public void testBloomFilterByte()
{
BloomFilter bf = new BloomFilter(10000);
byte val = Byte.MinValue;
byte val1 = 1;
byte val2 = 2;
byte val3 = Byte.MaxValue;
Assert.Equal(false, bf.testLong(val));
Assert.Equal(false, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(false, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val1);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val2);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(true, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val3);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(true, bf.testLong(val2));
Assert.Equal(true, bf.testLong(val3));
byte randVal = 0;
for (int i = 0; i < COUNT; i++)
{
randVal = (byte)rand.Next(Byte.MaxValue);
bf.addLong(randVal);
}
// last value should be present
Assert.Equal(true, bf.testLong(randVal));
// most likely this value should not exist
Assert.Equal(false, bf.testLong(unchecked((byte)-120)));
Assert.Equal(7800, bf.sizeInBytes());
}
public void testBloomFilterString()
{
BloomFilter bf = new BloomFilter(100000);
string val = "bloo";
string val1 = "bloom fil";
string val2 = "bloom filter";
string val3 = "cuckoo filter";
Assert.Equal(false, bf.testString(val));
Assert.Equal(false, bf.testString(val1));
Assert.Equal(false, bf.testString(val2));
Assert.Equal(false, bf.testString(val3));
bf.addString(val);
Assert.Equal(true, bf.testString(val));
Assert.Equal(false, bf.testString(val1));
Assert.Equal(false, bf.testString(val2));
Assert.Equal(false, bf.testString(val3));
bf.addString(val1);
Assert.Equal(true, bf.testString(val));
Assert.Equal(true, bf.testString(val1));
Assert.Equal(false, bf.testString(val2));
Assert.Equal(false, bf.testString(val3));
bf.addString(val2);
Assert.Equal(true, bf.testString(val));
Assert.Equal(true, bf.testString(val1));
Assert.Equal(true, bf.testString(val2));
Assert.Equal(false, bf.testString(val3));
bf.addString(val3);
Assert.Equal(true, bf.testString(val));
Assert.Equal(true, bf.testString(val1));
Assert.Equal(true, bf.testString(val2));
Assert.Equal(true, bf.testString(val3));
long randVal = 0;
for (int i = 0; i < COUNT; i++)
{
randVal = rand.NextLong();
bf.addString(randVal.ToString(CultureInfo.InvariantCulture));
}
// last value should be present
Assert.Equal(true, bf.testString(randVal.ToString(CultureInfo.InvariantCulture)));
// most likely this value should not exist
Assert.Equal(false, bf.testString((-120L).ToString(CultureInfo.InvariantCulture)));
Assert.Equal(77944, bf.sizeInBytes());
}
public void testBloomFilterFloat()
{
BloomFilter bf = new BloomFilter(10000);
float val = Single.MinValue;
float val1 = 1.1f;
float val2 = 2.2f;
float val3 = Single.MaxValue;
Assert.Equal(false, bf.testDouble(val));
Assert.Equal(false, bf.testDouble(val1));
Assert.Equal(false, bf.testDouble(val2));
Assert.Equal(false, bf.testDouble(val3));
bf.addDouble(val);
Assert.Equal(true, bf.testDouble(val));
Assert.Equal(false, bf.testDouble(val1));
Assert.Equal(false, bf.testDouble(val2));
Assert.Equal(false, bf.testDouble(val3));
bf.addDouble(val1);
Assert.Equal(true, bf.testDouble(val));
Assert.Equal(true, bf.testDouble(val1));
Assert.Equal(false, bf.testDouble(val2));
Assert.Equal(false, bf.testDouble(val3));
bf.addDouble(val2);
Assert.Equal(true, bf.testDouble(val));
Assert.Equal(true, bf.testDouble(val1));
Assert.Equal(true, bf.testDouble(val2));
Assert.Equal(false, bf.testDouble(val3));
bf.addDouble(val3);
Assert.Equal(true, bf.testDouble(val));
Assert.Equal(true, bf.testDouble(val1));
Assert.Equal(true, bf.testDouble(val2));
Assert.Equal(true, bf.testDouble(val3));
float randVal = 0;
for (int i = 0; i < COUNT; i++)
{
randVal = rand.NextFloat();
bf.addDouble(randVal);
}
// last value should be present
Assert.Equal(true, bf.testDouble(randVal));
// most likely this value should not exist
Assert.Equal(false, bf.testDouble(-120.2f));
