private static void TestOnTestTexts(DiacriticsReconstructor dr, int rPath) { foreach (var text in allTexts) { DiacriticsTester.Test(reconstructorsPath[rPath] + text, dr); } }
private static void TestWiki(DiacriticsReconstructor dr) { for (int i = 7; i < 8; i++) { var path = wiki_path_AA + "FileDR/" + wiki_ + string.Format("{0:00}", i); //Console.WriteLine(path); DiacriticsTester.Test(path, dr); } DiacriticsTester.PrintOverallStats(); }
internal static void Test(string path, DiacriticsReconstructor dr, bool writeStatistics = true) { long bytes = GC.GetTotalMemory(true); Console.WriteLine($"Memory (bytes): {bytes}"); if (writeStatistics) { statisticsPath = $"{TextFile.FileName(path)}_STATISTICS{TextFile.FileExtension(path)}"; File.WriteAllText(statisticsPath, $"Memory (bytes): {bytes}\n"); } Console.WriteLine($"Reading {path}"); string originalText = File.OpenText(path).ReadToEnd(); Console.WriteLine("Removing diacritics..."); string textWithoutDiacritics = StringRoutines.MyDiacriticsRemover(originalText); File.WriteAllText($"{TextFile.FileName(path)}_WITHOUT-DIACRITICS{TextFile.FileExtension(path)}", textWithoutDiacritics); Console.WriteLine("Reconstructing..."); var sw = Stopwatch.StartNew(); string reconstructedText = dr.Reconstruct(textWithoutDiacritics); sw.Stop(); string ngramsStat = dr.GetStatistic(); Console.Write(ngramsStat); if (writeStatistics) { File.AppendAllText(statisticsPath, ngramsStat); } dr.EraseStatistic(); Console.WriteLine($"Elapsed (milliseconds): {sw.Elapsed.TotalMilliseconds}"); if (writeStatistics) { File.AppendAllText(statisticsPath, $"Elapsed (milliseconds): {sw.Elapsed.TotalMilliseconds}\n"); } Console.WriteLine("Done."); File.WriteAllText($"{TextFile.FileName(path)}_RENCOSTRUCTED{TextFile.FileExtension(path)}", reconstructedText); Console.WriteLine("Testing..."); FindMistakes(originalText, reconstructedText, path, writeStatistics); Console.WriteLine("Done.\n"); }