gistfile1.txt

using HtmlAgilityPack;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;

namespace h2x
{
    class Program
    {
        static void Main(string[] args)
        {
            string html = "<p> test </p> <z> jjjjj </z>";

            //Replace Lead
            //Replace Body
            //Replace header
            //Replace figure
            //Replace image
            //Replace image title
            //Replace image credits
            //Replace href
            //Replace video
            //Replace video title
            //Replace video credits
            //Replace div, p, &nbsp; b, nobr
            //Replace quote text
            //Replace quote source

            string str = System.IO.File.ReadAllText("../../WithQuote.html", UTF8Encoding.Unicode);

            str = HtmlEntity.DeEntitize(str);

            HtmlDocument doc = new HtmlDocument();
            doc.LoadHtml(str);

            //lead
            var nodes = doc.DocumentNode.SelectNodes("//div[@class=\"Lead\"]");            
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "lead";
                }
            }
            
            //body
            nodes = doc.DocumentNode.SelectNodes("//div[@class=\"Body\"]");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "body";
                }
            }

            //header
            nodes = doc.DocumentNode.SelectNodes("//h3");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Remove();
                }
            }

            //figure
            nodes = doc.DocumentNode.SelectNodes("//figure");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "figure";
                }
            }

            //image
            nodes = doc.DocumentNode.SelectNodes("//img");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "image";
                }
            }

            //figure title
            nodes = doc.DocumentNode.SelectNodes("//div[@class=\"Figure-title\"]");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "figureTitle";
                }
            }

            //figure credits
            nodes = doc.DocumentNode.SelectNodes("//div[@class=\"Figure-credits\"]");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "figureCredits";
                }
            }

            //source quote
            nodes = doc.DocumentNode.SelectNodes("//blockquote[@class=\"SourceQuote\"]");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    var textNode = link.SelectNodes("//p[@class=\"SourceQuote-text\"]").First();
                    var sourceNode = link.SelectNodes("//span[@class=\"SourceQuote-origin\"]").First();

                    link.RemoveAllChildren();
                    link.Name = "quote";

                    //text
                    var quoteTextNode = new HtmlNode(HtmlNodeType.Element, link.OwnerDocument, 0);
                    quoteTextNode.Name = "quoteText";
                    quoteTextNode.InnerHtml = textNode.InnerText;
                    link.AppendChild(quoteTextNode);

                    //source
                    var quoteSourceNode = new HtmlNode(HtmlNodeType.Element, link.OwnerDocument, 1);
                    quoteSourceNode.Name = "quoteSource";
                    quoteSourceNode.InnerHtml = sourceNode.InnerText;
                    link.AppendChild(quoteSourceNode);
                }
            }

            //blockquote
            nodes = doc.DocumentNode.SelectNodes("//blockquote");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "quote";
                }
            }

            //a
            nodes = doc.DocumentNode.SelectNodes("//a");
            if (nodes != null)
            {
                foreach (HtmlNode link in nodes)
                {
                    link.Name = "link";
                }
            }


            //Remove unused tags

            var docStr = doc.DocumentNode.InnerHtml;

            docStr = Regex.Replace(docStr, "<(br|p|nobr|div|/p|/nobr|/div).*?>", string.Empty);

            docStr = "<document>" + docStr + "</document>";

            System.IO.File.WriteAllText("../../converted.html", docStr, UTF8Encoding.Unicode);

            //str = Regex.Replace(html, "img src=\\\"([^ ]+)\\\"", "img src=\"https://meduza.io$1\"");
            //str = HtmlRemoval.ReplaceTag(str, "/p", "/x");
        }
    }
}