Extending

Alias a tag

Reuse another tag's conversion with Tags.Aliases (key = tag to remap, value = tag to convert it as):

var config = new Config();
config.Tags.Aliases["u"] = "em"; // convert <u> as if it were <em>
var converter = new Converter(config);

^{snippet source | anchor}

Custom readers

Implement IMdReader, mark it with [MarkdownReader("tag", ...)], and pass its assembly to the converter. A custom reader overrides the built-in one for those tags. Readers need a parameterless constructor and the using ReverseMarkdown.Dom; and using ReverseMarkdown.Readers; namespaces.

// Render <a> as "text (href)" instead of a markdown link.
[MarkdownReader("a")]
public class PlainLinkReader : IMdReader
{
    public void Read(AngleSharp.Dom.IElement element, ReaderContext ctx)
    {
        var href = element.GetAttribute("href") ?? "";
        var text = element.TextContent.Trim();
        ctx.Emit(new MdText($"{text} ({href})") { SourceTag = "a" });
    }
}

^{snippet source | anchor}

Register it by passing the assembly to the converter:

var converter = new Converter(new Config(), typeof(PlainLinkReader).Assembly);

^{snippet source | anchor}

Trimming and Native AOT

ReverseMarkdown is trim- and Native AOT-compatible. The default conversion path uses no reflection.

The [MarkdownReader] + assembly form above discovers readers by scanning assemblies with reflection, which the trimmer can remove and Native AOT cannot analyze. Those overloads are marked [RequiresUnreferencedCode] / [RequiresDynamicCode], so the analyzer will warn if you use them in a trimmed/AOT app.

Under trimming or AOT, register readers explicitly instead - no attribute, no assembly, no reflection:

var converter = new ReverseMarkdown.Converter();
converter.RegisterReader("mark", new HighlightReader());

RegisterReader overrides the built-in reader for that tag, exactly like the scanned form. Call it before converting; it is not safe to call concurrently with a conversion.

Recipe: convert only a whitelist of tags, rest as plain text

Register a reader that reads an element's children (which strips the tag but keeps its text) for the tags you want flattened, and set Tags.Unknown = Bypass so any remaining tag is stripped too.

[MarkdownReader("strong", "b", "em", "i", "del", "s", "h1", "h2", "h3", "h4", "h5", "h6",
                "blockquote", "code", "pre", "span", "sub", "sup", "hr", "br", "img", "table")]
public class StripToTextReader : IMdReader
{
    public void Read(AngleSharp.Dom.IElement element, ReaderContext ctx) => ctx.ReadChildren(element);
}

^{snippet source | anchor}

var config = new Config
{
    Tags = { Unknown = Config.UnknownTagsOption.Bypass }
};
var converter = new Converter(config, typeof(PlainLinkReader).Assembly);

^{snippet source | anchor}

<p>, <li>, <ol>/<ul>, and <a> keep converting to markdown; everything else comes out as plain text:

<p>Hi <strong>bold</strong> <a href="http://x.com">click</a></p>
->  Hi bold click (http://x.com)

Transform the Markdown DOM

For anything the reader hooks don't cover, converter.Parse(html) returns a mutable MarkdownDocument you can traverse and transform before converter.Render(document).

var converter = new Converter();
var document = converter.Parse(html);
// inspect / filter / reshape the document here
var markdown = converter.Render(document);

^{snippet source | anchor}