/// <summary>
/// Determine the similarity between this estimator and the provided estimator data,
/// </summary>
/// <param name="estimatorData">The estimator data to compare against.</param>
/// <returns></returns>
public double?Similarity(IBitMinwiseHashEstimatorData estimatorData)
{
if (estimatorData == null)
{
return(0.0D);
}
return(Extract()
.Similarity(estimatorData));
}
/// <summary>
/// Create a bit array for the given estimator,
/// </summary>
/// <param name="estimator"></param>
/// <returns></returns>
private static FastBitArray CreateBitArray(IBitMinwiseHashEstimatorData estimator)
{
if (estimator == null)
{
throw new ArgumentNullException(nameof(estimator));
}
var estimatorBitArray = estimator.Values == null
? new FastBitArray((int)estimator.GetBitSize())
: new FastBitArray(estimator.Values)
{
Length = (int)estimator.GetBitSize()
};
if (estimator.Values == null)
{
estimatorBitArray.SetAll(true);
}
return(estimatorBitArray);
}
/// <summary>
/// Determine the similarity between two estimators.
/// </summary>
/// <param name="estimator"></param>
/// <param name="otherEstimatorData"></param>
/// <returns>Similarity (percentage similar, zero is completely different, one is completely the same)</returns>
/// <remarks>Zero is no similarity, one is completely similar.</remarks>
internal static double?Similarity(
this IBitMinwiseHashEstimatorData estimator,
IBitMinwiseHashEstimatorData otherEstimatorData)
{
if (estimator == null ||
otherEstimatorData == null ||
estimator.BitSize != otherEstimatorData.BitSize ||
estimator.HashCount != otherEstimatorData.HashCount)
{
return(null);
}
if (estimator.Values == null && otherEstimatorData.Values == null)
{
return(1.0D);
}
return(ComputeSimilarityFromSignatures(
CreateBitArray(estimator),
CreateBitArray(otherEstimatorData),
estimator.HashCount,
estimator.BitSize));
}
internal static long GetBitSize(this IBitMinwiseHashEstimatorData data)
{
return(data == null ? 0L : data.GetBlockSize() *data.BitSize);
}
internal static long GetBlockSize(this IBitMinwiseHashEstimatorData data)
{
return(data == null ? 0L : data.HashCount *data.Capacity);
}