几个C#关于Html解析的类

html-css08

几个C#关于Html解析的类,第1张

第一种方法:

用System.Net.WebClient下载Web Page存到本地文件或者String中,用正则表达式来分析。这个方法可以用在Web Crawler等需要分析很多Web Page的应用中。

using System

using System.Net

using System.Text

using System.Text.RegularExpressions

namespace HttpGet

{

    class Class1

    {

        [STAThread]

        static void Main(string[] args)

        {

            System.Net.WebClient client = new WebClient()

            byte[] page = client.DownloadData("http://www.google.com")

            string content = System.Text.Encoding.UTF8.GetString(page)

            string regex = "href=[\\\"\\\'](http:\\/\\/|\\.\\/|\\/)?\\w+(\\.\\w+)*(\\/\\w+(\\.\\w+)?)*(\\/|\\?\\w*=\\w*(&\\w*=\\w*)*)?[\\\"\\\']"

            Regex re = new Regex(regex)

            MatchCollection matches = re.Matches(content)

            System.Collections.IEnumerator enu = matches.GetEnumerator()

            while (enu.MoveNext() && enu.Current != null)

            {

                Match match = (Match)(enu.Current)

                Console.Write(match.Value + "\r\n")

            }

        }

    }

}

第二种方法:

利用Winista.Htmlparser.Net 解析Html。这是.NET平台下解析Html的开源代码,网上有源码下载

using System

using System.Collections.Generic

using System.ComponentModel

using System.Data

using System.Drawing

using System.Linq

using System.Text

using System.Windows.Forms

using Winista.Text.HtmlParser

using Winista.Text.HtmlParser.Lex

using Winista.Text.HtmlParser.Util

using Winista.Text.HtmlParser.Tags

using Winista.Text.HtmlParser.Filters

namespace HTMLParser

{

    public partial class Form1 : Form

    {

        public Form1()

        {

            InitializeComponent()

            AddUrl()

        }

        private void btnParser_Click(object sender, EventArgs e)

        {

            #region 获得网页的html

            try

            {

                txtHtmlWhole.Text = ""

                string url = CBUrl.SelectedItem.ToString().Trim()

                System.Net.WebClient aWebClient = new System.Net.WebClient()

                aWebClient.Encoding = System.Text.Encoding.Default

                string html = aWebClient.DownloadString(url)

                txtHtmlWhole.Text = html

            }

            catch (Exception ex)

            {

                MessageBox.Show(ex.Message)

            }

            #endregion

            #region 分析网页html节点

            Lexer lexer = new Lexer(this.txtHtmlWhole.Text)

            Parser parser = new Parser(lexer)

            NodeList htmlNodes = parser.Parse(null)

            this.treeView1.Nodes.Clear()

            this.treeView1.Nodes.Add("root")

            TreeNode treeRoot = this.treeView1.Nodes[0]

            for (int i = 0 i < htmlNodes.Count i++)

            {

                this.RecursionHtmlNode(treeRoot, htmlNodes[i], false)

            }

            #endregion

        }

        private void RecursionHtmlNode(TreeNode treeNode, INode htmlNode, bool siblingRequired)

        {

            if (htmlNode == null || treeNode == null) return

            TreeNode current = treeNode

            TreeNode content 

            //current node

            if (htmlNode is ITag)

            {

                ITag tag = (htmlNode as ITag)

                if (!tag.IsEndTag())

                {

                    string nodeString = tag.TagName

                    if (tag.Attributes != null && tag.Attributes.Count > 0)

                    {

                        if (tag.Attributes["ID"] != null)

                        {

                            nodeString = nodeString + " { id=\"" + tag.Attributes["ID"].ToString() + "\" }"

                        }

                        if (tag.Attributes["HREF"] != null)

                        {

                            nodeString = nodeString + " { href=\"" + tag.Attributes["HREF"].ToString() + "\" }"

                        }

                    }

                    

                    current = new TreeNode(nodeString)

                    treeNode.Nodes.Add(current)

                }

            }

            //获取节点间的内容

            if (htmlNode.Children != null && htmlNode.Children.Count > 0)

            {

                this.RecursionHtmlNode(current, htmlNode.FirstChild, true)

                content = new TreeNode(htmlNode.FirstChild.GetText())

                treeNode.Nodes.Add(content)

            }

            //the sibling nodes

            if (siblingRequired)

            {

                INode sibling = htmlNode.NextSibling

                while (sibling != null)

                {

                    this.RecursionHtmlNode(treeNode, sibling, false)

                    sibling = sibling.NextSibling

                }

            }

        }

        private void AddUrl()

        {

            CBUrl.Items.Add("http://www.xxx.com")

