‫استخراج تمام XPathهای یک محتوای HTMLایی به کمک کتابخانه HtmlAgilityPack

using System;
using System.Linq;
using System.Net;
using System.Text;
using HtmlAgilityPack;

namespace HapTests
{
    public class HtmlReader
    {
        public Action<string> ParseError { set; get; }

        public Func<HtmlNode, bool> ParserHtmlNode { set; get; }

        public void StartParsingHtml(Uri url)
        {
            using (var client = new WebClient { Encoding = Encoding.UTF8 })
            {
                client.Headers.Add("user-agent", "Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0)");
                StartParsingHtml(client.DownloadString(url));
            }
        }

        public void StartParsingHtml(string htmlContent)
        {
            if (string.IsNullOrWhiteSpace(htmlContent))
                throw new ArgumentNullException("content");

            var doc = new HtmlDocument
            {
                OptionCheckSyntax = true,
                OptionFixNestedTags = true,
                OptionAutoCloseOnEnd = true,
                OptionDefaultStreamEncoding = Encoding.UTF8
            };
            doc.LoadHtml(htmlContent);

            if (doc.ParseErrors != null && doc.ParseErrors.Any())
            {
                foreach (var error in doc.ParseErrors)
                {
                    if (ParseError != null)
                        ParseError(error.Code + " - " + error.Reason);
                }
            }

            if (!doc.DocumentNode.HasChildNodes)
                return;

            handleChildren(doc.DocumentNode.ChildNodes);
        }

        private void handleChildren(HtmlNodeCollection nodes)
        {
            foreach (var itm in nodes)
            {
                if (itm.Name.ToLower().Equals("html"))
                {
                    if (itm.Element("body") != null)
                        handleChildren(itm.Element("body").ChildNodes);
                }
                else
                    handleHtmlNode(itm);
            }
        }

        private void parserChildNodes(HtmlNode content)
        {
            foreach (var item in content.ChildNodes)
            {
                handleHtmlNode(item);
            }
        }

        private void handleHtmlNode(HtmlNode htmNode)
        {
            switch (htmNode.Name.ToLower())
            {
                case "html":
                case "body":
                    handleChildren(htmNode.ChildNodes);
                    break;

                default:
                    if (ParserHtmlNode == null)
                        throw new ArgumentNullException("ParserHtmlNode");

                    if (ParserHtmlNode(htmNode))
                        parserChildNodes(htmNode);

                    break;
            }
        }
    }
}

new HtmlReader
{
    ParseError = error => Console.WriteLine(error),
    ParserHtmlNode = htmlNode =>
    {
        //switch(htmlNode.Name) { }
        return true; //it's a nested node.
    }
}.StartParsingHtml(html);

using System;
using System.Diagnostics;
using System.IO;
using System.Text;
using HtmlAgilityPack;

