HTML解析利器HtmlAgilityPack

来源：互联网发布：python is empty 编辑：程序博客网时间：2024/05/21 19:32

用HtmlAgilityPack，自行网上下载dll文件，获取：

C# code

        HttpWebRequest httpWebRequest = WebRequest.Create(@"http://www.sooker.com/xuexiao/") as HttpWebRequest;        HttpWebResponse httpWebResponse = httpWebRequest.GetResponse() as HttpWebResponse;        Stream stream = httpWebResponse.GetResponseStream();        StreamReader reader = new StreamReader(stream, Encoding.GetEncoding("gb2312"));        string s = reader.ReadToEnd();        reader.Close();        httpWebResponse.Close();        HtmlDocument htmlDoc = new HtmlDocument();        htmlDoc.LoadHtml(s);        HtmlNodeCollection imgs = htmlDoc.DocumentNode.SelectNodes(@"//ul[@class='curriculumUl']/li//div[@class='pic']/a/img");        foreach (HtmlNode img in imgs)            Response.Write(img.Attributes["src"].Value + "<br/>");        HtmlNodeCollection anchors = htmlDoc.DocumentNode.SelectNodes(@"//ul[@class='curriculumUl']/li//a[@class='school-name']");        foreach (HtmlNode anchor in anchors)        {            Response.Write(anchor.Attributes["href"].Value + "<br/>");            Response.Write(anchor.InnerHtml + "<br/>");        }        Response.End();

http://zhoufoxcn.blog.51cto.com/792419/595344