Assert.Equal(7800, bf.sizeInBytes());
}
public void testBloomFilterLong()
{
BloomFilter bf = new BloomFilter(10000);
long val = Int64.MinValue;
long val1 = 1;
long val2 = 2;
long val3 = Int64.MaxValue;
Assert.Equal(false, bf.testLong(val));
Assert.Equal(false, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(false, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val1);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(false, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val2);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(true, bf.testLong(val2));
Assert.Equal(false, bf.testLong(val3));
bf.addLong(val3);
Assert.Equal(true, bf.testLong(val));
Assert.Equal(true, bf.testLong(val1));
Assert.Equal(true, bf.testLong(val2));
Assert.Equal(true, bf.testLong(val3));
long randVal = 0;
for (int i = 0; i < COUNT; i++)
{
randVal = rand.NextLong();
bf.addLong(randVal);
}
// last value should be present
Assert.Equal(true, bf.testLong(randVal));
// most likely this value should not exist
Assert.Equal(false, bf.testLong(-120));
Assert.Equal(7800, bf.sizeInBytes());
}
/**
* Merge the specified bloom filter with current bloom filter.
*
* @param that - bloom filter to merge
*/
public void merge(BloomFilter that)
{
if (this != that && this.numBits == that.numBits && this.numHashFunctions == that.numHashFunctions)
{
this.bitSet.putAll(that.bitSet);
}
else
{
throw new ArgumentException("BloomFilters are not compatible for merging." +
" this - " + this.ToString() + " that - " + that.ToString());
}
}
/**
* Create a tree writer.
* @param columnId the column id of the column to write
* @param inspector the object inspector to use
* @param schema the row schema
* @param streamFactory limited access to the Writer's data.
* @param nullable can the value be null?
* @
*/
protected TreeWriter(
int columnId,
ObjectInspector inspector,
TypeDescription schema,
StreamFactory streamFactory,
bool nullable)
{
this.streamFactory = streamFactory;
this.isCompressed = streamFactory.isCompressed();
this.id = columnId;
this.inspector = inspector;
if (nullable)
{
isPresentOutStream = streamFactory.createStream(id,
OrcProto.Stream.Types.Kind.PRESENT);
isPresent = new BitFieldWriter(isPresentOutStream, 1);
}
else
{
isPresent = null;
}
this.foundNulls = false;
createBloomFilter = streamFactory.getBloomFilterColumns()[columnId];
indexStatistics = ColumnStatisticsImpl.create(schema);
stripeColStatistics = ColumnStatisticsImpl.create(schema);
fileStatistics = ColumnStatisticsImpl.create(schema);
childrenWriters = new TreeWriter[0];
rowIndex = OrcProto.RowIndex.CreateBuilder();
rowIndexEntry = OrcProto.RowIndexEntry.CreateBuilder();
rowIndexPosition = new RowIndexPositionRecorder(rowIndexEntry);
stripeStatsBuilders = new List<OrcProto.StripeStatistics.Builder>();
if (streamFactory.buildIndex())
{
rowIndexStream = streamFactory.createStream(id, OrcProto.Stream.Types.Kind.ROW_INDEX);
}
else
{
rowIndexStream = null;
}
if (createBloomFilter)
{
bloomFilterEntry = OrcProto.BloomFilter.CreateBuilder();
bloomFilterIndex = OrcProto.BloomFilterIndex.CreateBuilder();
bloomFilterStream = streamFactory.createStream(id, OrcProto.Stream.Types.Kind.BLOOM_FILTER);
bloomFilter = new BloomFilter(streamFactory.getRowIndexStride(), streamFactory.getBloomFilterFPP());
}
else
{
bloomFilterEntry = null;
bloomFilterIndex = null;
bloomFilterStream = null;
bloomFilter = null;
}
}
public void testIntNullSafeEqualsBloomFilter()
{
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf(
PredicateLeaf.Operator.NULL_SAFE_EQUALS, PredicateLeaf.Type.LONG, "x", 15L, null);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addLong(i);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createIntStats(10, 100));
Assert.Equal(TruthValue.NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addLong(15);
