关于.net：在C＃中解析html的最佳方法是什么？

using mshtml;
...
object[] oPageText = { html };
HTMLDocument doc = new HTMLDocumentClass();
IHTMLDocument2 doc2 = (IHTMLDocument2)doc;
doc2.write(oPageText);

var wb = new WebBrowser()

var doc = wb.Browser.Document
var elem = doc.GetElementById(elementId);
object obj = elem.DomElement;
System.Reflection.MethodInfo mi = obj.GetType().GetMethod("click");
mi.Invoke(obj, new object[0]);

        IEnumerable<XNode> auctionNodes = Majestic12ToXml.Majestic12ToXml.ConvertNodesToXml(byteArrayOfAuctionHtml);

        foreach (XElement anchorTag in auctionNodes.OfType<XElement>().DescendantsAndSelf("a")) {

            if (anchorTag.Attribute("href") == null)
                continue;

            Console.WriteLine(anchorTag.Attribute("href").Value);
        }

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using Majestic12;
using System.IO;
using System.Xml.Linq;
using System.Diagnostics;
using System.Text.RegularExpressions;

namespace Majestic12ToXml {
public class Majestic12ToXml {

    static public IEnumerable<XNode> ConvertNodesToXml(byte[] htmlAsBytes) {

        HTMLparser parser = OpenParser();
        parser.Init(htmlAsBytes);

        XElement currentNode = new XElement("document");

        HTMLchunk m12chunk = null;

        int xmlnsAttributeIndex = 0;
        string originalHtml ="";

        while ((m12chunk = parser.ParseNext()) != null) {

            try {

                Debug.Assert(!m12chunk.bHashMode);  // popular default for Majestic-12 setting

                XNode newNode = null;
                XElement newNodesParent = null;

                switch (m12chunk.oType) {
                    case HTMLchunkType.OpenTag:

                        // Tags are added as a child to the current tag,
                        // except when the new tag implies the closure of
                        // some number of ancestor tags.

                        newNode = ParseTagNode(m12chunk, originalHtml, ref xmlnsAttributeIndex);

                        if (newNode != null) {
                            currentNode = FindParentOfNewNode(m12chunk, originalHtml, currentNode);

                            newNodesParent = currentNode;

                            newNodesParent.Add(newNode);

                            currentNode = newNode as XElement;
                        }

                        break;

                    case HTMLchunkType.CloseTag:

                        if (m12chunk.bEndClosure) {

                            newNode = ParseTagNode(m12chunk, originalHtml, ref xmlnsAttributeIndex);

                            if (newNode != null) {
                                currentNode = FindParentOfNewNode(m12chunk, originalHtml, currentNode);

                                newNodesParent = currentNode;
                                newNodesParent.Add(newNode);
                            }
                        }
                        else {
                            XElement nodeToClose = currentNode;

                            string m12chunkCleanedTag = CleanupTagName(m12chunk.sTag, originalHtml);

                            while (nodeToClose != null && nodeToClose.Name.LocalName != m12chunkCleanedTag)
                                nodeToClose = nodeToClose.Parent;

                            if (nodeToClose != null)
                                currentNode = nodeToClose.Parent;

                            Debug.Assert(currentNode != null);
                        }

                        break;

                    case HTMLchunkType.Script:

                        newNode = new XElement("script","REMOVED");
                        newNodesParent = currentNode;
                        newNodesParent.Add(newNode);
                        break;

                    case HTMLchunkType.Comment:

                        newNodesParent = currentNode;

                        if (m12chunk.sTag =="!--")
                            newNode = new XComment(m12chunk.oHTML);
                        else if (m12chunk.sTag =="![CDATA[")
                            newNode = new XCData(m12chunk.oHTML);
                        else
                            throw new Exception("Unrecognized comment sTag");

                        newNodesParent.Add(newNode);

                        break;

                    case HTMLchunkType.Text:

                        currentNode.Add(m12chunk.oHTML);
                        break;

                    default:
                        break;
                }
            }
            catch (Exception e) {
                var wrappedE = new Exception("Error using Majestic12.HTMLChunk, reason:" + e.Message, e);

                // the original html is copied for tracing/debugging purposes
                originalHtml = new string(htmlAsBytes.Skip(m12chunk.iChunkOffset)
                    .Take(m12chunk.iChunkLength)
                    .Select(B => (char)B).ToArray());

                wrappedE.Data.Add("source", originalHtml);

                throw wrappedE;
            }
        }

        while (currentNode.Parent != null)
            currentNode = currentNode.Parent;

        return currentNode.Nodes();
    }

    static XElement FindParentOfNewNode(Majestic12.HTMLchunk m12chunk, string originalHtml, XElement nextPotentialParent) {

        string m12chunkCleanedTag = CleanupTagName(m12chunk.sTag, originalHtml);

        XElement discoveredParent = null;

