private void ScanAlpha()
{
char cBegin = Convert.ToChar(txtAlphaBegin.Text);
char cEnd = Convert.ToChar(txtAlphaEnd.Text);
int begin = Convert.ToInt32(cBegin);
int end = Convert.ToInt32(cEnd);
string url = cboUrl.Text;
string tempUrl = url;
string path = PathSelect();
for (int i = begin; i <= end; i++)
{
tempUrl = tempUrl.Replace("{A}", Convert.ToChar(i).ToString());
//Get Term
// get info from hsx.vn
if (!WebLib.HttpPost(tempUrl, path, i))
{
txtComment.AppendText("Không lấy được dữ liệu từ" + tempUrl + " !");
txtComment.AppendText(Environment.NewLine);
}
else
{
txtComment.AppendText("Lấy xong dữ liệu từ " + tempUrl + "!");
txtComment.AppendText(Environment.NewLine);
}
tempUrl = url;
}
}
private void ScanGoolge()
{
string insite = cboUrl.Text;
string url = "https://www.google.com/search?q=" + txtGoolgeSearch.Text + "+site:" + insite + "&ie=utf-8&oe=utf-8";
//get search google content
string content;
int pos1, pos2, pos3;
string tempUrl, tempContent;
string path = PathSelect();
//click next page
string pageBegin = "<a class=\"fl\" href=\"/search?q=";
string pageEnd;
string search = "url?q=" + insite;
int page = 1, i = 1;
while ((content = RequestContent(url)) != "")
{
tempContent = content;
while ((pos1 = tempContent.IndexOf(search) + 6) != -1)
{
if ((pos2 = tempContent.IndexOf(".html", pos1)) == -1)
{
break;
}
tempUrl = tempContent.Substring(pos1, pos2 - pos1 + 5);
tempContent = tempContent.Substring(pos2, tempContent.Length - pos2);
if (tempUrl.StartsWith(insite))
{
content = tempContent;
if (!WebLib.HttpPost(tempUrl, path, i))
{
txtComment.AppendText("Không lấy được dữ liệu từ" + tempUrl + " !");
txtComment.AppendText(Environment.NewLine);
}
else
{
txtComment.AppendText("Lấy xong dữ liệu từ " + tempUrl + "!");
txtComment.AppendText(Environment.NewLine);
}
i++;
}
}
url = "https://www.google.com/search?q=" + txtGoolgeSearch.Text + "+site:" + insite + "&ie=utf-8&oe=utf-8&start=" + (10 * page);
page++;
if (page == 100)
{
break;
}
//if (page > 1)
//{
// pageEnd = "start=" + 10 * (page - 1) + "&sa=N";
// content = content.Substring(content.IndexOf(pageEnd) + pageEnd.Length);
//}
//pos1 = content.IndexOf(pageBegin) + pageBegin.Length - 10;
//if (pos1 == -1) return;
//pageEnd = "start=" + 10 * page + "&sa=N";
//pos2 = content.IndexOf(pageEnd, pos1) + pageEnd.Length;
//pos3 = content.IndexOf("\">", pos2);
//page++;
//url = "https://www.google.com" + content.Substring(pos1 , pos3 - pos1);
}
}