Assert.Equal(TruthValue.YES_NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testStringInBloomFilter()
{
List<object> args = new List<object>();
args.Add("str_15");
args.Add("str_19");
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf
(PredicateLeaf.Operator.IN, PredicateLeaf.Type.STRING,
"x", null, args);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addString("str_" + i);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createStringStats("str_10", "str_200"));
Assert.Equal(TruthValue.NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString("str_19");
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString("str_15");
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testDoubleEqualsBloomFilter()
{
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf(
PredicateLeaf.Operator.EQUALS, PredicateLeaf.Type.FLOAT, "x", 15.0, null);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addDouble(i);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createDoubleStats(10.0, 100.0));
Assert.Equal(TruthValue.NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addDouble(15.0);
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testIntInBloomFilter()
{
List<object> args = new List<object>();
args.Add(15L);
args.Add(19L);
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf
(PredicateLeaf.Operator.IN, PredicateLeaf.Type.LONG,
"x", null, args);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addLong(i);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createIntStats(10, 100));
Assert.Equal(TruthValue.NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addLong(19);
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addLong(15);
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testDecimalNullSafeEqualsBloomFilter()
{
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf(
PredicateLeaf.Operator.NULL_SAFE_EQUALS, PredicateLeaf.Type.DECIMAL, "x",
HiveDecimal.Parse("15"),
null);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addString(HiveDecimal.create(i).ToString());
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createDecimalStats("10", "200"));
Assert.Equal(TruthValue.NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString(HiveDecimal.create(15).ToString());
Assert.Equal(TruthValue.YES_NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testTimestampNullSafeEqualsBloomFilter()
{
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf(
PredicateLeaf.Operator.NULL_SAFE_EQUALS, PredicateLeaf.Type.TIMESTAMP, "x",
new Timestamp(15),
null);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addLong((new Timestamp(i)).Milliseconds);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createTimestampStats(10, 100));
Assert.Equal(TruthValue.NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addLong((new Timestamp(15)).Milliseconds);
Assert.Equal(TruthValue.YES_NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testStringNullSafeEqualsBloomFilter()
{
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf(
PredicateLeaf.Operator.NULL_SAFE_EQUALS, PredicateLeaf.Type.STRING, "x", "str_15", null);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addString("str_" + i);
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createStringStats("str_10", "str_200"));
Assert.Equal(TruthValue.NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString("str_15");
Assert.Equal(TruthValue.YES_NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testMerge()
{
BloomFilter bf = new BloomFilter(10000);
string val = "bloo";
string val1 = "bloom fil";
string val2 = "bloom filter";
string val3 = "cuckoo filter";
bf.addString(val);
bf.addString(val1);
bf.addString(val2);
bf.addString(val3);
BloomFilter bf2 = new BloomFilter(10000);
string v = "2_bloo";
string v1 = "2_bloom fil";
string v2 = "2_bloom filter";
