pdfHTML does not support fieldset tag yet and indeed your sample is converted into something like following:

You can plugin in your custom implementation into pdfHTML though to tweak the visual appearance to some degree (depending on your needs). Here is a basic example on how to tweak the visual appearance with a big number of assumptions:
private static class LegendTagWorker extends DivTagWorker {
public LegendTagWorker(IElementNode element, ProcessorContext context) {
super(element, context);
}
@Override
public IPropertyContainer getElementResult() {
IPropertyContainer result = super.getElementResult();
// Shift the position of the legend a bit above
result.setProperty(Property.POSITION, LayoutPosition.RELATIVE);
result.setProperty(Property.TOP, -14);
// Set the background of the text to white so that it does not overlap with the border of outer element
if (result instanceof Div && ((Div) result).getChildren().get(0) instanceof Paragraph) {
Paragraph p = (Paragraph) ((Div) result).getChildren().get(0);
((Text)p.getChildren().get(0)).setBackgroundColor(ColorConstants.WHITE);
}
return result;
}
}
private static class CustomTagWorkerFactory extends DefaultTagWorkerFactory {
@Override
public ITagWorker getCustomTagWorker(IElementNode tag, ProcessorContext context) {
if (tag.name().equals("legend")) {
return new LegendTagWorker(tag, context);
}
return super.getCustomTagWorker(tag, context);
}
}
Then you just need to pass your tag worker factory to converter properties:
ConverterProperties properties = new ConverterProperties();
properties.setTagWorkerFactory(new CustomTagWorkerFactory());
HtmlConverter.convertToPdf(inFile, outFile, properties);
Now we get the following result:

Please note that this is a very basic example and you will need to tweak it to your needs. Before this feature is implemented in pdfHTML you will have to use a workaround - either in code as I suggested, or you can do some tweaks with CSS, similar to the tweaks I am making in Java code