Пример #1
0
        public void HybridRemoveItemTest()
        {
            var addSize       = 10000;
            var testData      = DataGenerator.Generate().Take(addSize).ToArray();
            var errorRate     = 0.001F;
            var size          = testData.Length;
            var configuration = new HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(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));

            //tricky: assuming zero false positives.
            Assert.AreEqual(contained, containedAfterRemove * 2, "Wrong item count after removal.");
        }
Пример #2
0
        public void HybridIntersectDifferentFiltersTest()
        {
            var addSize       = 10000;
            var testData      = DataGenerator.Generate().Take(addSize).ToArray();
            var errorRate     = 0.001F;
            var size          = testData.Length;
            var configuration = new HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);

            bloomFilter.Initialize(2 * size, errorRate);
            foreach (var itm in testData)
            {
                bloomFilter.Add(itm);
            }
            var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);

            bloomFilter2.Initialize(2 * size, errorRate);
            foreach (var itm in testData.Skip(1000))
            {
                bloomFilter2.Add(itm);
            }
            bloomFilter.Intersect(bloomFilter2);
            Assert.AreEqual(9000, bloomFilter.ItemCount);
            var count = testData.Skip(1000).Count(bloomFilter.Contains);

            //Note: intersect introduces a horrible error rate when utilizing XOR, so don't actually use intersect.
            //There are however definitions of operations possible where the intersect would not have this horrible effect.
            Assert.IsTrue(count > 6000);
            Assert.IsTrue(testData.Take(1000).All(d => !bloomFilter.Contains(d)));
        }
Пример #3
0
        public void HybridAddDifferentSizesTest()
        {
            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 HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);

            bloomFilter.Initialize(4 * size, errorRate);
            foreach (var itm in testData)
            {
                bloomFilter.Add(itm);
            }
            var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);
            //We should create a foldable counter part.
            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(contained, 2 * addSize, "Not all items found in added Bloom filters");
        }
Пример #4
0
        public void HybridAddTest()
        {
            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 HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);

            bloomFilter.Initialize(2 * size, errorRate);
            foreach (var itm in testData)
            {
                bloomFilter.Add(itm);
            }
            var bloomFilter2 = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(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.Contains(item));

            Assert.AreEqual(contained, 2 * addSize, "Not all items found in added Bloom filters");
        }
Пример #5
0
        public void HybridRemoveKeyTest()
        {
            var addSize       = 10000;
            var testData      = DataGenerator.Generate().Take(addSize).ToArray();
            var errorRate     = 0.001F;
            var size          = testData.Length;
            var configuration = new HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(configuration);

            bloomFilter.Initialize(2 * size, errorRate);
            foreach (var itm in testData)
            {
                bloomFilter.Add(itm);
            }
            var contained = testData.Count(item => bloomFilter.Contains(item));

            try
            {
                foreach (var item in testData.Take(addSize / 2))
                {
                    bloomFilter.RemoveKey(item.Id);
                }
                Assert.Fail("RemoveKey should not be supported by a hybrid invertible Bloom filter");
            }
            catch (NotSupportedException)
            { };
        }
Пример #6
0
        public void HybridSimpleFold()
        {
            var addSize       = 50;
            var testData      = DataGenerator.Generate().Take(addSize).ToArray();
            var size          = testData.Length;
            var configuration = new HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(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(item => bloomFilter.Contains(item)), "False negative found");
        }
Пример #7
0
        public void HybridFalsePositiveTest()
        {
            var addSize       = 10000;
            var testData      = DataGenerator.Generate().Take(addSize).ToArray();
            var errorRate     = 0.001F;
            var size          = testData.Length;
            var configuration = new HybridDefaultBloomFilterConfiguration();
            var bloomFilter   = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(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 = testData.Count(itm => !bloomFilter.ContainsKey(itm.Id));
            Assert.IsTrue(notFoundCount == 0, "False negative error rate violated on ContainsKey");
            notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => bloomFilter.Contains(itm));
            Assert.IsTrue(notFoundCount <= errorRate * addSize, "False positive error rate violated");
            notFoundCount = DataGenerator.Generate().Skip(addSize).Take(addSize).Count(itm => bloomFilter.ContainsKey(itm.Id));
            Assert.IsTrue(notFoundCount <= errorRate * addSize, "False positive error rate violated on ContainsKey");
        }
Пример #8
0
        public void HybridCompressTest()
        {
            var addSize      = 10000;
            var errorRate    = 0.001F;
            var data         = DataGenerator.Generate().Take(addSize).ToArray();
            var hybridFilter = new InvertibleHybridBloomFilter <TestEntity, long, sbyte>(new HybridDefaultBloomFilterConfiguration());

            hybridFilter.Initialize(50 * data.Length, errorRate);
            Assert.AreEqual(hybridFilter.Capacity, 500000, "Unexpected size of hybrid Bloom filter.");
            foreach (var item in data)
            {
                hybridFilter.Add(item);
            }
            //check error rate.
            var notFoundCount = DataGenerator.Generate().Skip(addSize).Take(10000).Count(itm => hybridFilter.Contains(itm));

            Assert.IsTrue(notFoundCount <= errorRate * addSize, "Uncompressed hybrid Bloom filter exceeded error rate.");
            hybridFilter.Compress(true);
            Assert.AreEqual(hybridFilter.Capacity, 12820, "Unexpected size of compressed hybrid Bloom filter.");
            var compressNotFoundCount = DataGenerator.Generate().Skip(addSize).Take(10000).Count(itm => hybridFilter.Contains(itm));

            Assert.IsTrue(compressNotFoundCount <= errorRate * addSize, "Compressed hybrid Bloom filter exceeded error rate.");
        }