/// <summary>
/// This method performs the BurrowsWheelerTransform algorithm on the input SuffixArray.
/// </summary>
/// <returns></returns>
public BurrowsWheelerTransformBin(SuffixArrayBin suffixArray)
{
textLength = suffixArray.InputText.Length;
List<BitArray> bwtTemp =
(suffixArray.Suffixes.Select(
i => new { i, lenSuffix = i.Length }).Select(
@t => @t.i.Append(suffixArray.InputText.Substring(0, textLength - @t.lenSuffix))
)
).ToList();
bwtTemp.Sort(
(BitArray bA, BitArray bB) => {
int order = 0;
for (int i = 0; i < Math.Max(bA.Length, bB.Length); i++) {
if (i >= bA.Length) {
order = -1;
break;
}
if (i >= bB.Length) {
order = 1;
break;
}
if (!bA[i] && bB[i]) {
order = -1;
break;
}
if (bA[i] && !bB[i]) {
order = 1;
break;
}
}
return order;
}
);
BitArray F = new BitArray(bwtTemp.Count);
BitArray L = new BitArray(bwtTemp.Count);
for (int i = 0; i < bwtTemp.Count; i++) {
F[i] = bwtTemp[i][0];
L[i] = bwtTemp[i][bwtTemp[i].Length - 1];
}
waveletTree = new WaveletTreeBin(L);
WaveletTreeBin waveletTreeF = new WaveletTreeBin(F);
C = new Dictionary<bool, int>(2);
if (F.HasZeros())
C[false] = waveletTreeF.Select(false, 1);
else
C[false] = -1;
if (F.HasOnes())
C[true] = waveletTreeF.Select(true, 1);
else
C[true] = -1;
}
public static void RunBinaryTestOnce(BitArray text, BitArray pattern, int times = 10000)
{
Console.WriteLine("T: " + BitArrayToString(text));
Console.WriteLine("P: " + BitArrayToString(pattern));
Console.WriteLine();
SuffixArrayBin sa = new SuffixArrayBin(text);
BurrowsWheelerTransformBin bwt = new BurrowsWheelerTransformBin(sa);
int answer = 0;
// BWT
Stopwatch sw = Stopwatch.StartNew();
for (int i = 0; i < times; i++) {
answer = bwt.CountOccurrences(pattern);
}
double time = ((double)sw.ElapsedMilliseconds) / times;
Console.WriteLine("BWT:\n====");
Console.WriteLine(" Time: " + time + " ms");
Console.WriteLine(" Matches: " + answer.ToString());
Console.WriteLine();
// Brute Force
Stopwatch sw2 = Stopwatch.StartNew();
for (int i = 0; i < times; i++) {
answer = BruteForce.Run(text, pattern);
}
double time2 = ((double)sw2.ElapsedMilliseconds) / times;
Console.WriteLine("BF:\n====");
Console.WriteLine(" Time: " + time2 + " ms");
Console.WriteLine(" Matches: " + answer.ToString());
Console.WriteLine();
}
public static void RunBinaryTestFiles()
{
const string DirectoryString = @"..\..";
var gen = new Random();
var files = Directory.EnumerateFiles(DirectoryString + @"\GenFiles\", "*").Where(u => u.Contains(@"GenFiles\B_"));
StringBuilder output = new StringBuilder();
int fileNum = 0;
foreach (var file in files) {
var ob = new StatObj { FileName = file };
var commandLineInterpreter = new CommandInterpreter() { Encoding = Encoding.Default, Path = ob.FileName };
var fileReader = new BinaryFileReader(commandLineInterpreter);
var sa = new SuffixArrayBin(fileReader.Data);
var bwt = new BurrowsWheelerTransformBin(sa);
for (var i = 1; i < Math.Min(20, ob.FileSize / 2); i++) {
ob.PatternSize = i;
BitArray pattern = fileReader.Data.Substring(
(int)(
(ob.FileSize - ob.PatternSize) * gen.NextDouble()
), ob.PatternSize
);
// Settings and timers
//===============================================================
int times = 100;
Stopwatch sw;
//===============================================================
// BWT
//===============================================================
StatObj bwtStat = ob;
bwtStat.EndTimeTicks = 0;
bwtStat.Results = 0;
bwtStat.RuntimeTicks = 0;
bwtStat.StartTimeTicks = 0;
sw = Stopwatch.StartNew();
for (int t = 0; t < times; t++) {
bwtStat.Results = bwt.CountOccurrences(pattern);
}
sw.Stop();
bwtStat.RuntimeTicks = sw.ElapsedMilliseconds;
//===============================================================
// BF
//===============================================================
StatObj bfStat = ob;
bfStat.EndTimeTicks = 0;
bfStat.Results = 0;
bfStat.RuntimeTicks = 0;
bfStat.StartTimeTicks = 0;
sw = Stopwatch.StartNew();
for (int t = 0; t < times; t++) {
bfStat.Results = BruteForce.Run(fileReader.Data, pattern);
}
sw.Stop();
bfStat.RuntimeTicks = sw.ElapsedMilliseconds;
//===============================================================
// Print Results
//===============================================================
int m = ob.PatternSize;
int n = ob.FileSize;
output.AppendLine(String.Format("BWT,{0},{1},{2},{3},{4}", bwtStat.RuntimeTicks, m, n, bwtStat.Results, "Binary"));
output.AppendLine(String.Format("BruteForce,{0},{1},{2},{3},{4}", bfStat.RuntimeTicks, m, n, bfStat.Results, "Binary"));
//===============================================================
}
fileNum++;
Console.WriteLine(((((double)fileNum) / files.ToArray<string>().Length) * 100).ToString() + " %");
}
using (StreamWriter writer = new StreamWriter(DirectoryString + @"\GenFiles\BinaryResults.csv")) {
writer.Write(output.ToString());
}
}