public void InitOptions(ISynchronizedLinks <TLink> links)
{
if (UseSequenceMarker)
{
if (_equalityComparer.Equals(SequenceMarkerLink, links.Constants.Null))
{
SequenceMarkerLink = links.CreatePoint();
}
else
{
if (!links.Exists(SequenceMarkerLink))
{
var link = links.CreatePoint();
if (!_equalityComparer.Equals(link, SequenceMarkerLink))
{
throw new InvalidOperationException("Cannot recreate sequence marker link.");
}
}
}
if (MarkedSequenceMatcher == null)
{
MarkedSequenceMatcher = new MarkedSequenceCriterionMatcher <TLink>(links, SequenceMarkerLink);
}
}
var balancedVariantConverter = new BalancedVariantConverter <TLink>(links);
if (UseCompression)
{
if (LinksToSequenceConverter == null)
{
ICounter <TLink, TLink> totalSequenceSymbolFrequencyCounter;
if (UseSequenceMarker)
{
totalSequenceSymbolFrequencyCounter = new TotalMarkedSequenceSymbolFrequencyCounter <TLink>(links, MarkedSequenceMatcher);
}
else
{
totalSequenceSymbolFrequencyCounter = new TotalSequenceSymbolFrequencyCounter <TLink>(links);
}
var doubletFrequenciesCache = new LinkFrequenciesCache <TLink>(links, totalSequenceSymbolFrequencyCounter);
var compressingConverter = new CompressingConverter <TLink>(links, balancedVariantConverter, doubletFrequenciesCache);
LinksToSequenceConverter = compressingConverter;
}
}
else
{
if (LinksToSequenceConverter == null)
{
LinksToSequenceConverter = balancedVariantConverter;
}
}
if (UseIndex && Index == null)
{
Index = new SequenceIndex <TLink>(links);
}
if (Walker == null)
{
Walker = new RightSequenceWalker <TLink>(links, new DefaultStack <TLink>());
}
}
public static void CompressionTest()
{
using (var scope = new TempLinksTestScope(useSequences: true))
{
var links = scope.Links;
var sequences = scope.Sequences;
var e1 = links.Create();
var e2 = links.Create();
var sequence = new[]
{
e1, e2, e1, e2 // mama / papa / template [(m/p), a] { [1] [2] [1] [2] }
};
var balancedVariantConverter = new BalancedVariantConverter <ulong>(links.Unsync);
var totalSequenceSymbolFrequencyCounter = new TotalSequenceSymbolFrequencyCounter <ulong>(links.Unsync);
var doubletFrequenciesCache = new LinkFrequenciesCache <ulong>(links.Unsync, totalSequenceSymbolFrequencyCounter);
var compressingConverter = new CompressingConverter <ulong>(links.Unsync, balancedVariantConverter, doubletFrequenciesCache);
var compressedVariant = compressingConverter.Convert(sequence);
// 1: [1] (1->1) point
// 2: [2] (2->2) point
// 3: [1,2] (1->2) doublet
// 4: [1,2,1,2] (3->3) doublet
Assert.True(links.GetSource(links.GetSource(compressedVariant)) == sequence[0]);
Assert.True(links.GetTarget(links.GetSource(compressedVariant)) == sequence[1]);
Assert.True(links.GetSource(links.GetTarget(compressedVariant)) == sequence[2]);
Assert.True(links.GetTarget(links.GetTarget(compressedVariant)) == sequence[3]);
var source = _constants.SourcePart;
var target = _constants.TargetPart;
Assert.True(links.GetByKeys(compressedVariant, source, source) == sequence[0]);
Assert.True(links.GetByKeys(compressedVariant, source, target) == sequence[1]);
Assert.True(links.GetByKeys(compressedVariant, target, source) == sequence[2]);
Assert.True(links.GetByKeys(compressedVariant, target, target) == sequence[3]);
// 4 - length of sequence
Assert.True(links.GetSquareMatrixSequenceElementByIndex(compressedVariant, 4, 0) == sequence[0]);
Assert.True(links.GetSquareMatrixSequenceElementByIndex(compressedVariant, 4, 1) == sequence[1]);
