static void Main(string[] args)
{
InitIO();
var n = int.Parse(Console.ReadLine());
Console.Error.WriteLine("Input files: {0}", n);
var graph = new Graph <string>();
var comparer = new CodeComparer(new CodeAnalyzerBuilder());
var signatures = new Signatures();
var files = new string[n];
for (var i = 0; i < n; i++)
{
files[i] = Console.ReadLine();
var language = Path.GetExtension(files[i]);
var code = File.ReadAllText(files[i]);
code = signatures.FormatCode(code);
var codeAnalyzer = new SuffixTreeCodeAnalyzer(language, code);
graph.AddVertex(files[i]);
comparer.Register(files[i], language, code);
if ((i + 1) % 100 == 0)
{
OnProgress("Parsed: ", i + 1, n);
}
}
var maximum = (long)(n) * (long)(n - 1) / 2;
var progress = (long)0;
OnProgress("Compared: ", progress, maximum);
for (var i = 0; i < n; i++)
{
for (var j = i + 1; j < n; j++)
{
if (++progress % 10000 == 0)
{
OnProgress(progress, maximum);
}
if (files[i].Contains('-') && files[j].Contains('-'))
{
var id1 = files[i].Substring(0, files[i].IndexOf('-'));
var id2 = files[j].Substring(0, files[j].IndexOf('-'));
if (id1 == id2)
{
continue;
}
}
decimal compare = comparer.Compare(files[i], files[j]);
if (compare > 0.33M)
{
Console.Error.WriteLine("{0} | {1} -> {2:0.00}%", files[i], files[j], compare * 100);
graph.AddEdge(files[i], files[j]);
}
}
}
OnProgress(progress, maximum);
var res = from c in graph.GetConnectedComponents()
where c.Count >= 2
select c;
Console.WriteLine(res.Count());
Console.Error.WriteLine(res.Count());
foreach (var g in res)
{
foreach (var file in g.Verticies)
{
Console.Write(file + " ");
Console.Error.Write(file + " ");
}
Console.WriteLine();
Console.Error.WriteLine();
}
Console.Out.Dispose();
}