/// <summary>
/// Start main loop of processing the sources.
/// </summary>
private void StartProcessing()
{
while (true)
{
try
{
using (var db = new MMonitorContext())
{
var sources = db.TheSources
.Where(s =>
s.Enc == null &&
s.AutomaticalEncodingUpdateWasSuccess == null &&
s.IgnoreForAutomaticHelpersWork.Value != true)
.Take(numOfSourcesForParallerlProcessing);
if (sources.Count() == 0)
{
Thread.Sleep(1000 * 60 * 10);
continue;
}
log.Info($"{sources.Count()} sources were taken for processing");
Parallel.ForEach(sources, (s) =>
{
log.Info($"Starting processing of source {s.Url}");
Process(s);
log.Info($"Ended processing of source {s.Url}");
});
log.Info("Sources processing completed");
db.SaveChanges();
log.Info("Sources updated in db");
}
}
catch (Exception ex)
{
log.Error("Error with processing sources", ex);
Thread.Sleep(5000);
}
}
}
static void Main(string[] args)
{
md5 = MD5.Create();
string file = System.Configuration.ConfigurationManager.AppSettings["file_with_sources"];
int row = 0;
using (var db = new MMonitorContext())
{
var lines = File.ReadAllLines(file);
TheSource s;
Uri u;
string url = string.Empty;
string sourceType = string.Empty;
string l = string.Empty;
foreach (var line in lines)
{
string[] parts = line.Split('\t').Select(p => p.Trim()).ToArray();
if (parts[0].Contains("xn--"))
{
try
{
u = new Uri(parts[0]);
l = u.Host;
}
catch (Exception ex)
{
log.Error("Can't create source url from puny url", ex);
}
}
if (parts[0].Contains("https://"))
{
l = parts[0].Replace("https://", "");
}
else if (parts[0].Contains("http://"))
{
l = parts[0].Replace("http://", "");
}
else
{
l = parts[0];
}
if (l.Contains("www."))
{
l = l.Replace("www.", "");
}
if (!Uri.IsWellFormedUriString(l, UriKind.Relative) && !l.Contains(".рф"))
{
log.Error(string.Format("Bad format for Url {0} at line {1}", parts[0], row));
continue;
}
l = l.Trim().ToLower();
if (l.Contains(".рф") && !Regex.IsMatch(l, @"\S+\.рф$"))
{
log.Error(string.Format("Bad format for Url {0} at line {1}", parts[0], row));
continue;
}
s = new TheSource()
{
Url = l,
UrlHash = GetMD5Hash(l),
Lang = string.IsNullOrWhiteSpace(parts[1]) ? Langs.UNDEFINED : (Langs)Enum.Parse(typeof(Langs), parts[1]),
TheSourceType = (TheSourceType)Enum.Parse(typeof(TheSourceType), parts[2])
};
if (db.TheSources.Where(x => x.UrlHash == s.UrlHash).FirstOrDefault() == null)
{
db.TheSources.Add(s);
row++;
Console.WriteLine("row is " + row);
}
}
try
{
db.SaveChanges();
log.Info($"Loaded {row} sources to database");
}
catch (Exception ex)
{
log.Error("Can't upload sources", ex);
}
}
Console.ReadLine();
}