        }

    }

}

比如:

比如,数学符号 , ,可以直接获得:

escape 将特殊字符 & , < 和 > 替换为HTML安全序列。如果可选的 flags quote 为 True (默认值),则还会翻译引号字符,包括双引号( " )和单引号( ' )字符。

将字符串 s 中的所有命名和数字字符引用 (例如 >, >, >) 转换为相应的 Unicode 字符。此函数使用 HTML 5 标准为有效和无效字符引用定义的规则,以及 HTML 5 命名字符引用列表 。

这个模块定义了一个 HTMLParser 类,为 HTML(超文本标记语言)和 XHTML 文本文件解析提供基础。

class html.parser.HTMLParser(*, convert_charrefs=True) 创建一个能解析无效标记的解析器实例。查找标签(tags)和其他标记(markup)并调用 handler 函数。

用法:

通过调用 self.handle_starttag 处理开始标签,或通过调用 self.handle_startendtag 处理结束标签。标签之间的数据通过以 data 为参数调用 self.handle_data 从解析器传递到派生类(数据可以分成任意块)。如果 convert_charrefs 为 True ,则将字符引用自动转换为相应的 Unicode 字符(并且 self.handle_data 不再拆分成块),否则通过调用带有字符串的 self.handle_entityref 或 self.handle_charref 来传递它们以分别包含命名或数字引用作为参数。如果 convert_charrefs 为 True (默认值),则所有字符引用( script / style 元素中的除外)都会自动转换为相应的 Unicode 字符。

一个 HTMLParser 类的实例用来接受 HTML 数据,并在标记开始、标记结束、文本、注释和其他元素标记出现的时候调用对应的方法。要实现具体的行为,请使用 HTMLParser 的子类并重载其方法。

这个解析器不检查结束标记是否与开始标记匹配,也不会因外层元素完毕而隐式关闭了的元素引发结束标记处理。

下面是简单的 HTML 解析器的一个基本示例,使用 HTMLParser 类,当遇到开始标记、结束标记以及数据的时候将内容打印出来。

输出:

HTMLParser.reset() 重置实例。丢失所有未处理的数据。在实例化阶段被隐式调用。

HTMLParser.feed(data) 填充一些文本到解析器中。如果包含完整的元素,则被处理;如果数据不完整,将被缓冲直到更多的数据被填充,或者 close() 被调用。 data 必须为 str 类型。

HTMLParser.close() 如同后面跟着一个文件结束标记一样,强制处理所有缓冲数据。这个方法能被派生类重新定义,用于在输入的末尾定义附加处理,但是重定义的版本应当始终调用基类 HTMLParser 的 close() 方法。

HTMLParser.getpos() 返回当前行号和偏移值。

HTMLParser.get_starttag_text() 返回最近打开的开始标记中的文本。结构化处理时通常应该不需要这个,但在处理“已部署”的 HTML 或是在以最小改变来重新生成输入时可能会有用处(例如可以保留属性间的空格等)。

下列方法将在遇到数据或者标记元素的时候被调用。他们需要在子类中重载。基类的实现中没有任何实际操作(除了 handle_startendtag() ):

HTMLParser.handle_starttag 这个方法在标签开始的时候被调用(例如: <div id="main">)。 tag 参数是小写的标签名。 attrs 参数是一个 (name, value) 形式的列表,包含了所有在标记的 <> 括号中找到的属性。 name 转换为小写, value 的引号被去除,字符和实体引用都会被替换。比如,对于标签 <a href="https://www.cwi.nl/">,这个方法将以下列形式被调用 handle_starttag('a', [('href', 'https://www.cwi.nl/')]) 。 html.entities 中的所有实体引用,会被替换为属性值。

HTMLParser.handle_endtag(tag) 此方法被用来处理元素的结束标记(例如: </div>)。 tag 参数是小写的标签名。

HTMLParser.handle_startendtag(tag, attrs) 类似于 handle_starttag() , 只是在解析器遇到 XHTML 样式的空标记时被调用( <tag ... />)。这个方法能被需要这种特殊词法信息的子类重载;默认实现仅简单调用 handle_starttag() 和 handle_endtag() 。

HTMLParser.handle_data(data) 这个方法被用来处理任意数据(例如:文本节点和 <script>...</script> 以及 <style>...</style> 中的内容)。

HTMLParser.handle_entityref(name) 这个方法被用于处理 &name 形式的命名字符引用(例如 >),其中 name 是通用的实体引用(例如: 'gt' )。如果 convert_charrefs 为 True,该方法永远不会被调用。

