The Big License Swap, package restructure, other cleanup.
Every file is touched.
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicFinal.java
index 8bdd0dd..ef6b5b0 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicFinal.java
@@ -21,15 +21,14 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
+import org.apache.datasketches.cpc.CpcUnion;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.cpc.CpcSketch;
-import org.apache.datasketches.cpc.CpcUnion;
-
/**
* Class used to calculate the final pass of an <i>Algebraic</i> sketch
* operation. It will receive a bag of values returned by either the <i>Intermediate</i>
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicIntermediate.java
index 385837b..7f506b7 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/AlgebraicIntermediate.java
@@ -21,14 +21,13 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.cpc.CpcSketch;
-
/**
* Class used to calculate the intermediate combiner pass of an <i>Algebraic</i> sketch
* operation. This is called from the combiner, and may be called multiple times (from a mapper
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketch.java b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketch.java
index 2013c8c..30c656c 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketch.java
@@ -23,6 +23,7 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
@@ -33,8 +34,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.cpc.CpcSketch;
-
/**
* This is a Pig UDF that builds sketches from data.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicFinal.java
index 809cb0a..c97d1d1 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicFinal.java
@@ -23,6 +23,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class DataToSketchAlgebraicFinal extends AlgebraicFinal {
/**
@@ -47,7 +48,7 @@
/**
* Constructor for the final pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param seed for the hash function
*/
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicIntermediate.java
index d13d254..14e48a9 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/DataToSketchAlgebraicIntermediate.java
@@ -23,6 +23,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class DataToSketchAlgebraicIntermediate extends AlgebraicIntermediate {
/**
@@ -47,7 +48,7 @@
/**
* Constructor for the intermediate pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param seed for the hash function
*/
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/GetEstimate.java b/src/main/java/org/apache/datasketches/pig/cpc/GetEstimate.java
index 88c4a35..485e699 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/GetEstimate.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/GetEstimate.java
@@ -23,12 +23,11 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.cpc.CpcSketch;
-
/**
* This is a User Defined Function (UDF) for getting a distinct count estimate from a given CpcdSketch
*
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBounds.java b/src/main/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBounds.java
index 9b2ee6d..a9d68f8 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBounds.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBounds.java
@@ -23,6 +23,7 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.DataType;
@@ -31,8 +32,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.cpc.CpcSketch;
-
/**
* This is a User Defined Function (UDF) for obtaining the distinct count estimate
* along with a lower and upper bound from a given CpcSketch.
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/SketchToString.java b/src/main/java/org/apache/datasketches/pig/cpc/SketchToString.java
index 669eab6..e123820 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/SketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/SketchToString.java
@@ -23,12 +23,11 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.cpc.CpcSketch;
-
/**
* This is a User Defined Function (UDF) for printing a human-readable summary of a given CpcSketch
* @author Alexander Saydakov
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketch.java b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketch.java
index 5c26c3c..f311928 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketch.java
@@ -23,6 +23,8 @@
import java.io.IOException;
+import org.apache.datasketches.cpc.CpcSketch;
+import org.apache.datasketches.cpc.CpcUnion;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
@@ -33,9 +35,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.cpc.CpcSketch;
-import org.apache.datasketches.cpc.CpcUnion;
-
/**
* This is a Pig UDF that performs the Union operation on CpcSketches.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicFinal.java
index 3c247b9..04e741f 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicFinal.java
@@ -23,6 +23,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class UnionSketchAlgebraicFinal extends AlgebraicFinal {
/**
@@ -47,7 +48,7 @@
/**
* Constructor for the final pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param seed for the hash function
*/
diff --git a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicIntermediate.java
index 1669981..2f6b78d 100644
--- a/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/cpc/UnionSketchAlgebraicIntermediate.java
@@ -23,6 +23,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class UnionSketchAlgebraicIntermediate extends AlgebraicIntermediate {
/**
@@ -47,7 +48,7 @@
/**
* Constructor for the intermediate pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param seed for the hash function
*/
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketch.java b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketch.java
index c8c74f2..fe1345d 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketch.java
@@ -21,6 +21,8 @@
import java.io.IOException;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -28,9 +30,6 @@
import org.apache.pig.data.DataBag;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
/**
* This is a generic implementation to be specialized in concrete UDFs
* @param <T> type of item
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketchAlgebraicIntermediateFinal.java
index d2b42aa..4533c88 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentItemsSketchAlgebraicIntermediateFinal.java
@@ -21,15 +21,14 @@
import java.io.IOException;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
/**
* Class used to calculate the intermediate pass (combiner) or the final pass
* (reducer) of an Algebraic sketch operation. This may be called multiple times
@@ -49,6 +48,7 @@
*/
public DataToFrequentItemsSketchAlgebraicIntermediateFinal() {}
+ @SuppressWarnings("javadoc")
public DataToFrequentItemsSketchAlgebraicIntermediateFinal(
final int sketchSize, final ArrayOfItemsSerDe<T> serDe) {
sketchSize_ = sketchSize;
@@ -62,7 +62,7 @@
Logger.getLogger(getClass()).info("algebraic was used");
isFirstCall_ = false;
}
- final ItemsSketch<T> sketch = new ItemsSketch<T>(sketchSize_);
+ final ItemsSketch<T> sketch = new ItemsSketch<>(sketchSize_);
final DataBag bag = (DataBag) inputTuple.get(0);
for (Tuple dataTuple: bag) {
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketch.java b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketch.java
index bd56c5f..e7d5763 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketch.java
@@ -19,9 +19,8 @@
package org.apache.datasketches.pig.frequencies;
-import org.apache.pig.Algebraic;
-
import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.pig.Algebraic;
/**
* This UDF creates a FrequentItemsSketch<String> from raw data.
@@ -52,6 +51,7 @@
return IntermediateFinal.class.getName();
}
+ @SuppressWarnings("javadoc")
public static class Initial extends AlgebraicInitial {
/**
* Constructor for the initial pass of an Algebraic function. This will be passed the same
@@ -66,6 +66,7 @@
public Initial() {}
}
+ @SuppressWarnings("javadoc")
public static class IntermediateFinal extends DataToFrequentItemsSketchAlgebraicIntermediateFinal<String> {
/**
* Constructor for the intermediate and final passes of an Algebraic function. This will be
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimates.java b/src/main/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimates.java
index 7f17343..96f1d5a 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimates.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimates.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.frequencies.ErrorType;
+import org.apache.datasketches.frequencies.ItemsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.BagFactory;
import org.apache.pig.data.DataBag;
@@ -31,11 +35,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.frequencies.ErrorType;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
/**
* This UDF converts a FrequentItemsSketch<String> to estimates:
* {(item, estimate, upper bound, lower bound), ...}
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketch.java b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketch.java
index 9167a1a..9707f25 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketch.java
@@ -21,6 +21,8 @@
import java.io.IOException;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -28,9 +30,6 @@
import org.apache.pig.data.DataBag;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
/**
* This is a generic implementation to be specialized in concrete UDFs
* @param <T> Type of item
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketchAlgebraicIntermediateFinal.java
index 1473014..8744354 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentItemsSketchAlgebraicIntermediateFinal.java
@@ -21,15 +21,14 @@
import java.io.IOException;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
/**
* This is to calculate the intermediate pass (combiner) or the final pass
* (reducer) of an Algebraic sketch operation. This may be called multiple times
@@ -49,6 +48,7 @@
*/
public UnionFrequentItemsSketchAlgebraicIntermediateFinal() {}
+ @SuppressWarnings("javadoc")
public UnionFrequentItemsSketchAlgebraicIntermediateFinal(
final int sketchSize, final ArrayOfItemsSerDe<T> serDe) {
sketchSize_ = sketchSize;
@@ -62,7 +62,7 @@
Logger.getLogger(getClass()).info("algebraic is used");
isFirstCall_ = false;
}
- final ItemsSketch<T> sketch = new ItemsSketch<T>(sketchSize_);
+ final ItemsSketch<T> sketch = new ItemsSketch<>(sketchSize_);
final DataBag bag = (DataBag) inputTuple.get(0);
if (bag == null) {
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketch.java b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketch.java
index 5abf23b..fa6b6bc 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketch.java
@@ -19,9 +19,8 @@
package org.apache.datasketches.pig.frequencies;
-import org.apache.pig.Algebraic;
-
import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.pig.Algebraic;
/**
* This is to union FrequentItemsSketch<String>.
@@ -52,6 +51,7 @@
return IntermediateFinal.class.getName();
}
+ @SuppressWarnings("javadoc")
public static class Initial extends AlgebraicInitial {
/**
* Default constructor to make pig validation happy.
@@ -66,6 +66,7 @@
public Initial(final String sketchSize) {}
}
+ @SuppressWarnings("javadoc")
public static class IntermediateFinal extends UnionFrequentItemsSketchAlgebraicIntermediateFinal<String> {
/**
* Default constructor to make pig validation happy.
diff --git a/src/main/java/org/apache/datasketches/pig/frequencies/Util.java b/src/main/java/org/apache/datasketches/pig/frequencies/Util.java
index 5bd21cd..4fec7fe 100644
--- a/src/main/java/org/apache/datasketches/pig/frequencies/Util.java
+++ b/src/main/java/org/apache/datasketches/pig/frequencies/Util.java
@@ -19,15 +19,14 @@
package org.apache.datasketches.pig.frequencies;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.frequencies.ItemsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.frequencies.ItemsSketch;
-
final class Util {
static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/main/java/org/apache/datasketches/pig/hash/MurmurHash3.java b/src/main/java/org/apache/datasketches/pig/hash/MurmurHash3.java
index ab4fd42..5de7173 100644
--- a/src/main/java/org/apache/datasketches/pig/hash/MurmurHash3.java
+++ b/src/main/java/org/apache/datasketches/pig/hash/MurmurHash3.java
@@ -19,11 +19,11 @@
package org.apache.datasketches.pig.hash;
+import static java.lang.Math.min;
+import static java.nio.charset.StandardCharsets.UTF_8;
import static org.apache.datasketches.hash.MurmurHash3.hash;
import static org.apache.datasketches.hash.MurmurHash3Adaptor.hashToLongs;
import static org.apache.datasketches.hash.MurmurHash3Adaptor.modulo;
-import static java.lang.Math.min;
-import static java.nio.charset.StandardCharsets.UTF_8;
import java.io.IOException;
@@ -41,8 +41,6 @@
* parameter can be a positive Integer modulus divisor. If the divisor is provided, the Integer
* modulus remainder is computed on the entire 128-bit hash output treated as if it were a 128-bit
* positive value.
