//private string getTime()
//{
// string htmlResult = fetchedRes.InnerHtml.ReplaceFirst("<strong>기상실황표</strong>", "");
// Console.WriteLine("time result: " + htmlResult);
// htmlResult = htmlResult.Remove(htmlResult.Length, 3);
// return htmlResult;
// //DateTime local = DateTime.Now;
// //string localDateStr = local.ToString("yyyy-MM-dd-HH") + ".00";
// //return localDateStr;
//}
private void WriteTheLists(List <CityDetail> thelist)
{
//think about this later
//take only year, month, and day
string foldername = currentDateTime.Remove(currentDateTime.Length - 3, 3);
string outputpath = Directory.GetCurrentDirectory() + "\\output\\" + foldername + "\\";
bool exists = System.IO.Directory.Exists(outputpath);
if (!exists)
{
System.IO.Directory.CreateDirectory(outputpath);
}
string filename = outputpath + currentDateTime + ".00.txt";
//string filename = getTime() + ".txt";
//string[] arrFinalStr = new string[thelist.Count()];
List <string> finalStr = new List <string>();
foreach (var list in thelist)
{
finalStr.Add(list.PrintInString(";"));
}
File.WriteAllLines(filename, finalStr.ToArray());
RBoxDisplay.AppendText($"{"\nsaved in:" + outputpath + "\n\n"}");
}
private void DoScraping(IHtmlDocument document)
{
IEnumerable <IElement> articleLink = null;
articleLink = document.All.Where(x => x.ClassName == "table-col");
//articleLink = document.QuerySelectorAll("tbody");
if (articleLink.Any())
{
PrintResults(articleLink);
}
else
{
RBoxDisplay.AppendText($"{"-----------------------\nERROR: No data fetched, please check your internet connection\n\n"}");
}
}
private void UseLocalData()
{
//string thispath = Directory.GetCurrentDirectory();
//string filename = thispath + "\\output\\" + currentDateTime + ".00.txt";
string foldername = currentDateTime.Remove(currentDateTime.Length - 3, 3);
string outputpath = Directory.GetCurrentDirectory() + "\\output\\" + foldername + "\\";
outputpath = outputpath + currentDateTime + ".00.txt";
string readText = File.ReadAllText(outputpath);
string[] collectedData = splitLocalFileRes(readText);
RBoxDisplay.AppendText($"{"Fetch data from :" + outputpath + "\n\n"}");
RBoxDisplay.AppendText($"{"-----FETCHED FROM LOCAL DATA-----\n"}");
foreach (var cd in collectedData)
{
RBoxDisplay.AppendText($"{cd}");
}
}
public void PrintResults(IEnumerable <IElement> articleLink)
{
List <CityDetail> temporary_data = new List <CityDetail>();
foreach (var element in articleLink)
{
//if (element.InnerHtml.Contains("tbody"))
//{
// for(var i = 0, row; row = element.ChildElementCount())
//}
//Console.WriteLine("scrape result:" + element.InnerHtml);
int itt = 0;
List <string> header_str = new List <string>();
//Console.WriteLine("scrape result:" + element.FirstElementChild);
foreach (var b in element.Children)
{
Console.WriteLine("itterate: " + itt);
//table header
if (itt == 2)
{
//only run once, idk why. oh beacuse of the upper head lol
bool therealbody = false;
foreach (var c in b.Children)
{
if (!therealbody)
{
therealbody = true;
continue;
}
CityDetail temp_detail = new CityDetail();
//this is the column
foreach (var d in c.Children)
{
//Console.WriteLine("even this is c: " + d.InnerHtml);
d.InnerHtml = d.InnerHtml.Replace("<br>", " "); // doesnt work, just dont forget to do this
d.InnerHtml = d.InnerHtml.Replace(" ", "");
//string tempstr = d.InnerHtml;
//int location = d.InnerHtml.IndexOf(" ");
//if(location > -1)
//{
// d.InnerHtml = d.InnerHtml.Remove(location, 6);
//}
//d.i = d.InnerHtml.IndexOf("<br>")
//a.UniversalDetail.Add(d.InnerHtml);
temp_detail.UniversalDetail.Add(d.InnerHtml.HtmlEncode(Encoding.UTF8));
//string temp_str = d.InnerHtml.HtmlEncode(Encoding.UTF8);
header_str.Add(d.InnerHtml.ToString());
}
temporary_data.Add(temp_detail);
}
//int a = 10;
}
//table content
else if (itt == 3)
{
foreach (var c in b.Children)
{
//int itt1 = 0;
CityDetail temp_detail = new CityDetail();
bool justheader = false;
foreach (var d in c.Children)
{
//Console.WriteLine("even this is d: " + d.InnerHtml);
if (!justheader)
{
justheader = true;
foreach (var e in d.Children)
{
d.InnerHtml = d.InnerHtml.Replace(" ", "-");
//Console.WriteLine("just for city: " + e.InnerHtml);
temp_detail.UniversalDetail.Add(e.InnerHtml.HtmlEncode(Encoding.UTF8));
}
continue;
}
d.InnerHtml = d.InnerHtml.Replace(" ", "-");
//Console.WriteLine("even this is d: " + d.InnerHtml);
temp_detail.UniversalDetail.Add(d.InnerHtml.HtmlEncode(Encoding.UTF8));
//Console.WriteLine("itt1: " + itt1++);
}
temporary_data.Add(temp_detail);
}
}
itt++;
}
}
int itterator_helper = 0;
RBoxDisplay.AppendText($"{"-----FETCHED FROM ONLINE-----\n"}");
foreach (var q in temporary_data)
{
string disp;
disp = "\n" + itterator_helper + ". " + q.PrintInString(";");
RBoxDisplay.AppendText($"{disp}");
itterator_helper++;
}
//scrapping done. now move them into file
WriteTheLists(temporary_data);
}