namespace HapTests
{
    class Program
    {
        static void Main(string[] args)
        {
            var html = 
                @"<table width='750' border='0' style='font-size: 10pt; width: 736px' class='boxcar2 gerd'>
            <tbody><tr>
            <td height='70' colspan='4' class='boxcart1 gerd'>
            <iframe width='718' scrolling='no'>
            </iframe></td>
            </tr>
            <tr>
            <td height='70' colspan='4' class='boxcart1 gerd'>
    </td>
            </tr>
            <tr>
            <td width='193' height='36' class='boxcart2 gerd'>
            <a target='_self' href='Curr.cbi.2.php'>نرخ ارز مبادله ای بانک مرکزی</a></td>
            <td width='181' height='36' class='boxcart2 gerd'>
            <a target='_self' href='Curr.cbi.php'>نرخ ارز مرجع بانک مرکزی</a></td>
            <td width='149' height='36' class='boxcart2 gerd'>
            <a target='_self' href='curv.htm'>نمودار قیمت طلا</a></td>
            <td width='199' height='36' class='boxcart2 gerd'>
            <a target='_self' href='index.php'>قیمت طلا و سکه در بازار ایران</a></td>
            </tr>
            <tr>
            <td height='48' colspan='4' class='boxcart1 gerd'>
            <p dir='rtl'><span style='font-size: 13pt;'>تابلو آنلاین قیمت جهانی طلا و نقره ( دلار 
            )</span></p></td>
            </tr>
            <tr>
            <td height='57' colspan='2' class='boxcart1 gerd'>قیمت لحظه ای هر انس 
            نقره در بازارهای جهانی<br>
            <span style='font-size: 9pt;'>
            </span></td>
            <td height='57' colspan='2' class='boxcart1 gerd'>قیمت لحظه ای هر انس 
            طلا در بازارهای جهانی<br>
            <span style='font-size: 9pt;'>
            </span></td>
            </tr>
            <tr>
            <td height='48' colspan='4' class='boxcart1 gerd'>
            <p dir='rtl'><span style='font-size: 13pt'>تابلو آنلاین قیمت طلا ، سکه 
            و نقره در بازار ایران ( ریال )</span></p>
            </td>
            </tr>
            <tr>
            <td style='direction: rtl; font-size: 8pt' colspan='4'><div align='center'>
                            <table id='gold_tbl'><tbody><tr><th>قیمت طلا</th><th>قیمت زنده</th><th>تغییر</th>
                            <th>کمترین</th><th>بیشترین</th><th>زمان</th></tr><tr><td>انس طلا <sup>دلار</sup></td>
                            <td class='s0_1'>1,375.90</td><td class='c0_1 neg'>(-0.34%) -4.70</td>
                            <td class='l0_1'>1,374.90</td><td class='h0_1'>1,380.80</td><td class='z0_1 fa'>17:53</td>
                            </tr><tr><td>مثقال طلا</td><td class='s3_2'>5,290,000</td>
                            <td class='c3_2 pos'>(1.63%) 85,000</td><td class='l3_2'>5,200,000</td><td class='h3_2'>5,320,000</td><td class='z3_2 fa'>17:50</td></tr><tr><td>گرم طلای 18</td>
                            <td class='s3_3'>1,221,200</td><td class='c3_3 pos'>(1.63%) 19,600</td><td class='l3_3'>1,200,400</td><td class='h3_3'>1,228,100</td><td class='z3_3 fa'>17:50</td>
                            </tr><tr><td>انس نقره <sup>دلار</sup></td><td class='s0_5'>21.83</td><td class='c0_5'>(0.00%) 0.00</td><td class='l0_5'>21.67</td><td class='h0_5'>21.96</td>
                            <td class='z0_5 fa'>17:53</td></tr></tbody></table><br><table id='coin_tbl'><tbody><tr><th>سکه</th><th>قیمت زنده</th><th>تغییر</th><th>کمترین</th>
                            <th>بیشترین</th><th>ارزش طلا</th><th>زمان</th></tr><tr><td>بهار آزادی</td><td class='s3_10'>12,650,000</td><td class='c3_10 pos'>(2.68%) 330,000</td>
                            <td class='l3_10'>12,320,000</td><td class='h3_10'>12,650,000</td><td class='z4_10'>11,918,400</td><td class='z3_10 fa'>16:07</td></tr><tr><td>امامی</td>
                            <td class='s3_11'>12,960,000</td><td class='c3_11 pos'>(2.61%) 330,000</td><td class='l3_11'>12,630,000</td><td class='h3_11'>13,050,000</td><td class='z4_11'>11,918,400</td>
                            <td class='z3_11 fa'>17:43</td></tr><tr><td>نیم</td><td class='s3_12'>6,880,000</td><td class='c3_12 pos'>(2.69%) 180,000</td><td class='l3_12'>6,700,000</td>
                            <td class='h3_12'>6,900,000</td><td class='z4_12'>5,959,200</td><td class='z3_12 fa'>16:08</td></tr><tr><td>ربع</td><td class='s3_13'>4,250,000</td><td class='c3_13 pos'>(2.41%) 100,000</td>
                            <td class='l3_13'>4,150,000</td><td class='h3_13'>4,300,000</td><td class='z4_13'>2,978,100</td><td class='z3_13 fa'>17:42</td></tr><tr><td>گرمی</td><td class='s3_14'>2,940,000</td>   
                            <td class='c3_14 pos'>(3.16%) 90,000</td><td class='l3_14'>2,850,000</td><td class='h3_14'>2,940,000</td><td class='z4_14'>1,465,400</td><td class='z3_14 fa'>17:40</td></tr></tbody></table></div></td>
            </tr>
            </tbody></table>
                ";

            extractXPath(html);
            test(html);
        }

        /// <summary>
        /// Converts /#comment[1] to /comment()[1] 
        /// or /#text[1] to /text()[1]
        /// </summary>
        private static string GetValidXPath(string xpath)
        {
            var index = xpath.LastIndexOf("/");
            var lastPath = xpath.Substring(index);

            if (lastPath.Contains("#"))
            {
                xpath = xpath.Substring(0, index);
                lastPath = lastPath.Replace("#", "");
                lastPath = lastPath.Replace("[", "()[");
                xpath = xpath + lastPath;
            }

            return xpath;
        }

        private static void extractXPath(string html)
        {
            var sb = new StringBuilder();
            new HtmlReader
            {
                ParseError = error => Console.WriteLine(error),
                ParserHtmlNode = htmlNode =>
                {
                    if (htmlNode is HtmlTextNode)
                    {
                        sb.AppendLine("Text NodeName: " + htmlNode.Name.Trim());
                        sb.AppendLine("InnerText: " + htmlNode.InnerText.Trim());
                    }
                    else
                    {
                        sb.AppendLine("NodeName: " + htmlNode.Name.Trim());
                        var nodeText = new StringBuilder();
                        for (int i = 0; (i < htmlNode.OuterHtml.Length && htmlNode.OuterHtml[i] != '>'); i++)
                            nodeText.Append(htmlNode.OuterHtml[i]);

                        nodeText.Append(">");

                        sb.AppendLine("Node Start: " + nodeText.ToString());
                    }

                    sb.AppendLine("XPath: " + GetValidXPath(htmlNode.XPath.Trim()));
                    sb.AppendLine(Environment.NewLine);

                    return true; //it's a nested node.
                }
            }.StartParsingHtml(html);

            File.WriteAllText("xpath.txt", sb.ToString());
            Process.Start("xpath.txt");
        }

        private static void test(string html)
        {
            var doc = new HtmlDocument
            {
                OptionCheckSyntax = true,
                OptionFixNestedTags = true,
                OptionAutoCloseOnEnd = true,
                OptionDefaultStreamEncoding = Encoding.UTF8
            };
            doc.LoadHtml(html);
            var node = doc.DocumentNode.SelectSingleNode("/table[1]/tbody[1]/tr[7]/td[1]/div[1]/table[2]/tbody[1]/tr[6]/td[7]/text()[1]");
            Console.WriteLine(node.InnerText);
        }
    }
}

Text NodeName: #text
InnerText: 17:40
XPath: /table[1]/tbody[1]/tr[7]/td[1]/div[1]/table[2]/tbody[1]/tr[6]/td[7]/text()[1]