/// <summary>
/// Method used for compression of a suffix array based on Delta encoding algorithm.
/// This method is also using Psi function. See <see cref="PsiFunction.GeneratePsiFunction(int[], string)"/>.
/// </summary>
/// <param name="input">String which will be compressed.</param>
/// <returns>Dictionary of compressed values divided by characters from the original input.</returns>
public static Dictionary <char, List <Int16> > DeltaEncodingCompression(string input)
{
Dictionary <char, List <int> > incSequences = PsiFunction.GetIncreasingSequences(input);
Dictionary <char, List <Int16> > deltaEncoding = new Dictionary <char, List <Int16> >();
List <Int16> deltaSequence;
foreach (var key in incSequences.Keys)
{
deltaSequence = new List <Int16>();
deltaSequence.Add(Convert.ToInt16(incSequences[key].ElementAt(0)));
deltaEncoding.Add(key, deltaSequence);
}
foreach (var key in incSequences.Keys)
{
deltaSequence = new List <Int16>();
deltaSequence = deltaEncoding[key];
for (int i = 0; i < incSequences[key].Count - 1; i++)
{
deltaSequence.Add(Convert.ToInt16(incSequences[key].ElementAt(i + 1) - incSequences[key].ElementAt(i)));
}
deltaEncoding[key] = deltaSequence;
}
return(deltaEncoding);
}
/// <summary>
/// Method used for decompression of a compressed suffix array based on Delta encoding algorithm.
/// This method is also using method which generates suffix array from Psi function.
/// See <see cref="PsiFunction.PsiToSA(int[])"/>.
/// </summary>
/// <param name="delta">Dictionary of compressed values divided by characters from the original input.</param>
/// <param name="length">Length of the original input.</param>
/// <returns>Suffix array.</returns>
public static int[] DecompressDeltaEncoding(Dictionary <char, List <Int16> > delta, int length)
{
int[] psi = new int[length];
psi[0] = -1;
int j = 1;
foreach (var key in delta.Keys)
{
psi[j++] = Convert.ToInt32(delta[key][0]);
for (int i = 1; i < delta[key].Count; i++)
{
psi[j] = Convert.ToInt32(delta[key][i]) + psi[j - 1];
j++;
}
}
return(PsiFunction.PsiToSA(psi));
}
/// <summary>
/// Method used for decompression of list of byte arrays based on Elias-Fano algorithm.
/// </summary>
/// <param name="eliasFano">List of byte arrays. Result of Elias-Fano compression.</param>
/// <param name="bitLengthOfSeq">List of values that represent length of resulting bit sequences by increasing sequences.</param>
/// <param name="bitLengthOfOrigParts">List of integer arrays containing information about length of leading and lower bit sequences.</param>
/// <returns>Suffix array.</returns>
public static int[] DecompressEliasFano(List <byte[]> eliasFano, List <int> bitLengthOfSeq, List <int[]> bitLengthOfOrigParts)
{
int highLowBitsSeam = 0;
List <int> sequence;
List <int> psi = new List <int>();
psi.Insert(0, -1);
int k = 1;
for (int i = 0; i < eliasFano.Count; i++)
{
bool[] bitArray = ByteToBit(eliasFano[i], bitLengthOfSeq[i]);
highLowBitsSeam = GetHigherLowerBitsSeam(bitArray, bitLengthOfOrigParts[i][0]);
List <bool[]> leadingBits = GenerateLeadingBits(bitArray, highLowBitsSeam, bitLengthOfOrigParts[i][0]);
List <bool[]> lowerBits = GenerateLowerBits(bitArray, highLowBitsSeam, bitLengthOfOrigParts[i][1]);
if (bitLengthOfOrigParts[i][1] == 0)
{
sequence = new List <int>();
int index = 0;
foreach (var item in leadingBits)
{
sequence.Insert(index, GetNumberFromBits(item));
index++;
}
}
else
{
sequence = GenerateSequence(leadingBits, lowerBits);
}
foreach (var el in sequence)
{
psi.Insert(k, el);
k++;
}
}
return(PsiFunction.PsiToSA(psi.ToArray()));
}
/// <summary>
/// Method which is automatically called when starting this application.
/// Contains generating suffix array and calls of different methods of compression.
