//WORK WITH ROBOTS TXT
public UrlClass checkUrl(String url)
{
UrlClass urlClass = new UrlClass();
url = urlClass.checkUrl(url);
if (url == null)
{
MessageBox.Show("Нет доступа к сайту или не верно введена ссылка");
return(null);
}
var dom = CQ.CreateFromUrl(url);
string text = dom["meta"].ToString();
// MessageBox.Show(text.Contains("charset=\"UTF-8\"").ToString());
if (!text.ToLower().Contains("charset=utf-8"))
{
if (!text.ToLower().Contains("charset=\"utf-8\""))
{
MessageBox.Show("На данной веб-странице не используется UTF-8");
return(null);
}
}
return(urlClass);
}
public void start(String url)
{
UrlClass urlClass = checkUrl(url);
if (urlClass == null)
{
return;
}
UrlClass urlClass2 = new UrlClass(url);
urlClass2.start(1);
}
private void Button_Click2(object sender, RoutedEventArgs e)
{
string url = firstNameTxtBox_Copy2.Text;
DataSet ds = new DataSet();
DataTable dt = new DataTable();
try
{
conn.Open();
using (var cmd = new NpgsqlCommand())
{
cmd.Connection = conn;
DateTime dateTime;
TimeSpan diff;
// List<int> average = new List<int>();
Dictionary <string, string> dic = new Dictionary <string, string>();
cmd.CommandText = "SELECT newdate, link, hash, average FROM webCrawler where link=" + "\'" + url + "\'";
try
{
using (var reader = cmd.ExecuteReader())
{
while (reader.Read())
{
string link = (string)reader["link"];
string hash = (string)reader["hash"];
dic.Add(link, hash);
}
}
foreach (KeyValuePair <string, string> kvp in dic)
{
HtmlDocument doc = new HtmlDocument();
try
{
try
{
try
{
HtmlWeb hw = new HtmlWeb();
doc = hw.Load(kvp.Key);//загрузка html страницы
}
catch
{
using (WebClient wc = new WebClient())
{
byte[] respones = wc.DownloadData(kvp.Key);
String str = UrlClass.Unzip(respones);
doc.LoadHtml(str);
}
}
}
catch (Exception ex)
{
cmd.CommandText = "DELETE FROM webCrawler where link=" + "\'" + kvp.Key + "\'";
cmd.ExecuteNonQuery();
}
UrlClass urlClass = new UrlClass();
String text = urlClass.getTxtFromWebsite(doc);
string hash = HashMD5.MD5Hash(text);
if (kvp.Value != hash)
{
cmd.CommandText = "UPDATE webCrawler set hash=" + "\'" + hash + "\'" + ", text=" + "\'" + text + "\'" + " WHERE link=" + "\'" + kvp.Key + "\'";
cmd.ExecuteNonQuery();
}
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
// Console.WriteLine(hours.ToString());
}
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
}
// since we only showing the result we don't need connection anymore
conn.Close();
}
catch (Exception msg)
{
// something went wrong, and you wanna know why
// MessageBox.Show(msg.ToString());
}
}