public void TsvSplit()

in XForm/XForm/Accessory/PerformanceComparisons.cs [357:447]
77 lines of code
8 McCabe index (conditional complexity)

        public void TsvSplit()
        {
            Stream tsvStream = new MemoryStream();
            //Stream tsvStream = new FileStream("Sample.tsv", FileMode.Create);
            int rowCount = 1000 * 1000;
            WriteSampleTsv(tsvStream, 5, 1000 * 1000);

            byte[] content = new byte[64 * 1024];
            BitVector cells = new BitVector(content.Length);
            BitVector rows = new BitVector(content.Length);
            int[] rowEnds = new int[1024];

            byte[] allContent = new byte[tsvStream.Length];
            tsvStream.Seek(0, SeekOrigin.Begin);
            tsvStream.Read(allContent, 0, allContent.Length);
            BitVector allCells = new BitVector(allContent.Length);
            BitVector allRows = new BitVector(allContent.Length);

            using (Benchmarker b = new Benchmarker($"Tsv Parse [{rowCount:n0}] | count", DefaultMeasureMilliseconds))
            {
                b.Measure("Read only", (int)tsvStream.Length, () =>
                {
                    tsvStream.Seek(0, SeekOrigin.Begin);
                    while (true)
                    {
                        int lengthRead = tsvStream.Read(content, 0, content.Length);
                        if (lengthRead == 0) break;
                    }

                    return rowCount;
                });

                b.Measure("ReadLine | Split", (int)tsvStream.Length, () =>
                {
                    tsvStream.Seek(0, SeekOrigin.Begin);
                    int count = 0;
                    StreamReader reader = new StreamReader(tsvStream);
                    {
                        // Header row
                        reader.ReadLine();

                        while (!reader.EndOfStream)
                        {
                            string line = reader.ReadLine();
                            string[] cellSet = line.Split('\t');
                            count++;
                        }
                    }
                    return count;
                });

                b.Measure("Elfie TsvReader", (int)tsvStream.Length, () =>
                {
                    tsvStream.Seek(0, SeekOrigin.Begin);
                    int count = 0;
                    ITabularReader reader = TabularFactory.BuildReader(tsvStream, "Unused.tsv");
                    {
                        while (reader.NextRow()) count++;
                    }
                    return count;
                });


                Func<byte[], int, int, ulong[], ulong[], int> splitTsvN = NativeAccelerator.GetMethod<Func<byte[], int, int, ulong[], ulong[], int>>("XForm.Native.String8N", "SplitTsv");
                b.Measure("XForm Native Split", (int)tsvStream.Length, () =>
                {
                    tsvStream.Seek(0, SeekOrigin.Begin);

                    int count = -1;
                    while (true)
                    {
                        int lengthRead = tsvStream.Read(content, 0, content.Length);
                        if (lengthRead == 0) break;
                        if (lengthRead < content.Length) Array.Clear(content, lengthRead, content.Length - lengthRead);

                        int lineCount = splitTsvN(content, 0, lengthRead, cells.Array, rows.Array);
                        count += lineCount;

                        int fromRow = 0;
                        int countCopy = cells.Page(rowEnds, ref fromRow);
                    }

                    return count;
                });

                b.MeasureParallel("XForm Native Split Parallel", (int)tsvStream.Length, (index, length) =>
                {
                    return splitTsvN(allContent, index, length, allCells.Array, allRows.Array) - 1;
                });
            }
        }