/// </summary>
/// <param name="args"></param>
public static void Main(string[] args)
{
Console.Write("Unesite apsolutnu putanju do datoteke: ");
string path = Console.ReadLine();
string input = "";
try
{
input = File.ReadAllText(path);
}
catch (Exception e)
{
Console.WriteLine("\nDatoteka koju želite čitati ne postoji ili ju nije moguće otvoriti.");
Console.ReadKey();
Environment.Exit(-1);
}
int[] suffixArray = new int[input.Length];
var watch = Stopwatch.StartNew();
SuffixArray.SAIS.sufsort(input, suffixArray, input.Length);
watch.Stop();
Console.Write("SA: ");
WriteArray(suffixArray);
Console.WriteLine("Izgradnja SA se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine("SA zauzima " + GetMemorySizeOfObject(suffixArray) + " bajtova.");
Console.WriteLine();
//compress SA with Re-Pair
watch.Restart();
int[] csa = RePair.RePairMethod(suffixArray).ToArray();
watch.Stop();
Console.Write("CSA: ");
WriteArray(csa);
Console.WriteLine("Izgradnja CSA preko Re-Pair se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine("CSA zauzima " + GetMemorySizeOfObject(csa) + " bajtova.");
Console.WriteLine();
//decompress Re-Pair to SA
watch.Restart();
RePair.DecompressRePair(csa.ToList(), input.Length);
watch.Stop();
Console.WriteLine("Dekompresija Re-Pair CSA se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine();
//compress Psi with Delta encoding
watch.Restart();
PsiFunction.GeneratePsiFunction(suffixArray, input);
Console.Write("Psi: ");
WriteArray(PsiFunction.psi);
Dictionary <char, List <Int16> > delta = DeltaEncoding.DeltaEncodingCompression(input);
Console.WriteLine("Delta:");
WriteDictionary(delta);
watch.Stop();
Console.WriteLine("Izgradnja CSA preko Delta encoding se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine("CSA zauzima " + GetMemorySizeOfObject(delta) + " bajtova.");
Console.WriteLine();
//decompress Delta encoding to SA
watch.Restart();
int[] SA = DeltaEncoding.DecompressDeltaEncoding(delta, input.Length);
watch.Stop();
Console.WriteLine("Dekompresija Delta encoding CSA se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine();
WriteArray(SA);
//compress Psi with Elias-Fano
watch.Restart();
PsiFunction.GeneratePsiFunction(suffixArray, input);
Console.Write("Psi: ");
WriteArray(PsiFunction.psi);
List <int> bitLengthOfSeq = new List <int>();
List <int[]> bitLengthOfOrigParts = new List <int[]>();
List <byte[]> eliasFano = EliasFano.EliasFanoCompression(input, bitLengthOfSeq, bitLengthOfOrigParts);
watch.Stop();
Console.WriteLine("Elias-Fano:");
WriteListOfArrays(eliasFano);
Console.WriteLine("Duljine bitova:");
WriteListOfArrays(bitLengthOfOrigParts);
Console.WriteLine("Izgradnja CSA preko Elias-Fano se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine("CSA zauzima " + GetMemorySizeOfObject(eliasFano) + " bajtova.");
Console.WriteLine();
//decompress Elias-Fano to SA
watch.Restart();
int[] SAElias = EliasFano.DecompressEliasFano(eliasFano, bitLengthOfSeq, bitLengthOfOrigParts);
watch.Stop();
Console.WriteLine("Dekompresija Elias-Fano CSA se izvodila " + watch.ElapsedMilliseconds + " ms.");
Console.WriteLine();
WriteArray(SAElias);
}
/// <summary>
/// Method used for compression of a suffix array based on Elias-Fano algorithm.
/// </summary>
/// <param name="input">Original input.</param>
/// <param name="bitLengthOfSeq">List of values that represent length of resulting bit sequences by increasing sequences.</param>
/// <param name="bitLengthOfOrigParts">List of integer arrays containing information about length of leading and lower bit sequences.</param>
/// <returns>List of bytes which represent compressed values.</returns>
public static List <byte[]> EliasFanoCompression(string input, List <int> bitLengthOfSeq, List <int[]> bitLengthOfOrigParts)
{
Dictionary <char, List <int> > incSequences = PsiFunction.GetIncreasingSequences(input);
List <bool[]> eliasFano = new List <bool[]>();
int eliasIndex = 0;
int m;
int noOfBits = 0;
int noOfLeadingBits;
int noOfLowerBits;
int[] bucket;
bool[] lowerBits;
bool[] leadingBits;
bool[] number;
bool[] lowerEliasFano;
bool[] leadingEliasFano;
foreach (var val in incSequences.Values)
{
int lowerPointer = 0;
m = val[val.Count - 1];
noOfBits = BitsToExpressNumber(val[val.Count - 1]);
noOfLeadingBits = GetNumberOfLeadingBits(val.Count);
bucket = new int[(int)Math.Pow(2, noOfLeadingBits)];
noOfLowerBits = noOfBits - noOfLeadingBits;
int[] bits = new int[2];
bits[0] = noOfLeadingBits;
bits[1] = noOfLowerBits;
bitLengthOfOrigParts.Insert(eliasIndex, bits);
lowerEliasFano = new bool[val.Count * noOfLowerBits];
foreach (var item in val)
{
number = new bool[noOfBits];
lowerBits = new bool[noOfLowerBits];
leadingBits = new bool[noOfLeadingBits];
number = NumberToBits(item, noOfBits);
int j = 0;
for (int i = 0; i < noOfBits; i++)
{
if (i < noOfLeadingBits)
{
leadingBits[i] = number[i];
}
else
{
lowerBits[j++] = number[i];
}
}
bucket[GetNumberFromBits(leadingBits)]++;
for (int i = 0; i < noOfLowerBits; i++)
{
lowerEliasFano[lowerPointer++] = lowerBits[i];
}
}
int sizeOfLeading = 0;
foreach (var counter in bucket)
{
sizeOfLeading += counter + 1;
}
leadingEliasFano = new bool[sizeOfLeading];
GenerateLeadingSequence(bucket, leadingEliasFano);
bool[] resultVector = GenerateResultVectorOfBits(sizeOfLeading + val.Count * noOfLowerBits,
leadingEliasFano, lowerEliasFano);
eliasFano.Insert(eliasIndex, resultVector);
eliasIndex++;
}
int f = 0;
foreach (var seq in eliasFano)
{
bitLengthOfSeq.Insert(f, seq.Length);
f++;
}
return(EncodeToBytes(eliasFano));
}