- *
- * @author Lee Rhodes
*/
public class MurmurHash3 extends EvalFunc<Tuple> {
private static final TupleFactory mTupleFactory = TupleFactory.getInstance();
diff --git a/src/main/java/org/apache/datasketches/pig/hash/package-info.java b/src/main/java/org/apache/datasketches/pig/hash/package-info.java
index e7317cf..0a8bbd2 100644
--- a/src/main/java/org/apache/datasketches/pig/hash/package-info.java
+++ b/src/main/java/org/apache/datasketches/pig/hash/package-info.java
@@ -18,14 +18,12 @@
*/
/**
- * <p>The hash package contains a high-performing and extended Java implementation
- * of Austin Appleby's 128-bit MurmurHash3 hash function originally coded in C.
- * This core MurmurHash3.java class is used throughout all the sketch classes for consistentancy
+ * <p>The hash package contains a high-performing and extended Java implementation
+ * of Austin Appleby's 128-bit MurmurHash3 hash function originally coded in C.
+ * This core MurmurHash3.java class is used throughout all the sketch classes for consistentancy
* and as long as the user specifies the same seed will result in coordinated hash operations.
- * This package also contains an adaptor class that extends the basic class with more functions
+ * This package also contains an adaptor class that extends the basic class with more functions
* commonly associated with hashing.
* </p>
- *
- * @author Lee Rhodes
*/
package org.apache.datasketches.pig.hash;
diff --git a/src/main/java/org/apache/datasketches/pig/hll/AlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/hll/AlgebraicFinal.java
index 8801739..c8bcc88 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/AlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/AlgebraicFinal.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
+import org.apache.datasketches.memory.Memory;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
@@ -28,11 +32,6 @@
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
/**
* Class used to calculate the final pass of an <i>Algebraic</i> sketch
* operation. It will receive a bag of values returned by either the <i>Intermediate</i>
diff --git a/src/main/java/org/apache/datasketches/pig/hll/AlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/hll/AlgebraicIntermediate.java
index 079b7a5..db10f62 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/AlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/AlgebraicIntermediate.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
+import org.apache.datasketches.memory.Memory;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
@@ -29,11 +33,6 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
/**
* Class used to calculate the intermediate combiner pass of an <i>Algebraic</i> sketch
* operation. This is called from the combiner, and may be called multiple times (from a mapper
diff --git a/src/main/java/org/apache/datasketches/pig/hll/DataToSketch.java b/src/main/java/org/apache/datasketches/pig/hll/DataToSketch.java
index 6244fc3..6e38253 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/DataToSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/DataToSketch.java
@@ -21,6 +21,9 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
@@ -31,10 +34,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.hll.HllSketch;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
/**
* This is a Pig UDF that builds Sketches from data.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicFinal.java
index 968b818..69a9e58 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicFinal.java
@@ -22,12 +22,12 @@
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_HLL_TYPE;
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_LG_K;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataBag;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
+@SuppressWarnings("javadoc")
public class DataToSketchAlgebraicFinal extends AlgebraicFinal {
/**
@@ -52,7 +52,7 @@
/**
* Constructor for the final pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param tgtHllType HLL type of the resulting sketch
*/
diff --git a/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicIntermediate.java
index 3962967..bb184dc 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/DataToSketchAlgebraicIntermediate.java
@@ -22,12 +22,12 @@
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_HLL_TYPE;
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_LG_K;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataBag;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
+@SuppressWarnings("javadoc")
public class DataToSketchAlgebraicIntermediate extends AlgebraicIntermediate {
/**
@@ -52,7 +52,7 @@
/**
* Constructor for the intermediate pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param tgtHllType HLL type of the resulting sketch
*/
diff --git a/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimate.java b/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimate.java
index 089427a..81a1f89 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimate.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimate.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-
/**
* This is a User Defined Function (UDF) for getting a unique count estimate from an HllSketch
*
diff --git a/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBounds.java b/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBounds.java
index 9211e0f..e5e3907 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBounds.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBounds.java
@@ -21,6 +21,8 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.DataType;
@@ -29,9 +31,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-
/**
* This is a User Defined Function (UDF) for obtaining the unique count estimate
* along with a lower and upper bound from an HllSketch.
diff --git a/src/main/java/org/apache/datasketches/pig/hll/SketchToString.java b/src/main/java/org/apache/datasketches/pig/hll/SketchToString.java
index 17368e3..2dc0bb4 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/SketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/SketchToString.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-
/**
* This is a User Defined Function (UDF) for "pretty printing" the summary of an HllSketch
*
diff --git a/src/main/java/org/apache/datasketches/pig/hll/UnionSketch.java b/src/main/java/org/apache/datasketches/pig/hll/UnionSketch.java
index bec2396..a984db4 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/UnionSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/UnionSketch.java
@@ -24,6 +24,10 @@
import java.io.IOException;
+import org.apache.datasketches.hll.HllSketch;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
+import org.apache.datasketches.memory.Memory;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
@@ -34,11 +38,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.hll.HllSketch;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
/**
* This is a Pig UDF that performs the Union operation on HllSketches.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicFinal.java b/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicFinal.java
index 04389cf..0af607c 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicFinal.java
@@ -22,12 +22,12 @@
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_HLL_TYPE;
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_LG_K;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataBag;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
+@SuppressWarnings("javadoc")
public class UnionSketchAlgebraicFinal extends AlgebraicFinal {
/**
@@ -52,7 +52,7 @@
/**
* Constructor for the final pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param tgtHllType HLL type of the resulting sketch
*/
diff --git a/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicIntermediate.java b/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicIntermediate.java
index 666aae9..50e4c6f 100644
--- a/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicIntermediate.java
+++ b/src/main/java/org/apache/datasketches/pig/hll/UnionSketchAlgebraicIntermediate.java
@@ -22,12 +22,12 @@
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_HLL_TYPE;
import static org.apache.datasketches.pig.hll.DataToSketch.DEFAULT_LG_K;
+import org.apache.datasketches.hll.TgtHllType;
+import org.apache.datasketches.hll.Union;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataBag;
-import org.apache.datasketches.hll.TgtHllType;
-import org.apache.datasketches.hll.Union;
-
+@SuppressWarnings("javadoc")
public class UnionSketchAlgebraicIntermediate extends AlgebraicIntermediate {
/**
@@ -52,7 +52,7 @@
/**
* Constructor for the intermediate pass of an Algebraic function. Pig will call
* this and pass the same constructor arguments as the base UDF.
- *
+ *
* @param lgK parameter controlling the sketch size and accuracy
* @param tgtHllType HLL type of the resulting sketch
*/
diff --git a/src/main/java/org/apache/datasketches/pig/kll/DataToSketch.java b/src/main/java/org/apache/datasketches/pig/kll/DataToSketch.java
index 69dc7be..ee6977f 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/DataToSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/DataToSketch.java
@@ -21,6 +21,8 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -29,9 +31,6 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to build sketches from data.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces for
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetCdf.java b/src/main/java/org/apache/datasketches/pig/kll/GetCdf.java
index 331b80e..f39f03d 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetCdf.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetCdf.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get an approximation to the Cumulative Distribution Function (CDF) of the input stream
* given a sketch and a set of split points - an array of <i>m</i> unique, monotonically increasing
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetK.java b/src/main/java/org/apache/datasketches/pig/kll/GetK.java
index 98677bc..9ea07e7 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetK.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetK.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get the parameter K from a given sketch.
* This can be useful for debugging a work flow to make sure that resulting sketches
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetPmf.java b/src/main/java/org/apache/datasketches/pig/kll/GetPmf.java
index 8d376c9..f05885e 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetPmf.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetPmf.java
@@ -21,15 +21,14 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get an approximation to the Probability Mass Function (PMF) of the input stream
* given a sketch and a set of split points - an array of <i>m</i> unique, monotonically increasing
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetQuantile.java b/src/main/java/org/apache/datasketches/pig/kll/GetQuantile.java
index 8e3113e..5035294 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetQuantile.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetQuantile.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get a quantile value from a given sketch. A single value for a
* given fraction is returned. The fraction represents a normalized rank and must be
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetQuantiles.java b/src/main/java/org/apache/datasketches/pig/kll/GetQuantiles.java
index 60829f9..f2ac968 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetQuantiles.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetQuantiles.java
@@ -22,15 +22,14 @@
import java.io.IOException;
import java.util.Arrays;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get a list of quantile values from a given sketch and a list of
* fractions or a number of evenly spaced intervals. The fractions represent normalized ranks and
diff --git a/src/main/java/org/apache/datasketches/pig/kll/GetRank.java b/src/main/java/org/apache/datasketches/pig/kll/GetRank.java
index 093aefc..bb99c48 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/GetRank.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/GetRank.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get a normalized rank for a given value from a given sketch. A single
* rank for a given value is returned. The normalized rank is a double value
diff --git a/src/main/java/org/apache/datasketches/pig/kll/SketchToString.java b/src/main/java/org/apache/datasketches/pig/kll/SketchToString.java
index 565f5f9..52e25d9 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/SketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/SketchToString.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to get a human-readable summary of a given sketch.
*/
diff --git a/src/main/java/org/apache/datasketches/pig/kll/UnionSketch.java b/src/main/java/org/apache/datasketches/pig/kll/UnionSketch.java
index 5a66f6a..676b1eb 100644
--- a/src/main/java/org/apache/datasketches/pig/kll/UnionSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/kll/UnionSketch.java
@@ -21,6 +21,8 @@
import java.io.IOException;
+import org.apache.datasketches.kll.KllFloatsSketch;
+import org.apache.datasketches.memory.Memory;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -30,9 +32,6 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.kll.KllFloatsSketch;
-
/**
* This UDF is to merge sketches.
* This class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces for
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketch.java
index d955051..42aa8ce 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketch.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
+import org.apache.datasketches.quantiles.DoublesUnion;
+import org.apache.datasketches.quantiles.DoublesUnionBuilder;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -32,11 +36,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-import org.apache.datasketches.quantiles.DoublesUnion;
-import org.apache.datasketches.quantiles.DoublesUnionBuilder;
-
/**
* This is a Pig UDF that builds Sketches from data.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/DataToItemsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/DataToItemsSketch.java
index 7d55f55..8d7c9b0 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/DataToItemsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/DataToItemsSketch.java
@@ -22,6 +22,10 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
+import org.apache.datasketches.quantiles.ItemsUnion;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -33,11 +37,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-import org.apache.datasketches.quantiles.ItemsUnion;
-
/**
* Builds ItemsSketch from data.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/DataToStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/DataToStringsSketch.java
index d8e1f4e..58fbcb5 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/DataToStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/DataToStringsSketch.java
@@ -28,6 +28,7 @@
* Creates an ItemsSketch<String> from raw data.
