public void DirectCountingIsResetWhenMergingAlmostFullEstimators()
{
var addedEstimator = new CardinalityEstimator();
var mergedEstimator = new CardinalityEstimator();
for (int i = 0; i < 10_000; i++)
{
var guid = Guid.NewGuid().ToString();
addedEstimator.Add(guid);
// Simulate some intermediate estimators being merged together
var temporaryEstimator = new CardinalityEstimator();
temporaryEstimator.Add(guid);
mergedEstimator.Merge(temporaryEstimator);
}
var serializer = new CardinalityEstimatorSerializer();
var stream1 = new MemoryStream();
serializer.Serialize(stream1, addedEstimator, true);
var stream2 = new MemoryStream();
serializer.Serialize(stream2, mergedEstimator, true);
Assert.Equal(stream1.Length, stream2.Length);
}
public void TestSerializerCardinality10()
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(10);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
// Expected length is 93:
// 4 bytes for the major and minor versions
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in DirectCount
// 8 bytes for each element (ulong) in DirectCount
Assert.AreEqual(93, results.Length);
Assert.AreEqual(14, BitConverter.ToInt32(results.Skip(4).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.AreEqual(3, results[8]); // IsSparse = true AND IsDirectCount = true
Assert.AreEqual(10, BitConverter.ToInt32(results.Skip(9).Take(4).ToArray(), 0)); // Count = 10
}
public void TestSerializerCardinality1000()
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(1000);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll, false);
results = memoryStream.ToArray();
}
CardinalityEstimatorState data = hll.GetState();
// Expected length is:
// 4 bytes for the major and minor versions
// 1 byte for the HashFunctionId
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in lookupSparse
// 2+1 bytes for each element (ulong) in lookupSparse
// 8 bytes for CountAdded
Assert.AreEqual(22 + 3 * data.LookupSparse.Count, results.Length);
Assert.AreEqual((byte)HashFunctionId.Murmur3, results.Skip(4).Take(1).First());
Assert.AreEqual(14, BitConverter.ToInt32(results.Skip(5).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.AreEqual(2, results[9]); // IsSparse = true AND IsDirectCount = false
Assert.AreEqual(data.LookupSparse.Count, BitConverter.ToInt32(results.Skip(10).Take(4).ToArray(), 0));
Assert.AreEqual(1000UL, BitConverter.ToUInt64(results.Skip(14 + 3 * data.LookupSparse.Count).Take(8).ToArray(), 0)); // CountAdditions = 1000
}
private void TestSerializerCreatesSmallerData(int cardinality, out int customSize, out int defaultSize)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(cardinality);
var customSerializer = new CardinalityEstimatorSerializer();
byte[] customSerializerResults;
using (var memoryStream = new MemoryStream())
{
customSerializer.Serialize(memoryStream, hll, false);
customSerializerResults = memoryStream.ToArray();
customSize = customSerializerResults.Length;
}
var binaryFormatter = new BinaryFormatter();
byte[] defaultSerializerResults;
using (var memoryStream = new MemoryStream())
{
binaryFormatter.Serialize(memoryStream, hll);
defaultSerializerResults = memoryStream.ToArray();
defaultSize = defaultSerializerResults.Length;
}
Assert.IsTrue(customSerializerResults.Length <= defaultSerializerResults.Length);
}
public void TestSerializerCardinality1000()
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(1000);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
CardinalityEstimatorState data = hll.GetState();
// Expected length is 2908:
// 4 bytes for the major and minor versions
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in lookupSparse
// 2+1 bytes for each element (ulong) in lookupSparse
Assert.AreEqual(13 + 3 * data.LookupSparse.Count, results.Length);
Assert.AreEqual(14, BitConverter.ToInt32(results.Skip(4).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.AreEqual(2, results[8]); // IsSparse = true AND IsDirectCount = false
Assert.AreEqual(data.LookupSparse.Count, BitConverter.ToInt32(results.Skip(9).Take(4).ToArray(), 0));
}
public void TestSerializerCardinality10()
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(10);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
// Expected length is 93:
