public AnsCodingTable(FrequencyDictionary <int> observedFrequencies, int denominator = -1, byte[] encryptionKey = null)
{
if (denominator <= 0)
{
denominator = DEFAULT_DENOMINATOR;
}
Denominator = (int)observedFrequencies.TotalFrequency;
FrequencyDictionary = new Dictionary <int, Fraction>();
foreach (FrequencyStorage <int> feature in observedFrequencies.GetSortedFrequencies())
{
FrequencyDictionary[feature.ContainedObject] = new Fraction(feature.Frequency, Denominator);
}
List <Fraction> toSimplify = new List <Fraction>();
foreach (KeyValuePair <int, Fraction> pair in FrequencyDictionary)
{
toSimplify.Add(pair.Value);
}
toSimplify.Sort(Fraction.SORTER);
int sharedDenominator = denominator;
try
{
Fraction.ScaleFractionsToDenominator(toSimplify, sharedDenominator);
} catch (ArgumentOutOfRangeException e)
{
throw new ArgumentException("Denominator insufficient to construct table without losing frequency fidelity. Either reduce the block size, or increase the denominator", e);
}
Streamer = new CodingTableStreamer(FrequencyDictionary);
Denominator = toSimplify[0].Denominator;
GenerateTable(encryptionKey);
}
public void Test(string text, Dictionary <string, int> expectedResult)
{
var actualResult1 = FrequencyDictionary.GetFrequenciesAtText(text);
var actualResult2 = FrequencyDictionary.GetFrequenciesAtTextSecondVersion(text);
Assert.That(actualResult1, Is.EqualTo(expectedResult));
Assert.That(actualResult2, Is.EqualTo(expectedResult));
}
RankingDictionary <int> GetCandidateRanking(List <SudokuCell> block)
{
FrequencyDictionary <int> candidateFrequency = new FrequencyDictionary <int>();
foreach (SudokuCell cell in block)
{
candidateFrequency.AddRange(cell.Candidates.ToArray());
}
return(candidateFrequency.GetRankingDictionary());
}
public void AddTest(int result, params string[] values)
{
var frequencyDictionary = new FrequencyDictionary <string>();
foreach (var value in values)
{
frequencyDictionary.Add(value);
}
Assert.AreEqual(result, frequencyDictionary.GetValue(values.First()));
}
private void EncodeBlock(int blockContentLength, int targetDenominator)
{
Output.WriteLong(blockContentLength, 23);
FrequencyDictionary <int> frequencies = new FrequencyDictionary <int>();
for (int i = 0; i < blockContentLength; i++)
{
frequencies.AddData(Block[i]);
}
AnsCodingTable blockTable = new AnsCodingTable(frequencies, targetDenominator);
blockTable.WriteTable(Output);
WriteEncoding(blockContentLength, blockTable);
}
public void EncodeBlock(int[] blockIn, int count, int targetDenominator, byte[] encryptionKey = null)
{
Output.WriteLong(count, 23);
Block = blockIn;
FrequencyDictionary <int> frequencies = new FrequencyDictionary <int>();
int offsetCount = count;
for (int i = 0; i < offsetCount; i++)
{
frequencies.AddData(Block[i]);
}
AnsCodingTable blockTable = new AnsCodingTable(frequencies, targetDenominator, encryptionKey);
blockTable.WriteTable(Output);
WriteEncoding(count, blockTable);
}
/// <summary>
/// Кодирование исходного текста блочным способом c использованием метрики
/// </summary>
/// <param name="fileNameSource">Имя файла исходного текста</param>
/// <param name="sizeBlock">Размкр блока</param>
/// <returns>Закодированый текст блочным методом с использованием метрики, представленный в виде списка битов</returns>
public List <bool> EncodeBlockWithMetrics(string fileNameSource, int sizeBlock,
out List <double> distanсeEuclide, out List <double> distanceChebyshev, out List <double> distanceCityBlock)
{