* It supports all three ways: exec(), Accumulator and Algebraic.
*/
+@SuppressWarnings("javadoc")
public class DataToStringsSketch extends DataToItemsSketch<String> {
private static final Comparator<String> COMPARATOR = Comparator.naturalOrder();
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/DoublesSketchToString.java b/src/main/java/org/apache/datasketches/pig/quantiles/DoublesSketchToString.java
index ef7b653..98c0797 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/DoublesSketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/DoublesSketchToString.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-
/**
* This UDF is to get a human-readable summary of a given sketch.
*/
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketch.java
index 47a5f55..715a6da 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketch.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-
/**
* This UDF is to get the parameter K from a given sketch.
* This can be useful for debugging a work flow to make sure that resulting sketches
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketch.java
index 44bd541..cd2089b 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketch.java
@@ -22,14 +22,13 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-
/**
* This UDF is to get the parameter K from a given sketch.
* This can be useful for debugging a work flow to make sure that resulting sketches
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketch.java
index 75180ad..5a7d712 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketch.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-
/**
* This UDF is to get an approximation to the Probability Mass Function (PMF) of the input stream
* given a sketch and a set of split points - an array of <i>m</i> unique, monotonically increasing
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketch.java
index 68fbff8..87eb845 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketch.java
@@ -22,14 +22,13 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-
/**
* This UDF is to get an approximation to the Probability Mass Function (PMF) of the input stream
* given a sketch and a set of split points - an array of <i>m</i> unique, monotonically increasing
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketch.java
index 5987620..bab7eae 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketch.java
@@ -21,13 +21,12 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-
/**
* This UDF is to get a quantile value from a DoublesSketch. A single value for a
* given fraction is returned. The fraction represents a normalized rank and must be
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketch.java
index 5123551..90f496a 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketch.java
@@ -22,14 +22,13 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-
/**
* This UDF is to get a quantile value from an ItemsSketch<String>. A single value for a
* given fraction is returned. The fraction represents a normalized rank and must be
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketch.java
index 115e43f..cb7a889 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketch.java
@@ -22,14 +22,13 @@
import java.io.IOException;
import java.util.Arrays;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-
/**
* This UDF is to get a list of quantile values from an DoublesSketch given a list of
* fractions or a number of evenly spaced intervals. The fractions represent normalized ranks and
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketch.java
index 137d2d4..ad21a79 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketch.java
@@ -23,15 +23,14 @@
import java.util.Arrays;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-
/**
* This UDF is to get a list of quantile values from an ItemsSketch<String> given a list of
* fractions or a number of evenly spaced intervals. The fractions represent normalized ranks and
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/StringsSketchToString.java b/src/main/java/org/apache/datasketches/pig/quantiles/StringsSketchToString.java
index 558211f..2f52c40 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/StringsSketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/StringsSketchToString.java
@@ -22,14 +22,13 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfStringsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-
/**
* This UDF is to get a human-readable summary of a given sketch.
*/
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketch.java
index 2b39a2b..f280389 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketch.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.DoublesSketch;
+import org.apache.datasketches.quantiles.DoublesUnion;
+import org.apache.datasketches.quantiles.DoublesUnionBuilder;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -33,11 +37,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.quantiles.DoublesSketch;
-import org.apache.datasketches.quantiles.DoublesUnion;
-import org.apache.datasketches.quantiles.DoublesUnionBuilder;
-
/**
* This is a Pig UDF that merges Quantiles Sketches.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/UnionItemsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/UnionItemsSketch.java
index 4912ba7..1858ac1 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/UnionItemsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/UnionItemsSketch.java
@@ -22,6 +22,10 @@
import java.io.IOException;
import java.util.Comparator;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.quantiles.ItemsSketch;
+import org.apache.datasketches.quantiles.ItemsUnion;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -34,11 +38,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-import org.apache.datasketches.quantiles.ItemsSketch;
-import org.apache.datasketches.quantiles.ItemsUnion;
-
/**
* Computes union of ItemsSketch.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
diff --git a/src/main/java/org/apache/datasketches/pig/quantiles/UnionStringsSketch.java b/src/main/java/org/apache/datasketches/pig/quantiles/UnionStringsSketch.java
index 9e56b3d..e6e210b 100644
--- a/src/main/java/org/apache/datasketches/pig/quantiles/UnionStringsSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/quantiles/UnionStringsSketch.java
@@ -28,6 +28,7 @@
* Computes union of ItemsSketch<String>.
* It supports all three ways: exec(), Accumulator and Algebraic
*/
+@SuppressWarnings("javadoc")
public class UnionStringsSketch extends UnionItemsSketch<String> {
private static final Comparator<String> COMPARATOR = Comparator.naturalOrder();
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/ArrayOfTuplesSerDe.java b/src/main/java/org/apache/datasketches/pig/sampling/ArrayOfTuplesSerDe.java
index 0dca183..c508eba 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/ArrayOfTuplesSerDe.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/ArrayOfTuplesSerDe.java
@@ -24,14 +24,13 @@
import java.io.DataOutputStream;
import java.io.IOException;
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.memory.WritableMemory;
import org.apache.pig.data.DataReaderWriter;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.WritableByteArray;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.memory.WritableMemory;
-import org.apache.datasketches.ArrayOfItemsSerDe;
-
/**
* This <tt>ArrayOfItemsSerDe</tt> implementation takes advantage of the Pig methods used in
* Pig's own BinStorage to serialize arbitrary <tt>Tuple</tt> data.
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/DataToVarOptSketch.java b/src/main/java/org/apache/datasketches/pig/sampling/DataToVarOptSketch.java
index cc59c54..c4a36e4 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/DataToVarOptSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/DataToVarOptSketch.java
@@ -24,6 +24,7 @@
import java.io.IOException;
+import org.apache.datasketches.sampling.VarOptItemsSketch;
import org.apache.pig.AccumulatorEvalFunc;
import org.apache.pig.Algebraic;
import org.apache.pig.data.DataBag;
@@ -33,8 +34,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.sampling.VarOptItemsSketch;
-
/**
* Creates a binary version of a VarOpt sampling over input tuples. The resulting
* <tt>DataByteArray</tt> can be read in Pig with <tt>GetVarOptSamples</tt>, although the
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/GetVarOptSamples.java b/src/main/java/org/apache/datasketches/pig/sampling/GetVarOptSamples.java
index 8b2c46c..6493d9e 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/GetVarOptSamples.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/GetVarOptSamples.java
@@ -25,6 +25,8 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.sampling.VarOptItemsSketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
@@ -33,9 +35,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.sampling.VarOptItemsSketch;
-
/**
* This UDF extracts samples from the binary image of a VarOpt<Tuple> sketch. Because the
* input is a binary object, this UDF is unable to automatically determine the data schema at query
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/ReservoirSampling.java b/src/main/java/org/apache/datasketches/pig/sampling/ReservoirSampling.java
index 59acfad..82ccc3b 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/ReservoirSampling.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/ReservoirSampling.java
@@ -23,6 +23,9 @@
import java.util.ArrayList;
import java.util.List;
+import org.apache.datasketches.sampling.ReservoirItemsSketch;
+import org.apache.datasketches.sampling.ReservoirItemsUnion;
+import org.apache.datasketches.sampling.SamplingPigUtil;
import org.apache.pig.AccumulatorEvalFunc;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -35,16 +38,13 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.sampling.ReservoirItemsSketch;
-import org.apache.datasketches.sampling.ReservoirItemsUnion;
-import org.apache.datasketches.sampling.SamplingPigUtil;
-
/**
* This is a Pig UDF that applies reservoir sampling to input tuples. It implements both
* the <tt>Accumulator</tt> and <tt>Algebraic</tt> interfaces for efficient performance.
*
* @author Jon Malkin
*/
+@SuppressWarnings("javadoc")
public class ReservoirSampling extends AccumulatorEvalFunc<Tuple> implements Algebraic {
// defined for test consistency
static final String N_ALIAS = "n";
@@ -73,7 +73,7 @@
@Override
public Tuple exec(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return null;
}
@@ -88,7 +88,7 @@
@Override
public void accumulate(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return;
}
@@ -122,12 +122,12 @@
@Override
public Schema outputSchema(final Schema input) {
- if (input != null && input.size() > 0) {
+ if ((input != null) && (input.size() > 0)) {
try {
Schema source = input;
// if we have a bag, grab one level down to get a tuple
- if (source.size() == 1 && source.getField(0).type == DataType.BAG) {
+ if ((source.size() == 1) && (source.getField(0).type == DataType.BAG)) {
source = source.getField(0).schema;
}
@@ -200,7 +200,7 @@
@Override
public Tuple exec(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return null;
}
@@ -253,7 +253,7 @@
@Override
public Tuple exec(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return null;
}
@@ -264,7 +264,7 @@
final long n = (long) reservoir.get(0);
final int k = (int) reservoir.get(1);
- if (n <= k && k <= targetK_) {
+ if ((n <= k) && (k <= targetK_)) {
for (Tuple t : (DataBag) reservoir.get(2)) {
union.update(t);
}
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/ReservoirUnion.java b/src/main/java/org/apache/datasketches/pig/sampling/ReservoirUnion.java
index e905a8a..0fe52b4 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/ReservoirUnion.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/ReservoirUnion.java
@@ -23,6 +23,9 @@
import java.util.ArrayList;
import java.util.List;
+import org.apache.datasketches.sampling.ReservoirItemsSketch;
+import org.apache.datasketches.sampling.ReservoirItemsUnion;
+import org.apache.datasketches.sampling.SamplingPigUtil;
import org.apache.pig.AccumulatorEvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.BagFactory;
@@ -32,10 +35,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.sampling.ReservoirItemsSketch;
-import org.apache.datasketches.sampling.ReservoirItemsUnion;
-import org.apache.datasketches.sampling.SamplingPigUtil;
-
/**
* This is a Pig UDF that unions reservoir samples. It implements
* the <tt>Accumulator</tt> interface for more efficient performance. Input is
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/VarOptCommonImpl.java b/src/main/java/org/apache/datasketches/pig/sampling/VarOptCommonImpl.java
index 4667987..75992fc 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/VarOptCommonImpl.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/VarOptCommonImpl.java
@@ -21,6 +21,10 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.sampling.VarOptItemsSamples;
+import org.apache.datasketches.sampling.VarOptItemsSketch;
+import org.apache.datasketches.sampling.VarOptItemsUnion;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.BagFactory;
@@ -29,11 +33,6 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.sampling.VarOptItemsSamples;
-import org.apache.datasketches.sampling.VarOptItemsSketch;
-import org.apache.datasketches.sampling.VarOptItemsUnion;
-
/**
* A collection of methods and constants used across VarOpt UDFs.
