using System;

using HtmlAgilityPack;

using System.Text.RegularExpressions;

public class Program

{

    public static void Main()

    {

        string newspaper_url = "https://www.hk01.com";

        HtmlWeb hw = new HtmlWeb();

        HtmlDocument doc = new HtmlDocument();

        //doc = hw.Load(@"https://www.hk01.com/channel/310/%E6%94%BF%E6%83%85");  //load 

        doc = hw.Load(@"https://www.hk01.com/search?q=%E3%80%8A%E6%97%85%E9%81%8A%E6%A5%AD%E6%A2%9D%E4%BE%8B%E8%8D%89%E6%A1%88%E3%80%8B"); //load  《旅遊業條例草案》search result

        string str_prev_link = "";

        //class contains sc-bwzfXH hbZeap

        var body = doc.DocumentNode.SelectNodes("//div[@class='sc-bdVaJa jpNXjy']");

        foreach (HtmlNode divnode in body)

        {

            string theclass = divnode.GetAttributeValue("class",string.Empty);

            string keyclass = "sc-bdVaJa jpNXjy";

            Console.WriteLine(theclass);

            string the_html=divnode.InnerHtml;

            Console.WriteLine(the_html);

            int offset = the_html.IndexOf("\"");

            offset = the_html.IndexOf("\"", offset+1);

            int href_start = the_html.IndexOf("\"", offset+1);

            int href_end = the_html.IndexOf("\"", href_start+1);

//var result = myString.IndexOf(':', offset+1);

            Console.WriteLine(href_start + " - " + href_end);

            string art_url = the_html.Substring(href_start, href_end - href_start+1).Replace("\"","");

            Console.WriteLine(art_url);

            //string input= "<th><a href=\"Boot_53.html\">135 Boot</a></th>";

            /*

string regex= "<a\s+(?:[^>]*?\s+)?href="([^"]*)"";

Match match = Regex.Match(the_html, regex);

if (match.Success)

{

    string link= match.Groups[1].Value;

    Console.WriteLine(link);

}*/

            //Console.WriteLine(divnode.Descendants("a").First());

            //HtmlAttribute att = divnode.Attributes["href"];

      //Console.WriteLine(new Regex(@"(?<=[\?&]id=)\d+(?=\&|\#|$)").Match(divnode.InnerHtml).Value);

            if(theclass == keyclass){

        foreach (HtmlNode link in divnode.SelectNodes("//a[@href]"))

        {

            //string str_link = link.GetAttributeValue("href", string.Empty);

            string str_link = art_url;

            //if(str_link.StartsWith("/政情")){

                //Console.WriteLine(str_link);

                if(str_link != str_prev_link){

                    HtmlDocument doc2 = new HtmlDocument();

                    Console.WriteLine("Surf:" + newspaper_url + str_link + "...");

                    doc2 = hw.Load(newspaper_url+str_link); //<a href="/tag/

                    foreach (HtmlNode link2 in doc2.DocumentNode.SelectNodes("//a[@href]"))

                    {

                        string str_link2 = link2.GetAttributeValue("href", string.Empty);

                        if(str_link2.StartsWith("/tag")){

                            Console.WriteLine(str_link2 + " - " + link2.InnerText);

                        }

                    }

                    str_prev_link = str_link;

                }

            //} //end "/政情"

        }

            }

        }

    }

}