private static void TakeSubsetFromFile(string inputFile, int size, bool filter = false)
{
AspectManager AM = new AspectManager();
ParserWrapper pw = _parser._pool.GetParserWrapper();
AM.WorkingAspect = AM.DeserializeAspect(inputFile, pw);
System.Console.WriteLine("Reading finished");
List <PointOfInterest> points = AM.WorkingAspect.Items;
if (filter)
{
int AmbigRemoved = 0;
int AmbigRemains = 0;
int GoodRemoved = 0;
int GoodRemains = 0;
for (int i = 0; i < points.Count; ++i)
{
if (points[i]?.Title?.StartsWith("+ ") ?? false)
{
if (points[i].Title.StartsWith("+ using"))
{
points.RemoveAt(i);
i -= 1;
GoodRemoved += 1;
}
else
{
GoodRemains += 1;
}
}
else
{
if (points[i]?.Title?.StartsWith("using") ?? true)
{
points.RemoveAt(i);
i -= 1;
AmbigRemoved += 1;
}
else
{
AmbigRemains += 1;
}
}
}
Console.WriteLine("Removed: " + GoodRemoved + " good, " + AmbigRemoved + " ambiguous results");
Console.WriteLine("Remains: " + GoodRemains + " good, " + AmbigRemains + " ambiguous results");
}
if (points.Count > size)
{
Shuffle(points);
points.RemoveRange(size, points.Count - size);
}
AM.SerializeAspect(inputFile, true);
System.Console.WriteLine("Writing finished");
_parser._pool.ReleaseParserWrapper(pw);
}
private static void ProcessInputFile(string BaseDir, string inputFile)
{
AspectManager AM = new AspectManager();
ParserWrapper pw = _parser._pool.GetParserWrapper();
AM.WorkingAspect = AM.DeserializeAspect(inputFile, pw);
_parser._pool.ReleaseParserWrapper(pw);
int TotalNodes = AM.WorkingAspect.Items.Count;
int NotFound = 0;
int Errors = 0;
int Found = 0;
int NotChanged = 0;
TreeManager TM = new TreeManager();
PointOfInterest NewAspect = new PointOfInterest();
Semaphore S = new Semaphore(1, 1);
int counter = 0;
Parallel.For(0, AM.WorkingAspect.Items.Count, (i) =>
//for (int i = 0; i < AM.WorkingAspect.Items.Count; ++i)
{
Interlocked.Increment(ref counter);
if (counter % 10 == 0)
{
S.WaitOne();
Console.Write("\rNodes left: " + (AM.WorkingAspect.Items.Count - counter) + " ");
S.Release();
}
try
{
PointOfInterest Pt = AM.WorkingAspect.Items[i];
PointOfInterest Root = TM.GetTree(BaseDir + Pt.FileName);
string text = TM.GetText(BaseDir + Pt.FileName);
TreeSearchResult Search = TreeSearchEngine.FindPointInTree2(Root, Pt, text);
if (Search.Count == 0)
{
Interlocked.Increment(ref NotFound);
}
else if (Search.Singular)
{
if (Search.GetNodeSimilarity(0) == 1)
{
Interlocked.Increment(ref NotChanged);
return;
//continue;
}
else
{
Interlocked.Increment(ref Found);
AM.WorkingAspect.Items[i].Title = "+ " + AM.WorkingAspect.Items[i].Title;
}
}
else
{
if (Search.Count >= 2)
{
float d1 = 1 - Search.GetNodeSimilarity(0);
float d2 = 1 - Search.GetNodeSimilarity(1);
//float near = Math.Max(Pt.NearG, Pt.NearL);
//float threshold = ((near + 4) / 5 + 1) /2 ; //hardcoded
//if (Search.GetNodeSimilarity(0) >= threshold && Search.GetNodeSimilarity(1) < threshold)
if (d2 != 0 && d2 >= d1 * 2)
{
Interlocked.Increment(ref Found);
AM.WorkingAspect.Items[i].Title = "+ " + AM.WorkingAspect.Items[i].Title;
//return;
//continue;
}
}
}
S.WaitOne();
NewAspect.Items.Add(AM.WorkingAspect.Items[i]);
S.Release();
}
catch (Exception e)
{
S.WaitOne();
Console.WriteLine(e.Message);
Errors += 1;
S.Release();
}
}
);
AM.WorkingAspect = NewAspect;
int Ambiguous = AM.WorkingAspect.Items.Count - Found;
AM.SerializeAspect(inputFile, true);
File.WriteAllText(inputFile + ".report.txt", "Total: " + TotalNodes + ", Not changed: " + NotChanged + ", found: " + Found + ", Not found: " + NotFound + ", errors: " + Errors + ", ambiguous: " + Ambiguous);
Console.Write("\rNodes left: 0 ");
Console.WriteLine();
}