string v3 = "2_cuckoo filter";
bf2.addString(v);
bf2.addString(v1);
bf2.addString(v2);
bf2.addString(v3);
Assert.Equal(true, bf.testString(val));
Assert.Equal(true, bf.testString(val1));
Assert.Equal(true, bf.testString(val2));
Assert.Equal(true, bf.testString(val3));
Assert.Equal(false, bf.testString(v));
Assert.Equal(false, bf.testString(v1));
Assert.Equal(false, bf.testString(v2));
Assert.Equal(false, bf.testString(v3));
bf.merge(bf2);
Assert.Equal(true, bf.testString(val));
Assert.Equal(true, bf.testString(val1));
Assert.Equal(true, bf.testString(val2));
Assert.Equal(true, bf.testString(val3));
Assert.Equal(true, bf.testString(v));
Assert.Equal(true, bf.testString(v1));
Assert.Equal(true, bf.testString(v2));
Assert.Equal(true, bf.testString(v3));
}
public void testNullsInBloomFilter()
{
List<object> args = new List<object>();
args.Add(HiveDecimal.Parse("15"));
args.Add(null);
args.Add(HiveDecimal.Parse("19"));
PredicateLeaf pred = TestSearchArgumentImpl.createPredicateLeaf
(PredicateLeaf.Operator.IN, PredicateLeaf.Type.DECIMAL,
"x", null, args);
BloomFilter bf = new BloomFilter(10000);
for (int i = 20; i < 1000; i++)
{
bf.addString(HiveDecimal.create(i).ToString());
}
ColumnStatistics cs = ColumnStatisticsImpl.deserialize(createDecimalStats("10", "200", false));
// hasNull is false, so bloom filter should return NO
Assert.Equal(TruthValue.NO, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
cs = ColumnStatisticsImpl.deserialize(createDecimalStats("10", "200", true));
// hasNull is true, so bloom filter should return YES_NO_NULL
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString(HiveDecimal.create(19).ToString());
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
bf.addString(HiveDecimal.create(15).ToString());
Assert.Equal(TruthValue.YES_NO_NULL, RecordReaderImpl.evaluatePredicate(cs, pred, bf));
}
public void testBloomFilterBytes()
{
BloomFilter bf = new BloomFilter(10000);
byte[] val = new byte[] { 1, 2, 3 };
byte[] val1 = new byte[] { 1, 2, 3, 4 };
byte[] val2 = new byte[] { 1, 2, 3, 4, 5 };
byte[] val3 = new byte[] { 1, 2, 3, 4, 5, 6 };
Assert.Equal(false, bf.test(val));
Assert.Equal(false, bf.test(val1));
Assert.Equal(false, bf.test(val2));
Assert.Equal(false, bf.test(val3));
bf.add(val);
Assert.Equal(true, bf.test(val));
Assert.Equal(false, bf.test(val1));
Assert.Equal(false, bf.test(val2));
Assert.Equal(false, bf.test(val3));
bf.add(val1);
Assert.Equal(true, bf.test(val));
Assert.Equal(true, bf.test(val1));
Assert.Equal(false, bf.test(val2));
Assert.Equal(false, bf.test(val3));
bf.add(val2);
Assert.Equal(true, bf.test(val));
Assert.Equal(true, bf.test(val1));
Assert.Equal(true, bf.test(val2));
Assert.Equal(false, bf.test(val3));
bf.add(val3);
Assert.Equal(true, bf.test(val));
Assert.Equal(true, bf.test(val1));
Assert.Equal(true, bf.test(val2));
Assert.Equal(true, bf.test(val3));
byte[] randVal = new byte[COUNT];
for (int i = 0; i < COUNT; i++)
{
rand.NextBytes(randVal);
bf.add(randVal);
}
// last value should be present
Assert.Equal(true, bf.test(randVal));
// most likely this value should not exist
randVal[0] = 0;
randVal[1] = 0;
randVal[2] = 0;
randVal[3] = 0;
randVal[4] = 0;
Assert.Equal(false, bf.test(randVal));
Assert.Equal(7800, bf.sizeInBytes());
}
private static void writeBloomFilterStats(JsonWriter writer, BloomFilter bf)
{
int bitCount = bf.getBitSize();
int popCount = 0;
foreach (long l in bf.getBitSet())
{
popCount += Long.NumberOfOnes(l);
}
int k = bf.getNumHashFunctions();
float loadFactor = (float)popCount / (float)bitCount;
float expectedFpp = (float)Math.Pow(loadFactor, k);
writer.key("numHashFunctions").value(k);
writer.key("bitCount").value(bitCount);
writer.key("popCount").value(popCount);
writer.key("loadFactor").value(loadFactor);
writer.key("expectedFpp").value(expectedFpp);
}
