/// <summary>
/// Fold the strata estimator data.
/// </summary>
/// <typeparam name="TEntity">The entity type</typeparam>
/// <typeparam name="TId">The identifier type</typeparam>
/// <typeparam name="TCount">The count type</typeparam>
/// <param name="estimatorData"></param>
/// <param name="configuration"></param>
/// <param name="factor">The factor to fold by</param>
/// <returns>The <paramref name="estimatorData"/> folded by <paramref name="factor"/>.</returns>
internal static HybridEstimatorFullData <int, TCount> Fold <TEntity, TId, TCount>(
this IHybridEstimatorFullData <int, TCount> estimatorData,
IInvertibleBloomFilterConfiguration <TEntity, TId, int, TCount> configuration,
uint factor)
where TCount : struct
where TId : struct
{
if (estimatorData == null)
{
return(null);
}
var minWiseFold = Math.Max(
1L,
configuration
.FoldingStrategy?
.GetAllFoldFactors(estimatorData.BitMinwiseEstimator?.Capacity ?? 1L)
.OrderBy(f => f)
.FirstOrDefault(f => f > factor) ?? 1L);
return(new HybridEstimatorFullData <int, TCount>
{
ItemCount = estimatorData.ItemCount,
BitMinwiseEstimator = estimatorData.BitMinwiseEstimator?.Fold((uint)minWiseFold),
StrataEstimator =
estimatorData.StrataEstimator?.Fold(configuration.ConvertToEstimatorConfiguration(), factor)
});
}
/// <summary>
/// Intersect with the estimator.
/// </summary>
/// <param name="estimator"></param>
public void Intersect(IHybridEstimatorFullData <int, TCount> estimator)
{
IHybridEstimator <TEntity, int, TCount> self = this;
Rehydrate(self
.FullExtract()
.Intersect(_configuration, estimator));
}
/// <summary>
/// Rehydrate the hybrid estimator from full data.
/// </summary>
/// <param name="data">The data to restore</param>
public void Rehydrate(IHybridEstimatorFullData <int, TCount> data)
{
if (data == null)
{
return;
}
_minwiseEstimator?.Rehydrate(data.BitMinwiseEstimator);
_strataEstimator.Rehydrate(data.StrataEstimator);
_minwiseReplacementCount = Math.Max(0, data.ItemCount - (_strataEstimator.ItemCount + (_minwiseEstimator?.ItemCount ?? 0L)));
}
/// <summary>
/// Intersect two hybrid estimators
/// </summary>
/// <typeparam name="TEntity"></typeparam>
/// <typeparam name="TId"></typeparam>
/// <typeparam name="TCount"></typeparam>
/// <param name="estimatorData"></param>
/// <param name="configuration"></param>
/// <returns></returns>
internal static IHybridEstimatorFullData <int, TCount> Intersect <TEntity, TId, TCount>(
this IHybridEstimatorFullData <int, TCount> estimatorData,
IInvertibleBloomFilterConfiguration <TEntity, TId, int, TCount> configuration,
IHybridEstimatorFullData <int, TCount> otherEstimatorData)
where TId : struct
where TCount : struct
{
if (estimatorData == null && otherEstimatorData == null)
{
return(null);
}
var res = new HybridEstimatorFullData <int, TCount>();
res.BitMinwiseEstimator = estimatorData?.BitMinwiseEstimator.Intersect(otherEstimatorData?.BitMinwiseEstimator, configuration.FoldingStrategy);
res.StrataEstimator = estimatorData?.StrataEstimator.Intersect(otherEstimatorData?.StrataEstimator, configuration);
res.ItemCount = (res.BitMinwiseEstimator?.ItemCount ?? 0L) + (res.StrataEstimator?.ItemCount ?? 0L);
return(res);
}
/// <summary>
/// Compress the hybrid estimator.
/// </summary>
/// <typeparam name="TEntity"></typeparam>
/// <typeparam name="TId"></typeparam>
/// <typeparam name="TCount"></typeparam>
/// <param name="estimatorData"></param>
/// <param name="configuration"></param>
/// <returns></returns>
internal static HybridEstimatorFullData <int, TCount> Compress <TEntity, TId, TCount>(
this IHybridEstimatorFullData <int, TCount> estimatorData,
IInvertibleBloomFilterConfiguration <TEntity, TId, int, TCount> configuration)
where TCount : struct
where TId : struct
{
if (configuration?.FoldingStrategy == null || estimatorData == null)
{
return(null);
}
var fold = configuration.FoldingStrategy.FindCompressionFactor(
configuration,
estimatorData.StrataEstimator.BlockSize,
estimatorData.StrataEstimator.BlockSize,
estimatorData.ItemCount);
var res = fold.HasValue ? estimatorData.Fold(configuration, fold.Value) : null;
return(res);
}