*
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/VarOptSampling.java b/src/main/java/org/apache/datasketches/pig/sampling/VarOptSampling.java
index b6f4af2..85ed71a 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/VarOptSampling.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/VarOptSampling.java
@@ -28,6 +28,8 @@
import java.io.IOException;
+import org.apache.datasketches.sampling.VarOptItemsSketch;
+import org.apache.datasketches.sampling.VarOptItemsUnion;
import org.apache.pig.AccumulatorEvalFunc;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -37,15 +39,13 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.sampling.VarOptItemsSketch;
-import org.apache.datasketches.sampling.VarOptItemsUnion;
-
/**
* Applies VarOpt sampling to input tuples. Implements both the <tt>Accumulator</tt> and
* <tt>Algebraic</tt> interfaces for efficient performance.
*
* @author Jon Malkin
*/
+@SuppressWarnings("javadoc")
public class VarOptSampling extends AccumulatorEvalFunc<DataBag> implements Algebraic {
private final int targetK_;
private final int weightIdx_;
@@ -91,7 +91,7 @@
@Override
public void accumulate(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return;
}
@@ -124,7 +124,7 @@
@Override
public Schema outputSchema(final Schema input) {
try {
- if (input == null || input.size() == 0) {
+ if ((input == null) || (input.size() == 0)) {
throw new IllegalArgumentException("Degenerate input schema to VarOptSampling");
}
@@ -136,8 +136,8 @@
final Schema record = input.getField(0).schema; // record has a tuple in field 0
final Schema fields = record.getField(0).schema;
- if (fields.getField(weightIdx_).type != DataType.DOUBLE
- && fields.getField(weightIdx_).type != DataType.FLOAT) {
+ if ((fields.getField(weightIdx_).type != DataType.DOUBLE)
+ && (fields.getField(weightIdx_).type != DataType.FLOAT)) {
throw new IllegalArgumentException("weightIndex item of VarOpt tuple must be a "
+ "weight (double/float), found " + fields.getField(0).type
+ ": " + fields.toString());
@@ -214,7 +214,7 @@
@Override
public DataBag exec(final Tuple inputTuple) throws IOException {
- if (inputTuple == null || inputTuple.size() < 1 || inputTuple.isNull(0)) {
+ if ((inputTuple == null) || (inputTuple.size() < 1) || inputTuple.isNull(0)) {
return null;
}
diff --git a/src/main/java/org/apache/datasketches/pig/sampling/VarOptUnion.java b/src/main/java/org/apache/datasketches/pig/sampling/VarOptUnion.java
index 9f4c37e..108f9b0 100644
--- a/src/main/java/org/apache/datasketches/pig/sampling/VarOptUnion.java
+++ b/src/main/java/org/apache/datasketches/pig/sampling/VarOptUnion.java
@@ -23,6 +23,8 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.sampling.VarOptItemsUnion;
import org.apache.pig.AccumulatorEvalFunc;
import org.apache.pig.Algebraic;
import org.apache.pig.data.DataBag;
@@ -31,9 +33,6 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.sampling.VarOptItemsUnion;
-
/**
* Accepts binary VarOpt sketch images and unions them into a single binary output sketch.
* Due to using opaque binary objects, schema information is unavailable.
diff --git a/src/main/java/org/apache/datasketches/pig/theta/AexcludeB.java b/src/main/java/org/apache/datasketches/pig/theta/AexcludeB.java
index 01f77b8..5619d62 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/AexcludeB.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/AexcludeB.java
@@ -25,6 +25,11 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.AnotB;
+import org.apache.datasketches.theta.CompactSketch;
+import org.apache.datasketches.theta.SetOperation;
+import org.apache.datasketches.theta.Sketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataByteArray;
@@ -33,20 +38,12 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.theta.AnotB;
-import org.apache.datasketches.theta.CompactSketch;
-import org.apache.datasketches.theta.SetOperation;
-import org.apache.datasketches.theta.Sketch;
-
/**
* This is a Pig UDF that performs the A-NOT-B Set Operation on two given Sketches. Because this
* operation is fundamentally asymmetric, it is structured as a single stateless operation rather
* than stateful as are Union and Intersection UDFs, which can be iterative.
* The requirement to perform iterative A\B\C\... is rare. If needed, it can be rendered easily by
* the caller.
- *
- * @author Lee Rhodes
*/
public class AexcludeB extends EvalFunc<Tuple> {
private final long seed_;
@@ -78,7 +75,7 @@
*/
public AexcludeB(final long seed) {
super();
- this.seed_ = seed;
+ seed_ = seed;
}
// @formatter:off
diff --git a/src/main/java/org/apache/datasketches/pig/theta/DataToSketch.java b/src/main/java/org/apache/datasketches/pig/theta/DataToSketch.java
index 4a35642..434ed1c 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/DataToSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/DataToSketch.java
@@ -32,6 +32,11 @@
import java.io.IOException;
+import org.apache.datasketches.Util;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.CompactSketch;
+import org.apache.datasketches.theta.SetOperation;
+import org.apache.datasketches.theta.Union;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -42,17 +47,9 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.Util;
-import org.apache.datasketches.theta.CompactSketch;
-import org.apache.datasketches.theta.SetOperation;
-import org.apache.datasketches.theta.Union;
-
/**
* This is a Pig UDF that builds Sketches from data.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
- *
- * @author Lee Rhodes
*/
public class DataToSketch extends EvalFunc<Tuple> implements Accumulator<Tuple>, Algebraic {
//With the single exception of the Accumulator interface, UDFs are stateless.
@@ -125,10 +122,10 @@
*/
public DataToSketch(final int nomEntries, final float p, final long seed) {
super();
- this.nomEntries_ = nomEntries;
- this.p_ = p;
- this.seed_ = seed;
- this.emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ nomEntries_ = nomEntries;
+ p_ = p;
+ seed_ = seed;
+ emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
//Catch these errors during construction, don't wait for the exec to be called.
checkIfPowerOf2(nomEntries, "nomEntries");
checkProbability(p, "p");
@@ -510,10 +507,10 @@
* @param seed <a href="{@docRoot}/resources/dictionary.html#seed">See Update Hash Seed</a>.
*/
public IntermediateFinal(final int nomEntries, final float p, final long seed) {
- this.myNomEntries_ = nomEntries;
- this.myP_ = p;
- this.mySeed_ = seed;
- this.myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ myNomEntries_ = nomEntries;
+ myP_ = p;
+ mySeed_ = seed;
+ myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
}
@Override //IntermediateFinal exec
diff --git a/src/main/java/org/apache/datasketches/pig/theta/ErrorBounds.java b/src/main/java/org/apache/datasketches/pig/theta/ErrorBounds.java
index 0d15a17..4e4225c 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/ErrorBounds.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/ErrorBounds.java
@@ -23,6 +23,8 @@
import java.io.IOException;
+import org.apache.datasketches.Util;
+import org.apache.datasketches.theta.Sketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
@@ -30,9 +32,6 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.Util;
-import org.apache.datasketches.theta.Sketch;
-
//@formatter:off
/**
* This is a User Defined Function (UDF) for returning the Double value result along with a lower and
@@ -50,8 +49,6 @@
* </ul>
* </li>
* </ul>
- *
- * @author Lee Rhodes
*/
//@formatter:on
public class ErrorBounds extends EvalFunc<Tuple> {
diff --git a/src/main/java/org/apache/datasketches/pig/theta/Estimate.java b/src/main/java/org/apache/datasketches/pig/theta/Estimate.java
index 415ec0e..38c5631 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/Estimate.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/Estimate.java
@@ -23,16 +23,13 @@
import java.io.IOException;
+import org.apache.datasketches.Util;
+import org.apache.datasketches.theta.Sketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.Util;
-import org.apache.datasketches.theta.Sketch;
-
/**
* Returns the unique count estimate of a sketch as a Double.
- *
- * @author LeeRhodes
*/
public class Estimate extends EvalFunc<Double> {
private final long seed_;
diff --git a/src/main/java/org/apache/datasketches/pig/theta/Intersect.java b/src/main/java/org/apache/datasketches/pig/theta/Intersect.java
index 3ef2887..424f7f0 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/Intersect.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/Intersect.java
@@ -28,6 +28,11 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.CompactSketch;
+import org.apache.datasketches.theta.Intersection;
+import org.apache.datasketches.theta.SetOperation;
+import org.apache.datasketches.theta.Sketch;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -38,17 +43,9 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.theta.CompactSketch;
-import org.apache.datasketches.theta.Intersection;
-import org.apache.datasketches.theta.SetOperation;
-import org.apache.datasketches.theta.Sketch;
-
/**
* This is a Pig UDF that performs the Intersection Set Operation on Sketches.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
- *
- * @author Lee Rhodes
*/
public class Intersect extends EvalFunc<Tuple> implements Accumulator<Tuple>, Algebraic {
//With the single exception of the Accumulator interface, UDFs are stateless.
@@ -85,8 +82,8 @@
*/
public Intersect(final long seed) {
super();
- this.seed_ = seed;
- this.emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ seed_ = seed;
+ emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
}
//@formatter:off
@@ -343,8 +340,8 @@
* @param seed <a href="{@docRoot}/resources/dictionary.html#seed">See Update Hash Seed</a>.
*/
public IntermediateFinal(final long seed) {
- this.mySeed_ = seed;
- this.myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ mySeed_ = seed;
+ myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
}
@Override //IntermediateFinal exec
diff --git a/src/main/java/org/apache/datasketches/pig/theta/JaccardSimilarity.java b/src/main/java/org/apache/datasketches/pig/theta/JaccardSimilarity.java
index 2d39169..6d90349 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/JaccardSimilarity.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/JaccardSimilarity.java
@@ -24,15 +24,14 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.Sketch;
import org.apache.pig.EvalFunc;
import org.apache.pig.backend.executionengine.ExecException;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.theta.Sketch;
-
/**
* This is a Pig UDF that performs the JaccardSimilarity Operation on two given
* Sketches.
diff --git a/src/main/java/org/apache/datasketches/pig/theta/PigUtil.java b/src/main/java/org/apache/datasketches/pig/theta/PigUtil.java
index 98f9e19..b7f1257 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/PigUtil.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/PigUtil.java
@@ -21,21 +21,18 @@
import java.io.IOException;
+import org.apache.datasketches.ResizeFactor;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.CompactSketch;
+import org.apache.datasketches.theta.Sketch;
+import org.apache.datasketches.theta.UpdateSketch;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.ResizeFactor;
-import org.apache.datasketches.theta.CompactSketch;
-import org.apache.datasketches.theta.Sketch;
-import org.apache.datasketches.theta.UpdateSketch;
-
/**
* Common methods for the pig classes.
