c# xml 转 html,c#中如何将html中的table转化为xml

///

/// 解析Xml文件的帮助类

///

public class XMLHelper

{

///

/// 有效名称的正则表达式

///

static string validName = @"^[^\$\/;""\!#\)\.]+$";

#region CovertHtmlToXml

///

/// 转换html源码为xml格式

///

/// html源码

/// xml字符串

/// 需转换的标记名

public static string CovertHtmlToXml(string html, string targetTag)

{

try

{

XmlDocument doc = new XmlDocument();

XmlNode xmlDeclaration = doc.CreateXmlDeclaration("1.0", "utf-8", null);

doc.AppendChild(xmlDeclaration);

// 借助htmlparser解析html内容

Parser parser = Parser.CreateParser(html, "GBK");

// 筛选出指定的节点

TagNameFilter tnf = new TagNameFilter(targetTag);

NodeList nodes = parser.Parse(tnf);

// 创建根节点

XmlElement root = doc.CreateElement("Tags");

TagNode tagNode = null;

Hashtable ht = null;

XmlAttribute attr = null;

XmlElement parent = null;

for (int i = 0; i < nodes.Size(); i++)

{

tagNode = nodesas TagNode;

parent = doc.CreateElement(tagNode.TagName);

// 添加属性

ht = tagNode.Attributes;

foreach (DictionaryEntry ent in ht)

{

// 查看属性名是否合法

if (Regex.IsMatch(ent.Key.ToString(), validName))

{

attr = doc.CreateAttribute(ent.Key.ToString());

attr.Value = ent.Value.ToString();

parent.Attributes.Append(attr);

}

}// end foreach (DictionaryEntry ent in ht)

AppendChild(tagNode, parent, doc);

root.AppendChild(parent);

}

doc.AppendChild(root);

return doc.OuterXml;

//throw new Exception("给定的html文本必须至少包含一个" + targetTag + "节点");

}

catch (Exception ex)

{

throw new Exception("转换html内容出错:" + ex.Message);

}

}

///

/// 添加子节点

///

/// Html的父节点

/// Xml的父节点

/// Xml文档对象

private static void AppendChild(INode tagNode, XmlNode parent, XmlDocument doc)

{

INode node = null;

XmlNode xmlNode = null;

XmlAttribute attr = null;

Hashtable ht = null;

// 判断是否包含子节点

if (tagNode.Children != null && tagNode.Children.Size() > 0)

{

for (int i = 0; i < tagNode.Children.Size(); i++)

{

node = tagNode.Children;

xmlNode = null;

attr = null;

ht = null;

// 如果是html标记节点

if (node is TagNode)

{

TagNode tn = node as TagNode;

if (Regex.IsMatch(tn.TagName, validName))

{

xmlNode = doc.CreateElement(tn.TagName);

// 添加属性

ht = tn.Attributes;

foreach (DictionaryEntry ent in ht)

{

// 查看属性名是否合法

if (Regex.IsMatch(ent.Key.ToString(), validName))

{

attr = doc.CreateAttribute(ent.Key.ToString());

attr.Value = ent.Value.ToString();

xmlNode.Attributes.Append(attr);

}

}

}

}

// 如果是文本节点

if (node is TextNode)

{

xmlNode = doc.CreateTextNode((node as TextNode).ToPlainTextString());

}

if (xmlNode != null)

{

parent.AppendChild(xmlNode);

AppendChild(node, xmlNode, doc);

}

}

}

}

#endregion

}