HTMLParser.handle_charref(name) 这个方法被用来处理 NNN 和 NNN 形式的十进制和十六进制字符引用。例如, > 等效的十进制形式为 >,而十六进制形式为 > ;在这种情况下,方法将收到 '62' 或 'x3E' 。如果 convert_charrefs 为 True ,则该方法永远不会被调用。

HTMLParser.handle_comment(data) 这个方法在遇到注释的时候被调用(例如: )。例如, 这个注释会用 ' comment ' 作为参数调用此方法。

Internet Explorer 条件注释(condcoms)的内容也被发送到这个方法,因此,对于 ``,这个方法将接收到 '[if IE 9]>IE9-specific content<![endif]' 。

HTMLParser.handle_decl(decl) 这个方法用来处理 HTML doctype 申明(例如 <!DOCTYPE html>)。 decl 形参为 <!...> 标记中的所有内容(例如: 'DOCTYPE html' )。

HTMLParser.handle_pi(data) 此方法在遇到处理指令的时候被调用。 data 形参将包含整个处理指令。例如,对于处理指令 <?proc color='red'>,这个方法将以 handle_pi("proc color='red'") 形式被调用。它旨在被派生类重载;基类实现中无任何实际操作。

注解: HTMLParser 类使用 SGML 语法规则处理指令。使用 '?' 结尾的 XHTML 处理指令将导致 '?' 包含在 data 中。

HTMLParser.unknown_decl(data) 当解析器读到无法识别的声明时,此方法被调用。 data 形参为 <![...]> 标记中的所有内容。某些时候对派生类的重载很有用。基类实现中无任何实际操作。

因此,我们可以如此定义:

下面介绍如何解析 HTML 文档。

解析一个文档类型声明:

解析一个具有一些属性和标题的元素:

script 和 style 元素中的内容原样返回,无需进一步解析:

解析注释:

解析命名或数字形式的字符引用,并把他们转换到正确的字符(注意:这 3 种转义都是 '>' ):

填充不完整的块给 feed() 执行, handle_data() 可能会多次调用(除非 convert_charrefs 被设置为 True ):

解析无效的 HTML (例如:未引用的属性)也能正常运行:

解析HTML最好的类还是微软自己的

在站内搜索的项目要进行HTML的解析,发现程序运行非常慢,一开始以为是lucene的问题,一测试大吃一惊,每一步lucene中AddDocument等只用了几十毫秒,而HTML解析竟然用了9秒。

日志如下:

2010-06-26 15:51:25,171 [8] DEBUG SearchSite.StartIndex - DownloadString:00:00:00.0482329

2010-06-26 15:51:34,187 [8] DEBUG SearchSite.StartIndex - ThreadParser:00:00:09.0236490

2010-06-26 15:51:34,187 [8] DEBUG SearchSite.StartIndex - DeleteDocuments:00:00:00.0000069

2010-06-26 15:51:34,203 [8] DEBUG SearchSite.StartIndex - AddDocument:00:00:00.0191071

我使用的是Winista.HtmlParser这块网上找到的HTML解析器。遂准备更换解析器,但是用“.Net HTML Parser”在google上寻找,一直没找到合适的解析器,不是太难用,就是性能比Winista.HtmlParser还差,我才想到,微软的MSHTML不就是用来解析HTML的吗?折腾这么一圈却忘了微软自家的东西,微软的东西性能肯定差不了。

添加对Microsoft.mshtml这个程序集的引用,然后编写如下代码:

IHTMLDocument2 doc = new HTMLDocumentClass()

doc.write(new object[]{pageSource})

doc.close()

Title = doc.title

Body = doc.body.innerText

再测试,哇咔咔,快的要命,只有几十毫秒,2000个帖子一会儿就爬完了。

日志如下:

2010-06-26 16:26:35,546 [8] DEBUG SearchSite.StartIndex - DownloadString:00:00:00.0400263

2010-06-26 16:26:35,562 [8] DEBUG SearchSite.StartIndex - ThreadParser:00:00:00.0257731

2010-06-26 16:26:35,562 [8] DEBUG SearchSite.StartIndex - DeleteDocuments:00:00:00.0001913

2010-06-26 16:26:35,578 [8] DEBUG SearchSite.StartIndex - AddDocument:00:00:00.0010881

HTMLDocumentClass的方法比任何一个网上找到的HTML解析器都丰富,想怎么搞就怎么搞,而且调用方法就是操作Dom的方法,完全不用再去学,哇咔咔,爽呆了。

在使用的过程中遇到两个问题:

1、VS2010中引用Microsoft.mshtml之后,要修改这个引用的“嵌入互操作类型”为False。

2、调用doc.write方法的时候必须通过IHTMLDocument2接口来调用,否则报错“错误的类型”,在google上搜“type mismatch HTMLDocument write”