Ejemplo n.º 1
0
        private static void Main()
        {
            var groundTruth = DuplicateRates.FromLines(File.ReadLines(GroundTruthFile));
            var documents   = DirectoryTools.ReadStringDictionary(InputFile)
                              //.MapValues(SourceCodeNormalizer.NormalizeContent)
                              //.MapValues(SourceCodeNormalizer.NormalizeSpaces)
                              //.MapValues(SourceCodeNormalizer.NormalizeWords)
                              .MapValues(SourceCodeNormalizer.NormalizeSpaces);

            var hashes     = HashTools.HashValues(documents);
            var duplicates = FindDuplicates(hashes, groundTruth);

            var result = documents.RemoveKeys(duplicates);

            DirectoryTools.SaveAsJson(result, OutputFile);
        }