- *
- * @author Lee Rhodes
*/
class PigUtil {
static final ResizeFactor RF = ResizeFactor.X8;
diff --git a/src/main/java/org/apache/datasketches/pig/theta/SketchToString.java b/src/main/java/org/apache/datasketches/pig/theta/SketchToString.java
index 5f67b07..f49493f 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/SketchToString.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/SketchToString.java
@@ -23,11 +23,10 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.Util;
import org.apache.datasketches.theta.Sketch;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.Tuple;
/**
* This is a User Defined Function (UDF) for "pretty printing" the summary of a sketch
@@ -36,7 +35,6 @@
* <p>
* Refer to {@link DataToSketch#exec(Tuple)} for the definition of a Sketch Tuple.
* </p>
- * @author Lee Rhodes
*/
public class SketchToString extends EvalFunc<String> {
private boolean detailOut = false;
diff --git a/src/main/java/org/apache/datasketches/pig/theta/Union.java b/src/main/java/org/apache/datasketches/pig/theta/Union.java
index fff4bb2..3444430 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/Union.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/Union.java
@@ -32,6 +32,10 @@
import java.io.IOException;
+import org.apache.datasketches.Util;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.theta.CompactSketch;
+import org.apache.datasketches.theta.SetOperation;
import org.apache.pig.Accumulator;
import org.apache.pig.Algebraic;
import org.apache.pig.EvalFunc;
@@ -42,16 +46,9 @@
import org.apache.pig.impl.logicalLayer.FrontendException;
import org.apache.pig.impl.logicalLayer.schema.Schema;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.Util;
-import org.apache.datasketches.theta.CompactSketch;
-import org.apache.datasketches.theta.SetOperation;
-
/**
* This is a Pig UDF that performs the Union Set Operation on Sketches.
* To assist Pig, this class implements both the <i>Accumulator</i> and <i>Algebraic</i> interfaces.
- *
- * @author Lee Rhodes
*/
public class Union extends EvalFunc<Tuple> implements Accumulator<Tuple>, Algebraic {
//With the single exception of the Accumulator interface, UDFs are stateless.
@@ -130,10 +127,10 @@
*/
public Union(final int nomEntries, final float p, final long seed) {
super();
- this.nomEntries_ = nomEntries;
- this.p_ = p;
- this.seed_ = seed;
- this.emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ nomEntries_ = nomEntries;
+ p_ = p;
+ seed_ = seed;
+ emptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
//Catch these errors during construction, don't wait for the exec to be called.
checkIfPowerOf2(nomEntries, "nomEntries");
checkProbability(p, "p");
@@ -463,10 +460,10 @@
* @param seed <a href="{@docRoot}/resources/dictionary.html#seed">See Update Hash Seed</a>.
*/
public IntermediateFinal(final int nomEntries, final float p, final long seed) {
- this.myNomEntries_ = nomEntries;
- this.myP_ = p;
- this.mySeed_ = seed;
- this.myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
+ myNomEntries_ = nomEntries;
+ myP_ = p;
+ mySeed_ = seed;
+ myEmptyCompactOrderedSketchTuple_ = emptySketchTuple(seed);
}
@Override //IntermediateFinal exec
diff --git a/src/main/java/org/apache/datasketches/pig/theta/package-info.java b/src/main/java/org/apache/datasketches/pig/theta/package-info.java
index 2d9e79a..cd651d7 100644
--- a/src/main/java/org/apache/datasketches/pig/theta/package-info.java
+++ b/src/main/java/org/apache/datasketches/pig/theta/package-info.java
@@ -19,7 +19,5 @@
/**
* Pig UDFs for Theta sketch.
- *
- * @author Lee Rhodes
*/
package org.apache.datasketches.pig.theta;
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchStats.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchStats.java
index db14356..1a43423 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchStats.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchStats.java
@@ -20,7 +20,6 @@
package org.apache.datasketches.pig.tuple;
import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
-
import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesSketchIterator;
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBounds.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBounds.java
index 5f44ee9..000bdd8 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBounds.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBounds.java
@@ -22,15 +22,14 @@
import java.io.IOException;
import java.util.Arrays;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-
/**
* This is a User Defined Function (UDF) for obtaining the unique count estimate
* along with a lower and upper bound from an ArrayOfDoublesSketch.
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimates.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimates.java
index 1914e1d..ea5a410 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimates.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimates.java
@@ -21,14 +21,13 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesSketchIterator;
import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* This UDF converts an ArrayOfDoubles sketch to estimates.
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeans.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeans.java
index 608c8e7..ce624ba 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeans.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeans.java
@@ -22,15 +22,14 @@
import java.io.IOException;
import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-
/**
* This UDF converts an ArrayOfDoubles sketch to mean values.
* The result will be a tuple with N double values, where
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntries.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntries.java
index cff7937..0b2d506 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntries.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntries.java
@@ -21,13 +21,12 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* This is a User Defined Function (UDF) for obtaining the number of retained entries
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketch.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketch.java
index a210f87..c6b879b 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketch.java
@@ -21,10 +21,6 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.datasketches.quantiles.DoublesSketchBuilder;
@@ -32,6 +28,9 @@
import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesSketchIterator;
import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* This UDF converts a given column of double values from an ArrayOfDoubles sketch
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariances.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariances.java
index ff6fb94..7a4d28a 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariances.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariances.java
@@ -22,15 +22,14 @@
import java.io.IOException;
import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-
/**
* This UDF converts an ArrayOfDoubles sketch to variance values.
* The result will be a tuple with N double values, where
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimates.java b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimates.java
index e488f5c..1e68e08 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimates.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimates.java
@@ -23,16 +23,14 @@
import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
import org.apache.commons.math3.stat.inference.TTest;
-
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketch;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-
/**
* Calculate p-values given two ArrayOfDoublesSketch. Each value in the sketch
* is treated as a separate metric measurement, and a p-value will be generated
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketch.java
index d05b95e..a2156eb 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketch.java
@@ -25,6 +25,7 @@
* This UDF creates an ArrayOfDoublesSketch from raw data.
* It supports all three ways: exec(), Accumulator and Algebraic.
*/
+@SuppressWarnings("javadoc")
public class DataToArrayOfDoublesSketch extends DataToArrayOfDoublesSketchBase implements Algebraic {
/**
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchAlgebraicIntermediateFinal.java
index 7da7119..0d2c340 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchAlgebraicIntermediateFinal.java
@@ -23,18 +23,17 @@
import java.io.IOException;
-import org.apache.log4j.Logger;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataBag;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.tuple.ArrayOfDoublesSetOperationBuilder;
import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
import org.apache.datasketches.tuple.ArrayOfDoublesUnion;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+import org.apache.log4j.Logger;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataBag;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* Class used to calculate the intermediate pass (combiner) or the final pass
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchBase.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchBase.java
index 93ce4b1..7461f03 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchBase.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchBase.java
@@ -23,6 +23,8 @@
import java.io.IOException;
+import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
+import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -32,9 +34,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
-import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
-
abstract class DataToArrayOfDoublesSketchBase extends EvalFunc<Tuple> implements Accumulator<Tuple> {
private final int sketchSize_;
private final float samplingProbability_;
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketch.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketch.java
index 18ea21c..8ee0fbc 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketch.java
@@ -19,23 +19,23 @@
package org.apache.datasketches.pig.tuple;
-import org.apache.pig.Algebraic;
-
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummaryFactory;
import org.apache.datasketches.tuple.adouble.DoubleSummarySetOperations;
+import org.apache.pig.Algebraic;
/**
* This UDF creates a Sketch<DoubleSummary> from raw data.
* It supports all three ways: exec(), Accumulator and Algebraic.
*/
+@SuppressWarnings("javadoc")
public class DataToDoubleSummarySketch extends DataToSketch<Double, DoubleSummary> implements Algebraic {
/**
* Constructor with default sketch size and default mode (sum)
*/
public DataToDoubleSummarySketch() {
- super(new DoubleSummaryFactory());
+ super(new DoubleSummaryFactory(DoubleSummary.Mode.Sum));
}
/**
@@ -43,7 +43,7 @@
* @param sketchSize String representation of sketch size
*/
public DataToDoubleSummarySketch(final String sketchSize) {
- super(Integer.parseInt(sketchSize), new DoubleSummaryFactory());
+ super(Integer.parseInt(sketchSize), new DoubleSummaryFactory(DoubleSummary.Mode.Sum));
}
/**
@@ -102,7 +102,8 @@
* Default sketch size and default mode
*/
public IntermediateFinal() {
- super(new DoubleSummaryFactory(), new DoubleSummarySetOperations(), new DoubleSummaryDeserializer());
+ super(new DoubleSummaryFactory(DoubleSummary.Mode.Sum),
+ new DoubleSummarySetOperations(DoubleSummary.Mode.Sum), new DoubleSummaryDeserializer());
}
/**
@@ -111,7 +112,8 @@
* @param sketchSize String representation of sketch size
*/
public IntermediateFinal(final String sketchSize) {
- super(Integer.parseInt(sketchSize), new DoubleSummaryFactory(), new DoubleSummarySetOperations(),
+ super(Integer.parseInt(sketchSize), new DoubleSummaryFactory(DoubleSummary.Mode.Sum),
+ new DoubleSummarySetOperations(DoubleSummary.Mode.Sum),
new DoubleSummaryDeserializer());
}
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToSketch.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToSketch.java
index 1e1fad4..0d26398 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToSketch.java
@@ -23,6 +23,10 @@
import java.io.IOException;
+import org.apache.datasketches.tuple.SummaryFactory;
+import org.apache.datasketches.tuple.UpdatableSketch;
+import org.apache.datasketches.tuple.UpdatableSketchBuilder;
+import org.apache.datasketches.tuple.UpdatableSummary;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -32,11 +36,6 @@
import org.apache.pig.data.DataType;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.tuple.SummaryFactory;
-import org.apache.datasketches.tuple.UpdatableSketch;
-import org.apache.datasketches.tuple.UpdatableSketchBuilder;
-import org.apache.datasketches.tuple.UpdatableSummary;
-
/**
* This is a generic implementation to be specialized in concrete UDFs
* @param <U> Update type
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DataToSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/tuple/DataToSketchAlgebraicIntermediateFinal.java
index 3425def..749cd74 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DataToSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DataToSketchAlgebraicIntermediateFinal.java
@@ -23,12 +23,6 @@
import java.io.IOException;
-import org.apache.log4j.Logger;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataBag;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.tuple.Sketch;
import org.apache.datasketches.tuple.SummaryDeserializer;
import org.apache.datasketches.tuple.SummaryFactory;
@@ -37,6 +31,11 @@
import org.apache.datasketches.tuple.UpdatableSketch;
import org.apache.datasketches.tuple.UpdatableSketchBuilder;
import org.apache.datasketches.tuple.UpdatableSummary;
+import org.apache.log4j.Logger;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataBag;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* Class used to calculate the intermediate pass (combiner) or the final pass
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimates.java b/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimates.java
index efe56fb..c8940e7 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimates.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimates.java
@@ -21,11 +21,6 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-import org.apache.pig.data.TupleFactory;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.tuple.Sketch;
import org.apache.datasketches.tuple.SketchIterator;
@@ -33,6 +28,10 @@
import org.apache.datasketches.tuple.SummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
+import org.apache.pig.data.TupleFactory;
/**
* This UDF converts a Sketch<DoubleSummary> to estimates.