Assert.True(links.GetSquareMatrixSequenceElementByIndex(compressedVariant, 4, 2) == sequence[2]);
Assert.True(links.GetSquareMatrixSequenceElementByIndex(compressedVariant, 4, 3) == sequence[3]);
}
}
public static void CompressionEfficiencyTest()
{
var strings = _exampleLoremIpsumText.Split(new[] { '\n', '\r' }, StringSplitOptions.RemoveEmptyEntries);
var arrays = strings.Select(UnicodeMap.FromStringToLinkArray).ToArray();
var totalCharacters = arrays.Select(x => x.Length).Sum();
using (var scope1 = new TempLinksTestScope(useSequences: true))
using (var scope2 = new TempLinksTestScope(useSequences: true))
using (var scope3 = new TempLinksTestScope(useSequences: true))
{
scope1.Links.Unsync.UseUnicode();
scope2.Links.Unsync.UseUnicode();
scope3.Links.Unsync.UseUnicode();
var balancedVariantConverter1 = new BalancedVariantConverter <ulong>(scope1.Links.Unsync);
var totalSequenceSymbolFrequencyCounter = new TotalSequenceSymbolFrequencyCounter <ulong>(scope1.Links.Unsync);
var linkFrequenciesCache1 = new LinkFrequenciesCache <ulong>(scope1.Links.Unsync, totalSequenceSymbolFrequencyCounter);
var compressor1 = new CompressingConverter <ulong>(scope1.Links.Unsync, balancedVariantConverter1, linkFrequenciesCache1, doInitialFrequenciesIncrement: false);
//var compressor2 = scope2.Sequences;
var compressor3 = scope3.Sequences;
var constants = Default <LinksConstants <ulong> > .Instance;
var sequences = compressor3;
//var meaningRoot = links.CreatePoint();
//var unaryOne = links.CreateAndUpdate(meaningRoot, constants.Itself);
//var frequencyMarker = links.CreateAndUpdate(meaningRoot, constants.Itself);
//var frequencyPropertyMarker = links.CreateAndUpdate(meaningRoot, constants.Itself);
//var unaryNumberToAddressConverter = new UnaryNumberToAddressAddOperationConverter<ulong>(links, unaryOne);
//var unaryNumberIncrementer = new UnaryNumberIncrementer<ulong>(links, unaryOne);
//var frequencyIncrementer = new FrequencyIncrementer<ulong>(links, frequencyMarker, unaryOne, unaryNumberIncrementer);
//var frequencyPropertyOperator = new FrequencyPropertyOperator<ulong>(links, frequencyPropertyMarker, frequencyMarker);
//var linkFrequencyIncrementer = new LinkFrequencyIncrementer<ulong>(links, frequencyPropertyOperator, frequencyIncrementer);
//var linkToItsFrequencyNumberConverter = new LinkToItsFrequencyNumberConveter<ulong>(links, frequencyPropertyOperator, unaryNumberToAddressConverter);
var linkFrequenciesCache3 = new LinkFrequenciesCache <ulong>(scope3.Links.Unsync, totalSequenceSymbolFrequencyCounter);
var linkToItsFrequencyNumberConverter = new FrequenciesCacheBasedLinkToItsFrequencyNumberConverter <ulong>(linkFrequenciesCache3);
var sequenceToItsLocalElementLevelsConverter = new SequenceToItsLocalElementLevelsConverter <ulong>(scope3.Links.Unsync, linkToItsFrequencyNumberConverter);
var optimalVariantConverter = new OptimalVariantConverter <ulong>(scope3.Links.Unsync, sequenceToItsLocalElementLevelsConverter);
var compressed1 = new ulong[arrays.Length];
var compressed2 = new ulong[arrays.Length];
var compressed3 = new ulong[arrays.Length];
var START = 0;
var END = arrays.Length;
//for (int i = START; i < END; i++)
// linkFrequenciesCache1.IncrementFrequencies(arrays[i]);
var initialCount1 = scope2.Links.Unsync.Count();
var sw1 = Stopwatch.StartNew();
for (int i = START; i < END; i++)
{
linkFrequenciesCache1.IncrementFrequencies(arrays[i]);
compressed1[i] = compressor1.Convert(arrays[i]);
}
var elapsed1 = sw1.Elapsed;
var balancedVariantConverter2 = new BalancedVariantConverter <ulong>(scope2.Links.Unsync);
