iTextSharp XMLParser.Parse throws and swallows exceptions repeatedly

Question

I have a function which I am using to try to give iTextSharp some HTML and from it generate a PDF. This function successfully generates the PDF, complete with CSS styling, but it is not running fast enough for our requirements.

I have noted the one area in particular which is taking a long time to execute is the call to XMLParser.Parse which I have seen taking upwards of 14 seconds to complete for an 8 page document showing little more than a table of data headed by icons. During the execution of this method I have noticed (in the output window) three exceptions being thrown (and presumably caught) by iTextSharp or code which iTextSharp calls into. These exceptions are:

1. 'System.Collections.Generic.KeyNotFoundException' in mscorlib.dll
2. 'iTextSharp.tool.xml.exceptions.NoDataException' in itextsharp.xmlworker.dll
3. 'System.ArgumentException' in mscorlib.dll

The three exceptions repeat (albeit not necessarily in that order) until the Parse method has finished executing.

Whilst I realize I don't need to handle these exceptions myself, I mention them since I am trying to improve the performance of this method and understand that catching exceptions can be an expensive operation. What I am looking for is what the cause of these exceptions being thrown is and if it is dependent upon bad data I am passing in, what data would it be that is bad?

This is the HTML to PDF function as it currently stands. Note I have already tried using XMLWorkerFontProvider.DONTLOOKFORFONTS, as suggested by the itext_so manual, but not made any performance gains from doing so. I have also noted a NullReferenceException thrown and swallowed once byPdfWriter.GetInstance; I wondered if this may also related to the thrown exceptions in the Parse method.

public static byte[] GeneratePdfFromHtml(string html, Action<PdfWriter, Document> pdfSettings = null, string additionalFooterText = null)
{
    var tagProcessor = (DefaultTagProcessorFactory)Tags.GetHtmlTagProcessorFactory();

    tagProcessor.RemoveProcessor(HTML.Tag.IMG);
    tagProcessor.AddProcessor(HTML.Tag.IMG, new CustomProcessorImageTag());

    using (var workStream = new MemoryStream())
    using (var document = new Document())
    //NOTE: The NullReferenceException is thrown via this call.
    using (PdfWriter writer = PdfWriter.GetInstance(document, workStream))
    {
        PdfEventHelper pdfEventHelper = new PdfEventHelper(additionalFooterText);

        writer.PageEvent = pdfEventHelper;
        writer.CloseStream = false;
        pdfSettings?.Invoke(writer, document);
        document.Open();

        var xmlWorkerFontProvider = new XMLWorkerFontProvider(XMLWorkerFontProvider.DONTLOOKFORFONTS);

        //Not noticably faster without this font directory registration.
        xmlWorkerFontProvider.RegisterDirectory("~/Content/fonts", false);

        //TODO: If further performance is needed then this line is the next slowest (.43ms).
        var htmlContext = new HtmlPipelineContext(new CssAppliersImpl(xmlWorkerFontProvider));

        htmlContext.SetTagFactory(tagProcessor);

        Func<string, string> mapPath = HttpContext.Current.Server.MapPath;
        var cssResolver = XMLWorkerHelper.GetInstance().GetDefaultCssResolver(true);

        foreach (var cssFileName
            in new[]
            {
                "bootstrap.min.css",
                "Pdf.css",
                "Rdp.css"
            })
            cssResolver.AddCssFile(mapPath($"~/Content/{cssFileName}"), true);

        using (var reader = new StringReader(html))
        {
            new XMLParser(
                new XMLWorker(
                    new CssResolverPipeline(
                        cssResolver,
                        new HtmlPipeline(
                            htmlContext,
                            new PdfWriterPipeline(
                                document,
                                writer))),
                    true))
            //TODO: Speed up this line - this is the slowest line in the method by far.
            //NOTE: This throws a series of ArgumentExceptions, NoDataExceptions and KeyNotFoundExceptions.
            .Parse(reader);

            document.Close();

            return workStream.ToArray();
        }
    }
}