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentile.java b/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentile.java
index 1e15e15..67ea162 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentile.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentile.java
@@ -21,10 +21,6 @@
import java.io.IOException;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.datasketches.quantiles.UpdateDoublesSketch;
@@ -34,6 +30,9 @@
import org.apache.datasketches.tuple.SummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* This UDF is to get a percentile value from a Sketch<DoubleSummary>.
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketch.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketch.java
index 6ca24a8..91e3a56 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketch.java
@@ -25,6 +25,7 @@
* This is to union ArrayOfDoublesSketches.
* It supports all three ways: exec(), Accumulator and Algebraic
*/
+@SuppressWarnings("javadoc")
public class UnionArrayOfDoublesSketch extends UnionArrayOfDoublesSketchBase implements Algebraic {
/**
* Constructor with default sketch size and default number of values of 1.
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchAlgebraicIntermediateFinal.java
index 18505a5..20df310 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchAlgebraicIntermediateFinal.java
@@ -23,17 +23,16 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSetOperationBuilder;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
+import org.apache.datasketches.tuple.ArrayOfDoublesUnion;
import org.apache.log4j.Logger;
import org.apache.pig.EvalFunc;
import org.apache.pig.data.DataBag;
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSetOperationBuilder;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-import org.apache.datasketches.tuple.ArrayOfDoublesUnion;
-
/**
* This is to calculate the intermediate pass (combiner) or the final pass
* (reducer) of an Algebraic sketch operation. This may be called multiple times
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchBase.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchBase.java
index 4480028..2769b4f 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchBase.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchBase.java
@@ -23,6 +23,11 @@
import java.io.IOException;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.tuple.ArrayOfDoublesSetOperationBuilder;
+import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
+import org.apache.datasketches.tuple.ArrayOfDoublesUnion;
+import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -31,12 +36,6 @@
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.memory.Memory;
-import org.apache.datasketches.tuple.ArrayOfDoublesSetOperationBuilder;
-import org.apache.datasketches.tuple.ArrayOfDoublesSketches;
-import org.apache.datasketches.tuple.ArrayOfDoublesUnion;
-import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
-
abstract class UnionArrayOfDoublesSketchBase extends EvalFunc<Tuple> implements Accumulator<Tuple> {
private final int sketchSize_;
private final int numValues_;
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketch.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketch.java
index 21c81c6..61b03c5 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketch.java
@@ -19,23 +19,24 @@
package org.apache.datasketches.pig.tuple;
-import org.apache.pig.Algebraic;
-
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummarySetOperations;
+import org.apache.pig.Algebraic;
/**
* This is to union Sketch<DoubleSummary>.
* It supports all three ways: exec(), Accumulator and Algebraic
*/
+@SuppressWarnings("javadoc")
public class UnionDoubleSummarySketch extends UnionSketch<DoubleSummary> implements Algebraic {
/**
* Constructor with default sketch size and default mode (sum)
*/
public UnionDoubleSummarySketch() {
- super(new DoubleSummarySetOperations(), new DoubleSummaryDeserializer());
+ super(new DoubleSummarySetOperations(DoubleSummary.Mode.Sum),
+ new DoubleSummaryDeserializer());
}
/**
@@ -43,7 +44,8 @@
* @param sketchSize String representation of sketch size
*/
public UnionDoubleSummarySketch(final String sketchSize) {
- super(Integer.parseInt(sketchSize), new DoubleSummarySetOperations(), new DoubleSummaryDeserializer());
+ super(Integer.parseInt(sketchSize),
+ new DoubleSummarySetOperations(DoubleSummary.Mode.Sum), new DoubleSummaryDeserializer());
}
/**
@@ -101,7 +103,7 @@
* Default sketch size and default mode.
*/
public IntermediateFinal() {
- super(new DoubleSummarySetOperations(), new DoubleSummaryDeserializer());
+ super(new DoubleSummarySetOperations(DoubleSummary.Mode.Sum), new DoubleSummaryDeserializer());
}
/**
@@ -110,7 +112,8 @@
* @param sketchSize String representation of sketch size
*/
public IntermediateFinal(final String sketchSize) {
- super(Integer.parseInt(sketchSize), new DoubleSummarySetOperations(), new DoubleSummaryDeserializer());
+ super(Integer.parseInt(sketchSize),
+ new DoubleSummarySetOperations(DoubleSummary.Mode.Sum), new DoubleSummaryDeserializer());
}
/**
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionSketch.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionSketch.java
index 5c75ef3..1bba6fe 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionSketch.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionSketch.java
@@ -23,6 +23,12 @@
import java.io.IOException;
+import org.apache.datasketches.tuple.Sketch;
+import org.apache.datasketches.tuple.Sketches;
+import org.apache.datasketches.tuple.Summary;
+import org.apache.datasketches.tuple.SummaryDeserializer;
+import org.apache.datasketches.tuple.SummarySetOperations;
+import org.apache.datasketches.tuple.Union;
import org.apache.log4j.Logger;
import org.apache.pig.Accumulator;
import org.apache.pig.EvalFunc;
@@ -31,13 +37,6 @@
import org.apache.pig.data.DataByteArray;
import org.apache.pig.data.Tuple;
-import org.apache.datasketches.tuple.Sketch;
-import org.apache.datasketches.tuple.Sketches;
-import org.apache.datasketches.tuple.Summary;
-import org.apache.datasketches.tuple.SummaryDeserializer;
-import org.apache.datasketches.tuple.SummarySetOperations;
-import org.apache.datasketches.tuple.Union;
-
/**
* This is a generic implementation to be specialized in concrete UDFs
* @param <S> Summary type
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/UnionSketchAlgebraicIntermediateFinal.java b/src/main/java/org/apache/datasketches/pig/tuple/UnionSketchAlgebraicIntermediateFinal.java
index 2ea5ead..655dd71 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/UnionSketchAlgebraicIntermediateFinal.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/UnionSketchAlgebraicIntermediateFinal.java
@@ -23,17 +23,16 @@
import java.io.IOException;
-import org.apache.log4j.Logger;
-import org.apache.pig.EvalFunc;
-import org.apache.pig.data.DataBag;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-
import org.apache.datasketches.tuple.Sketch;
import org.apache.datasketches.tuple.Summary;
import org.apache.datasketches.tuple.SummaryDeserializer;
import org.apache.datasketches.tuple.SummarySetOperations;
import org.apache.datasketches.tuple.Union;
+import org.apache.log4j.Logger;
+import org.apache.pig.EvalFunc;
+import org.apache.pig.data.DataBag;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
/**
* This is to calculate the intermediate pass (combiner) or the final pass
diff --git a/src/main/java/org/apache/datasketches/pig/tuple/Util.java b/src/main/java/org/apache/datasketches/pig/tuple/Util.java
index 833275c..0717ce1 100644
--- a/src/main/java/org/apache/datasketches/pig/tuple/Util.java
+++ b/src/main/java/org/apache/datasketches/pig/tuple/Util.java
@@ -19,16 +19,15 @@
package org.apache.datasketches.pig.tuple;
-import org.apache.pig.backend.executionengine.ExecException;
-import org.apache.pig.data.DataByteArray;
-import org.apache.pig.data.Tuple;
-import org.apache.pig.data.TupleFactory;
-
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.tuple.Sketch;
import org.apache.datasketches.tuple.Sketches;
import org.apache.datasketches.tuple.Summary;
import org.apache.datasketches.tuple.SummaryDeserializer;
+import org.apache.pig.backend.executionengine.ExecException;
+import org.apache.pig.data.DataByteArray;
+import org.apache.pig.data.Tuple;
+import org.apache.pig.data.TupleFactory;
final class Util {
diff --git a/src/main/java/org/apache/datasketches/sampling/SamplingPigUtil.java b/src/main/java/org/apache/datasketches/sampling/SamplingPigUtil.java
index 59816e9..3f8e856 100644
--- a/src/main/java/org/apache/datasketches/sampling/SamplingPigUtil.java
+++ b/src/main/java/org/apache/datasketches/sampling/SamplingPigUtil.java
@@ -24,8 +24,11 @@
/**
* @author Jon Malkin
*/
+@SuppressWarnings("javadoc")
public final class SamplingPigUtil {
+
public static <T> ArrayList<T> getRawSamplesAsList(final ReservoirItemsSketch<T> sketch) {
return sketch.getRawSamplesAsList();
}
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/PigTestingUtil.java b/src/test/java/org/apache/datasketches/pig/PigTestingUtil.java
index 956c485..5a3af0a 100644
--- a/src/test/java/org/apache/datasketches/pig/PigTestingUtil.java
+++ b/src/test/java/org/apache/datasketches/pig/PigTestingUtil.java
@@ -27,16 +27,14 @@
import org.apache.datasketches.Family;
import org.apache.datasketches.theta.UpdateSketch;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class PigTestingUtil {
public static final String LS = System.getProperty("line.separator");
-
-
+
+
/**
* Returns a tuple constructed from the given array of objects.
- *
+ *
* @param in Array of objects.
* @throws ExecException this is thrown by Pig
* @return tuple
@@ -49,11 +47,11 @@
}
return tuple;
}
-
+
/**
* Returns a Pig DataByteArray constructed from a QuickSelectSketch.
- *
- * @param nomSize of the Sketch. Note, minimum size is 16.
+ *
+ * @param nomSize of the Sketch. Note, minimum size is 16.
* Cache size will autoscale from a minimum of 16.
