public void BloomFilterAddTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var testData2 = DataGenerator.Generate().Skip(addSize).Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new BloomFilter <TestEntity, long>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData2)
{
bloomFilter2.Add(itm);
}
bloomFilter.Add(bloomFilter2);
var contained = testData.Union(testData2).Count(item => bloomFilter.ContainsKey(item.Id));
Assert.AreEqual(contained, 2 * addSize, "Not all items found in added Bloom filters");
}
public void BloomFilterAddDifferentSizesTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var testData2 = DataGenerator.Generate().Skip(addSize).Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(4 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new BloomFilter <TestEntity, long>(configuration);
//We have to create a foldable version.
var data = bloomFilter.Extract();
var foldFactor = configuration.FoldingStrategy.GetAllFoldFactors(data.BlockSize).Where(f => f > 1).OrderBy(f => f).First();
bloomFilter2.Initialize(addSize, data.BlockSize / foldFactor, data.HashFunctionCount);
foreach (var itm in testData2)
{
bloomFilter2.Add(itm);
}
bloomFilter.Add(bloomFilter2);
var contained = testData.Union(testData2).Count(item => bloomFilter.Contains(item));
Assert.AreEqual(2 * addSize, contained, "Not all items found in added Bloom filters");
}
public void BloomFilterRemoveItemTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var contained = testData.Count(item => bloomFilter.Contains(item));
foreach (var item in testData.Take(addSize / 2))
{
bloomFilter.Remove(item);
}
var containedAfterRemove = testData.Count(item => bloomFilter.Contains(item));
//Bloom filter does not behave well under removal
Assert.AreEqual(containedAfterRemove, 4137, "Wrong item count after removal.");
}
public void BloomFilterIntersectDifferentFiltersTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new BloomFilter <TestEntity, long>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData.Skip(1000))
{
bloomFilter2.Add(itm);
}
bloomFilter.Intersect(bloomFilter2);
Assert.IsTrue(bloomFilter.ItemCount >= 9000);
var count = testData.Skip(1000).Count(bloomFilter.Contains);
Assert.AreEqual(9000, count);
}
public void BloomFilterIntersectEqualFiltersTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var bloomFilter2 = new BloomFilter <TestEntity, long>(configuration);
bloomFilter2.Initialize(2 * size, errorRate);
foreach (var itm in testData)
{
bloomFilter2.Add(itm);
}
bloomFilter.Intersect(bloomFilter2);
//item count will be off due to estimated size.
Assert.IsTrue(bloomFilter.ItemCount >= addSize);
Assert.IsTrue(testData.All(bloomFilter.Contains));
}
public void StandardCompressTest()
{
var addSize = 10000;
var errorRate = 0.001F;
var data = DataGenerator.Generate().Take(addSize).ToArray();
var filter = new BloomFilter <TestEntity, long>(new DefaultBloomFilterConfiguration());
filter.Initialize(data.Length, errorRate);
foreach (var item in data)
{
filter.Add(item);
}
var basecount = DataGenerator
.Generate()
.Skip(addSize)
.Take(addSize)
.Count(itm => filter.ContainsKey(itm.Id));
Assert.IsTrue(basecount <= errorRate * addSize);
filter.Initialize(50 * data.Length, errorRate);
Assert.AreEqual(filter.Capacity, 500000, "Unexpected size of reverse Bloom filter.");
foreach (var item in data)
{
filter.Add(item);
}
//check error rate.
var notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => filter.Contains(itm));
Assert.IsTrue(notFoundCount <= basecount, "Uncompressed Bloom filter exceeded error rate.");
Assert.IsTrue(data.All(d => filter.ContainsKey(d.Id)), "False negatives found in uncompressed filter");
filter = filter.Compress(true);
Assert.AreEqual(filter.Capacity, 21739, "Unexpected size of compressed Bloom filter.");
var compressNotFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => filter.ContainsKey(itm.Id));
Assert.IsTrue(data.All(d => filter.ContainsKey(d.Id)), "False negatives found in compressed filter");
Assert.IsTrue(compressNotFoundCount <= basecount, "Compressed Bloom filter exceeded error rate.");
}
public void BloomFilterQuasiDecodeTest()
{
var size = 100000;
var data = DataGenerator.Generate().Take(size).ToList();
var errorRate = 0.001F;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(size, errorRate);
foreach (var itm in data)
{
bloomFilter.Add(itm);
}
data = DataGenerator.Generate().Skip(500).Take(8000).ToList();
data.Modify(1000);
var estimate = bloomFilter.QuasiDecode(data);
Assert.IsTrue(estimate > 90500 && estimate < 95000, "Unexpected estimate for difference.");
}
public void HybridFalsePositiveTest()
{
var addSize = 10000;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var errorRate = 0.001F;
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(size, errorRate);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var notFoundCount = testData.Count(itm => !bloomFilter.Contains(itm));
Assert.IsTrue(notFoundCount == 0, "False negative error rate violated");
notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => bloomFilter.Contains(itm));
Assert.IsTrue(notFoundCount <= errorRate * addSize, "False positive error rate violated");
}
public void BloomFilterSimpleFold()
{
var addSize = 50;
var testData = DataGenerator.Generate().Take(addSize).ToArray();
var size = testData.Length;
var configuration = new DefaultBloomFilterConfiguration();
var bloomFilter = new BloomFilter <TestEntity, long>(configuration);
bloomFilter.Initialize(size, 1024, (uint)3);
foreach (var itm in testData)
{
bloomFilter.Add(itm);
}
var positiveCount = DataGenerator.Generate().Take(500).Count(itm => bloomFilter.Contains(itm));
var folded = bloomFilter.Fold(4);
var positiveCountAfterFold = DataGenerator.Generate().Take(500).Count(itm => bloomFilter.Contains(itm));
Assert.AreEqual(positiveCount, positiveCountAfterFold, "False positive count different after fold");
Assert.AreEqual(256, folded.BlockSize, "Folded block size is unexpected.");
Assert.IsTrue(testData.All(bloomFilter.Contains), "False negative found");
}