Html Agility Pack 是一個開源的.NET 方案HTML解析器。

開源地址：https://github.com/zzzprojects/html-agility-pack

用法：vs上通過Nuget搜尋Html Agility Pack並安裝

示例程式碼1：

        /// <summary>
        /// 獲取網頁內容
        /// </summary>
        /// <returns></returns>
        private static string GetHtml()
        {
            
 
string html = string.Empty;
            string url = "http://quote.eastmoney.com/stocklist.html";

            using (var client = new HttpClient())
            {
                client.BaseAddress = new Uri(url);

                //關鍵程式碼1：設定請求頭採用GZip和deflate兩種壓縮演算法
                client.DefaultRequestHeaders.Add("
 
Accept-Encoding", "gzip, deflate");
                var response = client.GetAsync(url).Result;

                var fileStream = response.Content.ReadAsStreamAsync().Result;

                //關鍵程式碼2：對檔案流採用GZip演算法解壓
                GZipStream gzip = new GZipStream(fileStream, CompressionMode.Decompress);

                
 
using (StreamReader reader = new StreamReader(gzip, Encoding.GetEncoding("gb2312")))//中文編碼處理
                {
                    html = reader.ReadToEnd();
                    //File.WriteAllText(@"C:\stock.html", reader.ReadToEnd(), Encoding.Default);
                }
            }

            return html;
        }

示例程式碼2：

        static void Main(string[] args)
        {
            string html= GetHtml();

            var doc = new HtmlDocument();
            doc.LoadHtml(html);

            //查詢dom節點div的Id為quotesearch下所有ul下的li下的所有a節點
            var nodes = doc.DocumentNode.SelectNodes("//div[@id='quotesearch']/ul/li/a");
            foreach (var node in nodes)
            {
                var arrays=node.InnerText.Split('(');
                Console.WriteLine(string.Format("股票名稱：{0}，股票程式碼：{1}", arrays[0], arrays[1].Replace(")","")));
            }

            //File.WriteAllText(@"C:\stock.html", matches.ToString(), Encoding.Default);

            Console.ReadKey();
        }