* @param start start value
* @param numValues number of values in the range
@@ -67,11 +65,11 @@
byte[] byteArr = skA.compact(true, null).toByteArray();
return new DataByteArray(byteArr);
}
-
+
/**
* Returns a Pig DataByteArray constructed from a AlphaSketch.
- *
- * @param nomSize of the Sketch. Note, minimum nominal size is 512.
+ *
+ * @param nomSize of the Sketch. Note, minimum nominal size is 512.
* Cache size will autoscale from a minimum of 512.
* @param start start value
* @param numValues number of values in the range
@@ -86,12 +84,12 @@
byte[] byteArr = skA.compact(true, null).toByteArray();
return new DataByteArray(byteArr);
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s);
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/cpc/DataToSketchTest.java b/src/test/java/org/apache/datasketches/pig/cpc/DataToSketchTest.java
index 7b85a51..b18bbf6 100644
--- a/src/test/java/org/apache/datasketches/pig/cpc/DataToSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/cpc/DataToSketchTest.java
@@ -33,6 +33,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class DataToSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -334,7 +335,7 @@
static CpcSketch getSketch(final DataByteArray dba, final long seed) throws Exception {
Assert.assertNotNull(dba);
Assert.assertTrue(dba.size() > 0);
- return CpcSketch.heapify(dba.get(), seed);
+ return CpcSketch.heapify(dba.get(), seed);
}
}
diff --git a/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBoundsTest.java b/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBoundsTest.java
index 30ce74b..66036b1 100644
--- a/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBoundsTest.java
+++ b/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateAndErrorBoundsTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class GetEstimateAndErrorBoundsTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateTest.java b/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateTest.java
index 150dcbd..e889c4e 100644
--- a/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateTest.java
+++ b/src/test/java/org/apache/datasketches/pig/cpc/GetEstimateTest.java
@@ -27,6 +27,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class GetEstimateTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/cpc/SketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/cpc/SketchToStringTest.java
index f7e4540..eb0a5fe 100644
--- a/src/test/java/org/apache/datasketches/pig/cpc/SketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/cpc/SketchToStringTest.java
@@ -27,6 +27,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class SketchToStringTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/cpc/UnionSketchTest.java b/src/test/java/org/apache/datasketches/pig/cpc/UnionSketchTest.java
index 627662a..55e8b78 100644
--- a/src/test/java/org/apache/datasketches/pig/cpc/UnionSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/cpc/UnionSketchTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.cpc.CpcSketch;
+@SuppressWarnings("javadoc")
public class UnionSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketchTest.java
index ebcaba8..63e4052 100644
--- a/src/test/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/frequencies/DataToFrequentStringsSketchTest.java
@@ -34,7 +34,9 @@
import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.datasketches.pig.tuple.PigUtil;
+@SuppressWarnings("javadoc")
public class DataToFrequentStringsSketchTest {
+
@Test
public void execNullInputTuple() throws Exception {
EvalFunc<Tuple> func = new DataToFrequentStringsSketch("8");
@@ -189,7 +191,7 @@
bag.add(PigUtil.objectsToTuple(PigUtil.tuplesToBag(PigUtil.objectsToTuple("a"))));
// this is to simulate the output from a prior call of IntermediateFinal
- ItemsSketch<String> s = new ItemsSketch<String>(8);
+ ItemsSketch<String> s = new ItemsSketch<>(8);
s.update("b", 1L);
s.update("a", 2L);
s.update("b", 3L);
diff --git a/src/test/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimatesTest.java b/src/test/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimatesTest.java
index c570026..d981e83 100644
--- a/src/test/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimatesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/frequencies/FrequentStringsSketchToEstimatesTest.java
@@ -36,7 +36,9 @@
import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.datasketches.pig.tuple.PigUtil;
+@SuppressWarnings("javadoc")
public class FrequentStringsSketchToEstimatesTest {
+
@Test
public void nullInput() throws Exception {
EvalFunc<DataBag> func = new FrequentStringsSketchToEstimates();
@@ -54,7 +56,7 @@
@Test
public void emptySketch() throws Exception {
EvalFunc<DataBag> func = new FrequentStringsSketchToEstimates();
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe())));
DataBag bag = func.exec(inputTuple);
Assert.assertNotNull(bag);
@@ -64,7 +66,7 @@
@Test
public void exact() throws Exception {
EvalFunc<DataBag> func = new FrequentStringsSketchToEstimates();
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("a");
sketch.update("b");
@@ -91,7 +93,7 @@
@Test
public void estimation() throws Exception {
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("1", 1000);
sketch.update("2", 500);
sketch.update("3", 200);
diff --git a/src/test/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketchTest.java
index f1392c9..3645aba 100644
--- a/src/test/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/frequencies/UnionFrequentStringsSketchTest.java
@@ -35,7 +35,9 @@
import org.apache.datasketches.frequencies.ItemsSketch;
import org.apache.datasketches.pig.tuple.PigUtil;
+@SuppressWarnings("javadoc")
public class UnionFrequentStringsSketchTest {
+
@Test
public void execNullInput() throws Exception {
EvalFunc<Tuple> func = new UnionFrequentStringsSketch("8");
@@ -55,13 +57,13 @@
EvalFunc<Tuple> func = new UnionFrequentStringsSketch("8");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("b");
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
}
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("b");
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
@@ -166,7 +168,7 @@
Accumulator<Tuple> func = new UnionFrequentStringsSketch("8");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
}
func.accumulate(PigUtil.objectsToTuple(bag));
@@ -185,7 +187,7 @@
Accumulator<Tuple> func = new UnionFrequentStringsSketch("8");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("b");
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
@@ -194,7 +196,7 @@
bag = BagFactory.getInstance().newDefaultBag();
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("b");
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
@@ -235,7 +237,7 @@
// this is to simulate the output from Initial
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("b");
DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
@@ -244,7 +246,7 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a", 2L);
sketch.update("b", 3L);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe()))));
@@ -269,7 +271,7 @@
// this is to simulate the output from Initial
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a", 10);
sketch.update("b");
sketch.update("c");
@@ -284,7 +286,7 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
- ItemsSketch<String> sketch = new ItemsSketch<String>(8);
+ ItemsSketch<String> sketch = new ItemsSketch<>(8);
sketch.update("a");
sketch.update("a");
sketch.update("g", 5);
@@ -310,7 +312,9 @@
// only 2 items ("a" and "g") should have counts more than 1
int count = 0;
for (ItemsSketch.Row<String> item: items) {
- if (item.getLowerBound() > 1) count++;
+ if (item.getLowerBound() > 1) {
+ count++;
+ }
}
Assert.assertEquals(count, 2);
}
diff --git a/src/test/java/org/apache/datasketches/pig/hash/MurmurHash3Test.java b/src/test/java/org/apache/datasketches/pig/hash/MurmurHash3Test.java
index 847b8b2..caf2c5f 100644
--- a/src/test/java/org/apache/datasketches/pig/hash/MurmurHash3Test.java
+++ b/src/test/java/org/apache/datasketches/pig/hash/MurmurHash3Test.java
@@ -36,15 +36,13 @@
/**
* Tests the MurmurHash3 class.
- *
- * @author Lee Rhodes
*/
-@SuppressWarnings({ "unused", "unchecked" })
+@SuppressWarnings({ "unused", "unchecked", "javadoc" })
public class MurmurHash3Test {
private static final TupleFactory mTupleFactory = TupleFactory.getInstance();
-
+
private String hashUdfName = "org.apache.datasketches.pig.hash.MurmurHash3";
-
+
@Test
public void checkExceptions1() throws IOException {
EvalFunc<Tuple> hashUdf =
@@ -55,7 +53,7 @@
out = hashUdf.exec(in);
Assert.assertNull(out);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkExceptions2() throws IOException {
EvalFunc<Tuple> hashUdf =
@@ -67,7 +65,7 @@
in.set(1, new Double(9001));
out = hashUdf.exec(in);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkExceptions3() throws IOException {
EvalFunc<Tuple> hashUdf =
@@ -78,7 +76,7 @@
in.set(0, in);
out = hashUdf.exec(in);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkExceptions4() throws IOException {
EvalFunc<Tuple> hashUdf =
@@ -91,7 +89,7 @@
in.set(2, new Long(8));
out = hashUdf.exec(in);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkExceptions5() throws IOException {
EvalFunc<Tuple> hashUdf =
@@ -104,62 +102,62 @@
in.set(2, new Integer(0));
out = hashUdf.exec(in);
}
-
+
@Test
public void check1ValidArg() throws IOException {
EvalFunc<Tuple> hashUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(hashUdfName));
Tuple in, out;
-
+
//test Integer, Long, Float, Double, DataByteArray, String
in = mTupleFactory.newTuple(1);
-
+
in.set(0, null);
out = hashUdf.exec(in);
Assert.assertNull(out.get(0));
Assert.assertNull(out.get(1));
Assert.assertNull(out.get(2));
-
+
in.set(0, new Integer(1));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new Long(1));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new Float(1));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new Double(0.0));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new Double( -0.0));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, Double.NaN);
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("1"));
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("")); //empty
out = hashUdf.exec(in);
Assert.assertNull(out.get(0));
Assert.assertNull(out.get(1));
Assert.assertNull(out.get(2));
-
+
byte[] bArr = { 1, 2, 3, 4 };
DataByteArray dba = new DataByteArray(bArr);
in.set(0, dba);
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
bArr = new byte[0]; //empty
dba = new DataByteArray(bArr);
in.set(0, dba);
@@ -168,69 +166,69 @@
Assert.assertNull(out.get(1));
Assert.assertNull(out.get(2));
}
-
+
@Test
public void check2ValidArg() throws IOException {
EvalFunc<Tuple> hashUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(hashUdfName));
Tuple in, out;
-
+
//test String, seed
in = mTupleFactory.newTuple(2);
-
+
in.set(0, new String("1"));
//2nd is null
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("1"));
in.set(1, 9001);
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("1"));
in.set(1, 9001L);
out = hashUdf.exec(in);
checkOutput(out, false);
}
-
+
@Test
public void check3ValidArg() throws IOException {
EvalFunc<Tuple> hashUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(hashUdfName));
Tuple in, out;
-
+
//test String, seed
in = mTupleFactory.newTuple(3);
-
+
in.set(0, new String("1"));
//2nd is null
//3rd is null
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("1"));
in.set(1, 9001);
//3rd is null
out = hashUdf.exec(in);
checkOutput(out, false);
-
+
in.set(0, new String("1"));
in.set(1, 9001);
in.set(2, 7);
out = hashUdf.exec(in);
checkOutput(out, true);
}
-
+
@Test
public void check3ValidArgs() throws IOException {
EvalFunc<Tuple> hashUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(hashUdfName));
Tuple in, out;
-
+
//test multiple integers, seed
in = mTupleFactory.newTuple(3);
-
+
for (int i = 0; i < 10; i++ ) {
in.set(0, i);
in.set(1, 9001);
@@ -239,7 +237,7 @@
checkOutput(out, true);
}
}
-
+
private static void checkOutput(Tuple out, boolean checkMod) throws IOException {
long h0 = (Long) out.get(0);
long h1 = (Long) out.get(1);
@@ -250,7 +248,7 @@
Assert.assertTrue(r >= 0, "" + r);
}
}
-
+
/**
* Test the outputSchema method for MurmurHash3.