        // Get a list of all ancestors
        List<XElement> ancestors = new List<XElement>();
        XElement ancestor = nextPotentialParent;
        while (ancestor != null) {
            ancestors.Add(ancestor);
            ancestor = ancestor.Parent;
        }

        // Check if the new tag implies a previous tag was closed.
        if ("form" == m12chunkCleanedTag) {

            discoveredParent = ancestors
                .Where(XE => m12chunkCleanedTag == XE.Name)
                .Take(1)
                .Select(XE => XE.Parent)
                .FirstOrDefault();
        }
        else if ("td" == m12chunkCleanedTag) {

            discoveredParent = ancestors
                .TakeWhile(XE =>"tr" != XE.Name)
                .Where(XE => m12chunkCleanedTag == XE.Name)
                .Take(1)
                .Select(XE => XE.Parent)
                .FirstOrDefault();
        }
        else if ("tr" == m12chunkCleanedTag) {

            discoveredParent = ancestors
                .TakeWhile(XE => !("table" == XE.Name
                                    ||"thead" == XE.Name
                                    ||"tbody" == XE.Name
                                    ||"tfoot" == XE.Name))
                .Where(XE => m12chunkCleanedTag == XE.Name)
                .Take(1)
                .Select(XE => XE.Parent)
                .FirstOrDefault();
        }
        else if ("thead" == m12chunkCleanedTag
                  ||"tbody" == m12chunkCleanedTag
                  ||"tfoot" == m12chunkCleanedTag) {


            discoveredParent = ancestors
                .TakeWhile(XE =>"table" != XE.Name)
                .Where(XE => m12chunkCleanedTag == XE.Name)
                .Take(1)
                .Select(XE => XE.Parent)
                .FirstOrDefault();
        }

        return discoveredParent ?? nextPotentialParent;
    }

    static string CleanupTagName(string originalName, string originalHtml) {

        string tagName = originalName;

        tagName = tagName.TrimStart(new char[] { '?' });  // for nodes <?xml >

        if (tagName.Contains(':'))
            tagName = tagName.Substring(tagName.LastIndexOf(':') + 1);

        return tagName;
    }

    static readonly Regex _startsAsNumeric = new Regex(@"^[0-9]", RegexOptions.Compiled);

    static bool TryCleanupAttributeName(string originalName, ref int xmlnsIndex, out string result) {

        result = null;
        string attributeName = originalName;

        if (string.IsNullOrEmpty(originalName))
            return false;

        if (_startsAsNumeric.IsMatch(originalName))
            return false;

        //
        // transform xmlns attributes so they don't actually create any XML namespaces
        //
        if (attributeName.ToLower().Equals("xmlns")) {

            attributeName ="xmlns_" + xmlnsIndex.ToString(); ;
            xmlnsIndex++;
        }
        else {
            if (attributeName.ToLower().StartsWith("xmlns:")) {
                attributeName ="xmlns_" + attributeName.Substring("xmlns:".Length);
            }  

            //
            // trim trailing "
            //
            attributeName = attributeName.TrimEnd(new char[] { '"' });

            attributeName = attributeName.Replace(":","_");
        }

        result = attributeName;

        return true;
    }

    static Regex _weirdTag = new Regex(@"^<!\[.*\]>$");       // matches"<![if !supportEmptyParas]>"
    static Regex _aspnetPrecompiled = new Regex(@"^<%.*%>$"); // matches"<%@ ... %>"
    static Regex _shortHtmlComment = new Regex(@"^<!-.*->$"); // matches"<!-Extra_Images->"

    static XElement ParseTagNode(Majestic12.HTMLchunk m12chunk, string originalHtml, ref int xmlnsIndex) {

        if (string.IsNullOrEmpty(m12chunk.sTag)) {

            if (m12chunk.sParams.Length > 0 && m12chunk.sParams[0].ToLower().Equals("doctype"))
                return new XElement("doctype");

            if (_weirdTag.IsMatch(originalHtml))
                return new XElement("REMOVED_weirdBlockParenthesisTag");

            if (_aspnetPrecompiled.IsMatch(originalHtml))
                return new XElement("REMOVED_ASPNET_PrecompiledDirective");

            if (_shortHtmlComment.IsMatch(originalHtml))
                return new XElement("REMOVED_ShortHtmlComment");

            // Nodes like"<br" will end up with a m12chunk.sTag==""...  We discard these nodes.
            return null;
        }

        string tagName = CleanupTagName(m12chunk.sTag, originalHtml);

        XElement result = new XElement(tagName);