FrequencyDictionary frequencyDictionary = new FrequencyDictionary(fileNameSource);
Dictionary <string, double> p = ComparisonVector.GetVectorZerosFromFrequencyDictionary(frequencyDictionary);
Dictionary <string, double> q = ComparisonVector.GetVectorZerosFromFrequencyDictionary(frequencyDictionary);
List <double> distanceE = new List <double>();
List <double> distanceC = new List <double>();
List <double> distanceCB = new List <double>();
//Открываем поток для считывания с файла
StreamReader fileRead = new StreamReader(@fileNameSource, Encoding.Default);
List <bool> encodedSource = new List <bool>();
//Считываем весь текст с файла
string text = fileRead.ReadToEnd();
//int indexText = 0;
int indexBegin = 0;
int indexEnd = 0;
//пока не конец текста
while (indexEnd < text.Length)
{
//indexBegin = 0;
//indexEnd = 0;
List <string> listWords = new List <string>();
//идём по тексту, пока не наберём слов количеством в размер блока и пока не конец текста
for (int i = 0; (i < sizeBlock) && (indexEnd < text.Length); i++)
{
//пока это буква увеличиваем индекс конца слова
while (indexEnd < text.Length && char.IsLetter(text[indexEnd]))
{
indexEnd++;
}
//если мы прошлись по слову
string word;
if (indexBegin != indexEnd)
{
word = text.Substring(indexBegin, indexEnd - indexBegin);
}
else //это знак препенания, пробел или знак табуляции
{
indexEnd++;
word = text.Substring(indexBegin, indexEnd - indexBegin);
}
indexBegin = indexEnd;
//добавляем слово в блок
listWords.Add(word);
}
//кодирум блок слов и добавляем получившиеся биты в список битов
encodedSource.AddRange(EncodeBlock(listWords));
//работа с растояниями
q = ComparisonVector.GetVectorFrequencyFromBlock(frequencyDictionary, listWords);
distanceE.Add(ComparisonVector.GetEuclideanDistance(p, q));
distanceC.Add(ComparisonVector.GetChebyshevDistance(p, q));
distanceCB.Add(ComparisonVector.GetCityBlockDistance(p, q));
p = q;
}
//for (int i = 0; i < distanceE.Count; i++)
//{
// Console.Write(distanceE[i].ToString() + " " + distanceC[i].ToString() +
// " " + distanceCB[i].ToString() + "\n");
//}
distanсeEuclide = distanceE;
distanceChebyshev = distanceC;
distanceCityBlock = distanceCB;
//возращаем закодированный блок слов в виде списка бит
return(encodedSource);
}
RankingDictionary<int> GetCandidateRanking(List<SudokuCell> block)
{
FrequencyDictionary<int> candidateFrequency = new FrequencyDictionary<int>();
foreach(SudokuCell cell in block)
{
candidateFrequency.AddRange(cell.Candidates.ToArray());
}
return candidateFrequency.GetRankingDictionary();
}
/// <summary> /// Returns distortion between necessary and calculated value. /// For example between theoretical and practical values. /// </summary> /// <param name="chain"> /// The chain. /// </param> /// <param name="alphabet"> /// The alphabet. /// </param> /// <returns> /// The <see cref="double"/>. /// </returns> public abstract double Distortion(ComplexChain chain, FrequencyDictionary alphabet);
/// <summary> /// Returns the state of criterion. True, if everything is done, false - otherwise. /// </summary> /// <param name="chain"> /// The chain. /// </param> /// <param name="alphabet"> /// Chains alphabet. /// </param> /// <returns> /// The state of criterion. /// </returns> public abstract bool State(ComplexChain chain, FrequencyDictionary alphabet);
/// <summary>
/// Updates data for computing a new value of the criterion.
/// </summary>
/// <param name="chain">
/// A new chain.
/// </param>
/// <param name="alphabet">
/// A new alphabet.