var initialCount2 = scope2.Links.Unsync.Count();
var sw2 = Stopwatch.StartNew();
for (int i = START; i < END; i++)
{
compressed2[i] = balancedVariantConverter2.Convert(arrays[i]);
}
var elapsed2 = sw2.Elapsed;
for (int i = START; i < END; i++)
{
linkFrequenciesCache3.IncrementFrequencies(arrays[i]);
}
var initialCount3 = scope3.Links.Unsync.Count();
var sw3 = Stopwatch.StartNew();
for (int i = START; i < END; i++)
{
//linkFrequenciesCache3.IncrementFrequencies(arrays[i]);
compressed3[i] = optimalVariantConverter.Convert(arrays[i]);
}
var elapsed3 = sw3.Elapsed;
Console.WriteLine($"Compressor: {elapsed1}, Balanced variant: {elapsed2}, Optimal variant: {elapsed3}");
// Assert.True(elapsed1 > elapsed2);
// Checks
for (int i = START; i < END; i++)
{
var sequence1 = compressed1[i];
var sequence2 = compressed2[i];
var sequence3 = compressed3[i];
var decompress1 = UnicodeMap.FromSequenceLinkToString(sequence1, scope1.Links.Unsync);
var decompress2 = UnicodeMap.FromSequenceLinkToString(sequence2, scope2.Links.Unsync);
var decompress3 = UnicodeMap.FromSequenceLinkToString(sequence3, scope3.Links.Unsync);
var structure1 = scope1.Links.Unsync.FormatStructure(sequence1, link => link.IsPartialPoint());
var structure2 = scope2.Links.Unsync.FormatStructure(sequence2, link => link.IsPartialPoint());
var structure3 = scope3.Links.Unsync.FormatStructure(sequence3, link => link.IsPartialPoint());
//if (sequence1 != Constants.Null && sequence2 != Constants.Null && arrays[i].Length > 3)
// Assert.False(structure1 == structure2);
//if (sequence3 != Constants.Null && sequence2 != Constants.Null && arrays[i].Length > 3)
// Assert.False(structure3 == structure2);
Assert.True(strings[i] == decompress1 && decompress1 == decompress2);
Assert.True(strings[i] == decompress3 && decompress3 == decompress2);
}
Assert.True((int)(scope1.Links.Unsync.Count() - initialCount1) < totalCharacters);
Assert.True((int)(scope2.Links.Unsync.Count() - initialCount2) < totalCharacters);
Assert.True((int)(scope3.Links.Unsync.Count() - initialCount3) < totalCharacters);
Console.WriteLine($"{(double)(scope1.Links.Unsync.Count() - initialCount1) / totalCharacters} | {(double)(scope2.Links.Unsync.Count() - initialCount2) / totalCharacters} | {(double)(scope3.Links.Unsync.Count() - initialCount3) / totalCharacters}");
Assert.True(scope1.Links.Unsync.Count() - initialCount1 < scope2.Links.Unsync.Count() - initialCount2);
Assert.True(scope3.Links.Unsync.Count() - initialCount3 < scope2.Links.Unsync.Count() - initialCount2);
var duplicateProvider1 = new DuplicateSegmentsProvider <ulong>(scope1.Links.Unsync, scope1.Sequences);
var duplicateProvider2 = new DuplicateSegmentsProvider <ulong>(scope2.Links.Unsync, scope2.Sequences);
var duplicateProvider3 = new DuplicateSegmentsProvider <ulong>(scope3.Links.Unsync, scope3.Sequences);
var duplicateCounter1 = new DuplicateSegmentsCounter <ulong>(duplicateProvider1);
var duplicateCounter2 = new DuplicateSegmentsCounter <ulong>(duplicateProvider2);
var duplicateCounter3 = new DuplicateSegmentsCounter <ulong>(duplicateProvider3);
var duplicates1 = duplicateCounter1.Count();
ConsoleHelpers.Debug("------");
var duplicates2 = duplicateCounter2.Count();
ConsoleHelpers.Debug("------");
var duplicates3 = duplicateCounter3.Count();
Console.WriteLine($"{duplicates1} | {duplicates2} | {duplicates3}");
linkFrequenciesCache1.ValidateFrequencies();
linkFrequenciesCache3.ValidateFrequencies();
}
}