continue this project. Here are things I am likely to take this up next.

• Performance work.
• Filling and extracting text from Adobe and XFA forms -- another feature I could use at work.
• Document outline and page thumbnails
• More work on the Wpf viewer control to support at least zoom and rotation.

Recently released work is detailed on the Release history page.

I am extremely fortunate to have good employment as a physician at the Medical University of South Carolina meaning that programming can be just a hobby for me. I will probably work on the things above as time allows. My commit history ought to convince you that I really like programming. I take my amateur status seriously, though. It means I take the time to write the code correctly. I am not committing to any specific roadmap or schedule.

How can I help?

The biggest thing I need right now is bug reports. I am especially interested in examples of PDF files that either crash the renderer or render incorrectly. You can post such files as issues on git hub or email them to me at [email protected].

If one of the topics above interests you and you are a C# programmer, please drop me some email. Right now the codebase is largely the product of a single contributor and the code is pretty cohesive. I would enjoy collaborating with other developers who want to code in a similar style, and don't mind rather heavy edits from an opinionated maintainer.

More than anything else what I crave right now is users. Try out the library and tell me what works and what doesn't. Tell me your good ideas, I may like them more than my own. If there is some small issue that is blocking you let me know. Right now I'm a amateur with no budget, no schedule, and no users. If there is somthing that would win me some real users -- you just might find yourself at the top of the backlog!