// 4 bytes for the major and minor versions
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in DirectCount
// 8 bytes for each element (ulong) in DirectCount
Assert.AreEqual(93, results.Length);
Assert.AreEqual(14, BitConverter.ToInt32(results.Skip(4).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.AreEqual(3, results[8]); // IsSparse = true AND IsDirectCount = true
Assert.AreEqual(10, BitConverter.ToInt32(results.Skip(9).Take(4).ToArray(), 0)); // Count = 10
}
public static byte[] SerializeCardinality(this CardinalityEstimator estimator, CardinalityEstimatorSerializer serializer)
{
using (var stream = new MemoryStream())
{
serializer.Serialize(stream, estimator);
return(stream.ToArray());
}
}
public void DeserializedEstimatorUsesSameHashAsOriginal()
{
// Prepare some elements
IList <int> elements = new List <int>();
for (int i = 0; i < 150; i++)
{
elements.Add(Rand.Next());
}
foreach (HashFunctionId hashFunctionId in Enum.GetValues(typeof(HashFunctionId)))
{
// Add elements to an estimator using the given hashFunctionId
CardinalityEstimator original = new CardinalityEstimator(hashFunctionId: hashFunctionId);
foreach (int element in elements)
{
original.Add(element);
}
// Serialize
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, original, false);
results = memoryStream.ToArray();
}
// Deserialize
CardinalityEstimator deserialized;
using (var memoryStream = new MemoryStream(results))
{
deserialized = serializer.Deserialize(memoryStream, false);
}
// Add the elements again, should have no effect on state
foreach (int element in elements)
{
deserialized.Add(element);
}
Assert.AreEqual(original.Count(), deserialized.Count());
}
}
private void TestDeserializer(int cardinality)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(cardinality);
CardinalityEstimator hll2;
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
using (var memoryStream = new MemoryStream(results))
{
hll2 = serializer.Deserialize(memoryStream);
}
CardinalityEstimatorState data = hll.GetState();
CardinalityEstimatorState data2 = hll2.GetState();
Assert.AreEqual(data.BitsPerIndex, data2.BitsPerIndex);
Assert.AreEqual(data.IsSparse, data2.IsSparse);
Assert.IsTrue((data.DirectCount != null && data2.DirectCount != null) || (data.DirectCount == null && data2.DirectCount == null));
Assert.IsTrue((data.LookupSparse != null && data2.LookupSparse != null) ||
(data.LookupSparse == null && data2.LookupSparse == null));
Assert.IsTrue((data.LookupDense != null && data2.LookupDense != null) || (data.LookupDense == null && data2.LookupDense == null));
if (data.DirectCount != null)
{
// DirectCount are subsets of each-other => they are the same set
Assert.IsTrue(data.DirectCount.IsSubsetOf(data2.DirectCount) && data2.DirectCount.IsSubsetOf(data.DirectCount));
}
if (data.LookupSparse != null)
{
Assert.IsTrue(data.LookupSparse.DictionaryEqual(data2.LookupSparse));
}
if (data.LookupDense != null)
{
Assert.IsTrue(data.LookupDense.SequenceEqual(data2.LookupDense));
}
}
public void TestSerializerMultipleCardinalityAndBitsCombinations()
{
for (int bits = 4; bits <= 16; bits++)
{
for (int cardinality = 1; cardinality <= 1000; cardinality++)
{
var estimator = CreateAndFillCardinalityEstimator(cardinality, bits);
CardinalityEstimatorSerializer serializer = new CardinalityEstimatorSerializer();
using (var stream = new MemoryStream())
{
serializer.Serialize(stream, estimator, true);
stream.Seek(0, SeekOrigin.Begin);
var deserializedEstimator = serializer.Deserialize(stream);
Assert.AreEqual(estimator.Count(), deserializedEstimator.Count(), "Estimators should have same count before and after serialization");
}
}
}
}
private void TestSerializerCardinality100000Parameterized(bool useBinWriter)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(100000);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
if (useBinWriter)
{
using (var bw = new BinaryWriter(memoryStream))
{
serializer.Write(bw, hll);
}
}
else
{
serializer.Serialize(memoryStream, hll, false);
}
results = memoryStream.ToArray();
}
CardinalityEstimatorState data = hll.GetState();
// Expected length is:
// 4 bytes for the major and minor versions
// 1 byte for the HashFunctionId
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in lookupDense
