获取指定URL网页的HTML代码

最新推荐文章于 2021-06-05 00:31:41 发布

yuxuan0476

最新推荐文章于 2021-06-05 00:31:41 发布

阅读量562

点赞数

分类专栏： .NET C#类文章标签： url html string regex

.NET C#类专栏收录该内容

111 篇文章 0 订阅

订阅专栏

/// <summary>
   /// 获取指定URL网页的HTML代码
   /// </summary>
   /// <param name="URL">URL</param>
   /// <param name="charset">网页编码如 gb2312,utf-8</param>
   /// <returns>HTML代码</returns>
   public static string GetPageSource(string URL,string charset)
   {
    Uri uri =new Uri(URL);

HttpWebRequest hwReq = (HttpWebRequest)WebRequest.Create(uri);
HttpWebResponse hwRes = (HttpWebResponse)hwReq.GetResponse();

hwReq.Method = "Get";

hwReq.KeepAlive = false;

StreamReader reader = new StreamReader(hwRes.GetResponseStream(),System.Text.Encoding.GetEncoding(charset));

    return reader.ReadToEnd();
   }
//
   public static MatchCollection GetNewsLink(string Htmlstr,string Restr)
   {
    Regex reNewsLink = new Regex(Restr,System.Text.RegularExpressions.RegexOptions.Singleline);
    MatchCollection tem = reNewsLink.Matches(Htmlstr);
    return tem;
   }