/// </param>
public void Renew(ComplexChain chain, FrequencyDictionary alphabet)
{
this.chain = chain;
this.alphabet = alphabet;
}
public void GetTest(int result, string value)
{
var frequencyDictionary = new FrequencyDictionary <string>();
Assert.AreEqual(result, frequencyDictionary.GetValue(value));
}
private void buttonRun_Click(object sender, EventArgs e)
{
//Сначала для классического алгритма
if (textBoxOutC.Text != "")
{
labelOutС.Enabled = true;
textBoxOutC.Enabled = true;
//получаем частотный словарь
FrequencyDictionary frequencyDictionary = new FrequencyDictionary(openFileDialogInput.FileName);
SizeDictionaryC = frequencyDictionary.CountWord;
textBoxSizeDictionaryС.Text = SizeDictionaryC.ToString() + " слов";
//построение дерева
var startTime = System.Diagnostics.Stopwatch.StartNew();
HuffmanTree huffmanTree = new HuffmanTree(frequencyDictionary);
startTime.Stop();
var resultTime = startTime.Elapsed.TotalSeconds;
//string elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:000}",
// resultTime.Hours,
// resultTime.Minutes,
// resultTime.Seconds,
// resultTime.Milliseconds);
textBoxTimeBuildTree.Text = resultTime.ToString();
//Кодирование
startTime = System.Diagnostics.Stopwatch.StartNew();
huffmanTree.Encode(@openFileDialogInput.FileName, @textBoxOutC.Text);
startTime.Stop();
resultTime = resultTime + startTime.Elapsed.TotalSeconds;
//elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:000}",
// resultTime.Hours,
// resultTime.Minutes,
// resultTime.Seconds,
// resultTime.Milliseconds);
textBoxSpeedEncodeС.Text = resultTime.ToString();
//сбрасываем дерево
huffmanTree = new HuffmanTree();
//Декодируем
startTime = System.Diagnostics.Stopwatch.StartNew();
huffmanTree.Decode(@textBoxOutC.Text, @textBoxDecodeFileC.Text);
startTime.Stop();
resultTime = startTime.Elapsed.TotalSeconds;
//elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:000}",
// resultTime.Hours,
// resultTime.Minutes,
// resultTime.Seconds,
// resultTime.Milliseconds);
textBoxSpeedDecodeС.Text = resultTime.ToString();
//Получаем другие статитстики
SizeFileEncodeC = new FileInfo(@textBoxOutC.Text).Length;
textBoxSizeCompressС.Text = SizeFileEncodeC.ToString() + " байт";
textBoxCoefficientCompressС.Text = ((SizeFileInput * 1.0) / SizeFileEncodeC).ToString();
groupBoxClassicHuffmanAlgo.Enabled = true;
}
//Потом для адаптивного
if (textBoxOutA.Text != "")
{
labelOutA.Enabled = true;
textBoxOutA.Enabled = true;
HuffmanAdaptiveTree huffmanAdaptiveTree = new HuffmanAdaptiveTree();
//Кодирование
var startTime = System.Diagnostics.Stopwatch.StartNew();
huffmanAdaptiveTree.Encode(@openFileDialogInput.FileName, @textBoxOutA.Text);
startTime.Stop();
var resultTime = startTime.Elapsed.TotalSeconds;
//var elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:000}",
// resultTime.Hours,
// resultTime.Minutes,
// resultTime.Seconds,
// resultTime.Milliseconds);
textBoxSpeedEncodeA.Text = resultTime.ToString();
//сбрасываем дерево
huffmanAdaptiveTree = new HuffmanAdaptiveTree();
huffmanAdaptiveTree.Reset();
//Декодируем
startTime = System.Diagnostics.Stopwatch.StartNew();
huffmanAdaptiveTree.Decode(@textBoxOutA.Text, @textBoxDecodeFileA.Text);
startTime.Stop();
resultTime = startTime.Elapsed.TotalSeconds;
//elapsedTime = String.Format("{0:00}:{1:00}:{2:00}.{3:000}",
// resultTime.Hours,
// resultTime.Minutes,
// resultTime.Seconds,
// resultTime.Milliseconds);
textBoxSpeedDecodeA.Text = resultTime.ToString();
//Получаем другие статитстики
SizeFileEncodeA = new FileInfo(@textBoxOutA.Text).Length;
textBoxSizeCompressA.Text = SizeFileEncodeA.ToString() + " байт";
textBoxCoefficientCompressA.Text = ((SizeFileInput * 1.0) / SizeFileEncodeA).ToString();
groupBoxAdaptiveHuffmanAlgo.Enabled = true;
}
}