So you're using C# and you want to convert some RTF to text. The solution is easy: You reach for the WinForms RichTextBox. Load your RTF in, access the Text property, and presto, it's all there. Mostly. Except smiley faces become the letter J. And sometimes non-ASCII text becomes gibberish even though old versions used to display it fine. And it's really, really slow. Also it leaks native memory.

You try the WPF version. Wait, did that one file take twenty-five seconds to load just because it had a 240x180 image in it?!

Forget it! You need something better. You need...

... the converter that's consistently over a hundred times faster than RichTextBox. 1.48 megs a second? That's unreasonable. 214 megs a second is slightly less unreasonable! That's like step 2½ out of 8 in Context is Everything!

• Wingdings 1, 2 and 3, Webdings, Symbol, and Zapf Dingbats all converted to equivalent Unicode characters.
• Non-ASCII text correctly converted where RichTextBox can't.
• Got huge files with tons of images? No problem. We blaze past image data so fast it may as well not exist.

BenchmarkDotNet v0.13.12, Windows 10 (10.0.19045.3448/22H2/2022Update)
AMD Ryzen 9 3950X, 1 CPU, 32 logical and 16 physical cores
.NET SDK 8.0.204
  [Host]     : .NET 8.0.4 (8.0.424.16909), X64 RyuJIT AVX2
  DefaultJob : .NET 8.0.4 (8.0.424.16909), X64 RyuJIT AVX2

• All basic plain text, hex-encoded chars, Unicode-encoded chars
• Symbol fonts (the abovementioned ones) converted to Unicode equivalents
• Characters specified as "SYMBOL" field instructions
• Undocumented use of the \langN keyword to specify character encoding - old versions of RichTextBox used to support this

• Tables: Cells and rows have spaces between them, but not much functionality beyond that.
• Lists: Numbers and bullets show up (that's better than RichTextBox most of the time), but indentation usually doesn't.

• Footnotes
• "HYPERLINK" field instruction value
• Math objects