        List<XAttribute> attributes = new List<XAttribute>();

        for (int i = 0; i < m12chunk.iParams; i++) {

            if (m12chunk.sParams[i] =="<!--") {

                // an HTML comment was embedded within a tag.  This comment and its contents
                // will be interpreted as attributes by Majestic-12... skip this attributes
                for (; i < m12chunk.iParams; i++) {

                    if (m12chunk.sTag =="--" || m12chunk.sTag =="-->")
                        break;
                }

                continue;
            }

            if (m12chunk.sParams[i] =="?" && string.IsNullOrEmpty(m12chunk.sValues[i]))
                continue;

            string attributeName = m12chunk.sParams[i];

            if (!TryCleanupAttributeName(attributeName, ref xmlnsIndex, out attributeName))
                continue;

            attributes.Add(new XAttribute(attributeName, m12chunk.sValues[i]));
        }

        // If attributes are duplicated with different values, we complain.
        // If attributes are duplicated with the same value, we remove all but 1.
        var duplicatedAttributes = attributes.GroupBy(A => A.Name).Where(G => G.Count() > 1);

        foreach (var duplicatedAttribute in duplicatedAttributes) {

            if (duplicatedAttribute.GroupBy(DA => DA.Value).Count() > 1)
                throw new Exception("Attribute value was given different values");

            attributes.RemoveAll(A => A.Name == duplicatedAttribute.Key);
            attributes.Add(duplicatedAttribute.First());
        }

        result.Add(attributes);

        return result;
    }

    static HTMLparser OpenParser() {
        HTMLparser oP = new HTMLparser();

        // The code+comments in this function are from the Majestic-12 sample documentation.

        // ...

        // This is optional, but if you want high performance then you may
        // want to set chunk hash mode to FALSE. This would result in tag params
        // being added to string arrays in HTMLchunk object called sParams and sValues, with number
        // of actual params being in iParams. See code below for details.
        //
        // When TRUE (and its default) tag params will be added to hashtable HTMLchunk (object).oParams
        oP.SetChunkHashMode(false);

        // if you set this to true then original parsed HTML for given chunk will be kept -
        // this will reduce performance somewhat, but may be desireable in some cases where
        // reconstruction of HTML may be necessary
        oP.bKeepRawHTML = false;

        // if set to true (it is false by default), then entities will be decoded: this is essential
        // if you want to get strings that contain final representation of the data in HTML, however
        // you should be aware that if you want to use such strings into output HTML string then you will
        // need to do Entity encoding or same string may fail later
        oP.bDecodeEntities = true;

        // we have option to keep most entities as is - only replace stuff like  
        // this is called Mini Entities mode - it is handy when HTML will need
        // to be re-created after it was parsed, though in this case really
        // entities should not be parsed at all
        oP.bDecodeMiniEntities = true;

        if (!oP.bDecodeEntities && oP.bDecodeMiniEntities)
            oP.InitMiniEntities();

        // if set to true, then in case of Comments and SCRIPT tags the data set to oHTML will be
        // extracted BETWEEN those tags, rather than include complete RAW HTML that includes tags too
        // this only works if auto extraction is enabled
        oP.bAutoExtractBetweenTagsOnly = true;

        // if true then comments will be extracted automatically
        oP.bAutoKeepComments = true;

        // if true then scripts will be extracted automatically:
        oP.bAutoKeepScripts = true;

        // if this option is true then whitespace before start of tag will be compressed to single
        // space character in string:"", if false then full whitespace before tag will be returned (slower)
        // you may only want to set it to false if you want exact whitespace between tags, otherwise it is just
        // a waste of CPU cycles
        oP.bCompressWhiteSpaceBeforeTag = true;

        // if true (default) then tags with attributes marked as CLOSED (/ at the end) will be automatically
        // forced to be considered as open tags - this is no good for XML parsing, but I keep it for backwards
        // compatibility for my stuff as it makes it easier to avoid checking for same tag which is both closed
        // or open
        oP.bAutoMarkClosedTagsWithParamsAsOpen = false;

        return oP;
    }
}
}

using System;
using System.Collections.Generic;
using System.Text;
using System.Windows.Forms;
using System.Threading;

class ParseHTML
{
    public ParseHTML() { }
    private string ReturnString;

    public string doParsing(string html)
    {
        Thread t = new Thread(TParseMain);
        t.ApartmentState = ApartmentState.STA;
        t.Start((object)html);
        t.Join();
        return ReturnString;
    }

    private void TParseMain(object html)
    {
        WebBrowser wbc = new WebBrowser();
        wbc.DocumentText ="feces of a dummy";        //;magic words        
        HtmlDocument doc = wbc.Document.OpenNew(true);
        doc.Write((string)html);
        this.ReturnString = doc.Body.InnerHtml +" do here something";
        return;
    }
}

string myhtml ="<HTML><BODY>This is a new HTML document.</BODY></HTML>";
Console.WriteLine("before:" + myhtml);
myhtml = (new ParseHTML()).doParsing(myhtml);
Console.WriteLine("after:" + myhtml);

script SS_URLs.txt URL("http://stackoverflow.com/questions/56107/what-is-the-best-way-to-parse-html-in-c")

http://sstatic.net/so/all.css
http://sstatic.net/so/favicon.ico
http://sstatic.net/so/apple-touch-icon.webp
.
.
.