// 1 bytes for each element (ulong) in lookupDense
// 8 bytes for CountAdded
Assert.Equal(22 + data.LookupDense.Length, results.Length);
Assert.Equal((byte)HashFunctionId.Murmur3, results.Skip(4).Take(1).First());
Assert.Equal(14, BitConverter.ToInt32(results.Skip(5).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.Equal(0, results[9]); // IsSparse = false AND IsDirectCount = false
Assert.Equal(data.LookupDense.Length, BitConverter.ToInt32(results.Skip(10).Take(4).ToArray(), 0));
Assert.Equal(100000UL, BitConverter.ToUInt64(results.Skip(14 + data.LookupDense.Length).Take(8).ToArray(), 0)); // CountAdditions = 100000
}
private void TestDeserializer(int cardinality)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(cardinality);
CardinalityEstimator hll2;
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll, false);
results = memoryStream.ToArray();
}
using (var memoryStream = new MemoryStream(results))
{
hll2 = serializer.Deserialize(memoryStream, false);
}
CompareHLL(hll, hll2);
}
public void SerializeToStream(Stream stream)
{
Serializer.Serialize(stream, _hyperLogLog);
}
public void TestSerializerCardinality1000()
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(1000);
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
CardinalityEstimatorState data = hll.GetState();
// Expected length is 2908:
// 4 bytes for the major and minor versions
// 4 bytes for the Bits in Index
// 1 byte for the IsSparse and IsDirectCount flags
// 4 bytes for the number of elements in lookupSparse
// 2+1 bytes for each element (ulong) in lookupSparse
Assert.AreEqual(13 + 3*data.LookupSparse.Count, results.Length);
Assert.AreEqual(14, BitConverter.ToInt32(results.Skip(4).Take(4).ToArray(), 0)); // Bits in Index = 14
Assert.AreEqual(2, results[8]); // IsSparse = true AND IsDirectCount = false
Assert.AreEqual(data.LookupSparse.Count, BitConverter.ToInt32(results.Skip(9).Take(4).ToArray(), 0));
}
private void TestDeserializer(int cardinality)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(cardinality);
CardinalityEstimator hll2;
var serializer = new CardinalityEstimatorSerializer();
byte[] results;
using (var memoryStream = new MemoryStream())
{
serializer.Serialize(memoryStream, hll);
results = memoryStream.ToArray();
}
using (var memoryStream = new MemoryStream(results))
{
hll2 = serializer.Deserialize(memoryStream);
}
CardinalityEstimatorState data = hll.GetState();
CardinalityEstimatorState data2 = hll2.GetState();
Assert.AreEqual(data.BitsPerIndex, data2.BitsPerIndex);
Assert.AreEqual(data.IsSparse, data2.IsSparse);
Assert.IsTrue((data.DirectCount != null && data2.DirectCount != null) || (data.DirectCount == null && data2.DirectCount == null));
Assert.IsTrue((data.LookupSparse != null && data2.LookupSparse != null) ||
(data.LookupSparse == null && data2.LookupSparse == null));
Assert.IsTrue((data.LookupDense != null && data2.LookupDense != null) || (data.LookupDense == null && data2.LookupDense == null));
if (data.DirectCount != null)
{
// DirectCount are subsets of each-other => they are the same set
Assert.IsTrue(data.DirectCount.IsSubsetOf(data2.DirectCount) && data2.DirectCount.IsSubsetOf(data.DirectCount));
}
if (data.LookupSparse != null)
{
Assert.IsTrue(data.LookupSparse.DictionaryEqual(data2.LookupSparse));
}
if (data.LookupDense != null)
{
Assert.IsTrue(data.LookupDense.SequenceEqual(data2.LookupDense));
}
}
private void TestSerializerCreatesSmallerData(int cardinality, out int customSize, out int defaultSize)
{
CardinalityEstimator hll = CreateAndFillCardinalityEstimator(cardinality);
var customSerializer = new CardinalityEstimatorSerializer();
byte[] customSerializerResults;
using (var memoryStream = new MemoryStream())
{
customSerializer.Serialize(memoryStream, hll);
customSerializerResults = memoryStream.ToArray();
customSize = customSerializerResults.Length;
}
var binaryFormatter = new BinaryFormatter();
byte[] defaultSerializerResults;
using (var memoryStream = new MemoryStream())
{
binaryFormatter.Serialize(memoryStream, hll);
defaultSerializerResults = memoryStream.ToArray();
defaultSize = defaultSerializerResults.Length;
}
Assert.IsTrue(customSerializerResults.Length <= defaultSerializerResults.Length);
}