* @throws IOException thrown by Pig
@@ -259,54 +257,54 @@
public void outputSchemaTestMurmurHash3Udf() throws IOException {
EvalFunc<Tuple> hashUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(hashUdfName));
-
+
Schema inputSchema = null;
-
+
Schema nullOutputSchema = null;
-
+
Schema outputSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputInnerFs0 = null;
Schema.FieldSchema outputInnerFs1 = null;
Schema.FieldSchema outputInnerFs2 = null;
-
+
nullOutputSchema = hashUdf.outputSchema(null);
-
+
//CHARARRAY is one of many different input types
inputSchema = Schema.generateNestedSchema(DataType.BAG, DataType.CHARARRAY);
-
+
outputSchema = hashUdf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
-
+
outputInnerSchema = outputOuterFs0.schema;
outputInnerFs0 = outputInnerSchema.getField(0);
outputInnerFs1 = outputInnerSchema.getField(1);
outputInnerFs2 = outputInnerSchema.getField(2);
-
+
Assert.assertNull(nullOutputSchema, "Should be null");
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "long";
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "long";
Assert.assertNotNull(outputInnerFs1, "innerSchema.getField(1) may not be null");
result = DataType.findTypeName(outputInnerFs1.type);
Assert.assertEquals(result, expected);
-
+
expected = "int";
Assert.assertNotNull(outputInnerFs2, "innerSchema.getField(2) may not be null");
result = DataType.findTypeName(outputInnerFs2.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -325,17 +323,17 @@
//@formatter:on
//end print schemas
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/hll/DataToSketchTest.java b/src/test/java/org/apache/datasketches/pig/hll/DataToSketchTest.java
index ae0dcb7..407d780 100644
--- a/src/test/java/org/apache/datasketches/pig/hll/DataToSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/hll/DataToSketchTest.java
@@ -32,6 +32,7 @@
import org.apache.datasketches.hll.HllSketch;
import org.apache.datasketches.hll.TgtHllType;
+@SuppressWarnings("javadoc")
public class DataToSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
@@ -307,7 +308,7 @@
static HllSketch getSketch(DataByteArray dba) throws Exception {
Assert.assertNotNull(dba);
Assert.assertTrue(dba.size() > 0);
- return HllSketch.heapify(dba.get());
+ return HllSketch.heapify(dba.get());
}
}
diff --git a/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBoundsTest.java b/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBoundsTest.java
index 925779d..745b7ef 100644
--- a/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBoundsTest.java
+++ b/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateAndErrorBoundsTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.hll.HllSketch;
+@SuppressWarnings("javadoc")
public class SketchToEstimateAndErrorBoundsTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateTest.java b/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateTest.java
index 4d7cca7..4c99a27 100644
--- a/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateTest.java
+++ b/src/test/java/org/apache/datasketches/pig/hll/SketchToEstimateTest.java
@@ -27,6 +27,7 @@
import org.apache.datasketches.hll.HllSketch;
+@SuppressWarnings("javadoc")
public class SketchToEstimateTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/hll/SketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/hll/SketchToStringTest.java
index f3c4987..f977be7 100644
--- a/src/test/java/org/apache/datasketches/pig/hll/SketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/hll/SketchToStringTest.java
@@ -27,6 +27,7 @@
import org.apache.datasketches.hll.HllSketch;
+@SuppressWarnings("javadoc")
public class SketchToStringTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/hll/UnionSketchTest.java b/src/test/java/org/apache/datasketches/pig/hll/UnionSketchTest.java
index 37a853d..36cc0b6 100644
--- a/src/test/java/org/apache/datasketches/pig/hll/UnionSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/hll/UnionSketchTest.java
@@ -32,6 +32,7 @@
import org.apache.datasketches.hll.HllSketch;
import org.apache.datasketches.hll.TgtHllType;
+@SuppressWarnings("javadoc")
public class UnionSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/kll/DataToSketchTest.java b/src/test/java/org/apache/datasketches/pig/kll/DataToSketchTest.java
index c8f0057..18f3115 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/DataToSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/DataToSketchTest.java
@@ -32,6 +32,7 @@
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.kll.KllFloatsSketch;
+@SuppressWarnings("javadoc")
public class DataToSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
private static final BagFactory BAG_FACTORY = BagFactory.getInstance();
@@ -78,7 +79,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 1);
}
-
+
@Test
public void execMixedNullCase() throws Exception {
final EvalFunc<DataByteArray> func = new DataToSketch();
@@ -127,7 +128,7 @@
sketch = getSketch(result);
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
-
+
// mixed null case
bag = BAG_FACTORY.newDefaultBag();
bag.add(TUPLE_FACTORY.newTuple(Float.valueOf(1)));
@@ -208,7 +209,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
}
-
+
@Test
public void algebraicIntermediateMixedNullCase() throws Exception {
@SuppressWarnings("unchecked")
@@ -288,7 +289,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
}
-
+
@Test
public void algebraicFinalMixedNullCase() throws Exception {
@SuppressWarnings("unchecked")
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetCdfTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetCdfTest.java
index 40ffd49..ab00c32 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetCdfTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetCdfTest.java
@@ -32,6 +32,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetCdfTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -47,7 +48,9 @@
public void normalCase() throws Exception {
final EvalFunc<Tuple> func = new GetCdf();
final KllFloatsSketch sketch = new KllFloatsSketch();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
final Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 2f, 7f)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetKTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetKTest.java
index 838c1d5..a3cd730 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetKTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetKTest.java
@@ -31,6 +31,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetKTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetPmfTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetPmfTest.java
index 0cca066..6d6dd4e 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetPmfTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetPmfTest.java
@@ -32,6 +32,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetPmfTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -47,7 +48,9 @@
public void normalCase() throws Exception {
final EvalFunc<Tuple> func = new GetPmf();
final KllFloatsSketch sketch = new KllFloatsSketch();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
final Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 2f, 7f)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetQuantileTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetQuantileTest.java
index cd62d4d..bde55d2 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetQuantileTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetQuantileTest.java
@@ -31,6 +31,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantileTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetQuantilesTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetQuantilesTest.java
index 4581b3d..8e1827f 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetQuantilesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetQuantilesTest.java
@@ -32,6 +32,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantilesTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -75,7 +76,9 @@
public void oneFraction() throws Exception {
final EvalFunc<Tuple> func = new GetQuantiles();
final KllFloatsSketch sketch = new KllFloatsSketch();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
final Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 0.5)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 1);
@@ -86,7 +89,9 @@
public void severalFractions() throws Exception {
final EvalFunc<Tuple> func = new GetQuantiles();
final KllFloatsSketch sketch = new KllFloatsSketch();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
final Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 0.0, 0.5, 1.0)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
@@ -99,7 +104,9 @@
public void numberOfEvenlySpacedIntervals() throws Exception {
final EvalFunc<Tuple> func = new GetQuantiles();
final KllFloatsSketch sketch = new KllFloatsSketch();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
final Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 3)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/kll/GetRankTest.java b/src/test/java/org/apache/datasketches/pig/kll/GetRankTest.java
index 681202f..88fc9d7 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/GetRankTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/GetRankTest.java
@@ -31,6 +31,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetRankTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/kll/SketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/kll/SketchToStringTest.java
index ef33132..742336e 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/SketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/SketchToStringTest.java
@@ -31,6 +31,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class SketchToStringTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/kll/UnionSketchTest.java b/src/test/java/org/apache/datasketches/pig/kll/UnionSketchTest.java
index e64bcad..b5c7448 100644
--- a/src/test/java/org/apache/datasketches/pig/kll/UnionSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/kll/UnionSketchTest.java
@@ -32,6 +32,7 @@
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.kll.KllFloatsSketch;
+@SuppressWarnings("javadoc")
public class UnionSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
private static final BagFactory BAG_FACTORY = BagFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketchTest.java
index 36669a1..b3b160d 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/DataToDoublesSketchTest.java
@@ -35,6 +35,7 @@
import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.datasketches.quantiles.UpdateDoublesSketch;
+@SuppressWarnings("javadoc")
public class DataToDoublesSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
private static final BagFactory bagFactory = BagFactory.getInstance();
@@ -81,7 +82,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 1);
}
-
+
@Test
public void execMixedNullCase() throws Exception {
EvalFunc<Tuple> func = new DataToDoublesSketch();
@@ -130,7 +131,7 @@
sketch = getSketch(resultTuple);
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
-
+
// mixed null case
bag = bagFactory.newDefaultBag();
bag.add(tupleFactory.newTuple(1.0));
@@ -199,7 +200,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
}
-
+
@Test
public void algebraicIntermediateFinalMixedNullCase() throws Exception {
EvalFunc<Tuple> func = new DataToDoublesSketch.IntermediateFinal();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/DataToStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/DataToStringsSketchTest.java
index 2b8a25c..af1fbd2 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/DataToStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/DataToStringsSketchTest.java
@@ -38,6 +38,7 @@
import org.apache.datasketches.ArrayOfStringsSerDe;
import org.apache.datasketches.quantiles.ItemsSketch;
+@SuppressWarnings("javadoc")
public class DataToStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -88,7 +89,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 1);
}
-
+
@Test
public void execMixedNormalCase() throws Exception {
EvalFunc<Tuple> func = new DataToStringsSketch();
@@ -137,7 +138,7 @@
sketch = getSketch(resultTuple);
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
-
+
// mixed null case
bag = BAG_FACTORY.newDefaultBag();
bag.add(TUPLE_FACTORY.newTuple("a"));
@@ -215,7 +216,7 @@
Assert.assertFalse(sketch.isEmpty());
Assert.assertEquals(sketch.getN(), 2);
}
-
+
@Test
public void algebraicIntermediateFinalMixedNullCase() throws Exception {
@SuppressWarnings("unchecked")
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/DoublesSketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/DoublesSketchToStringTest.java
index 50a5025..794368b 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/DoublesSketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/DoublesSketchToStringTest.java
@@ -31,6 +31,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class DoublesSketchToStringTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketchTest.java
index 3a18222..3a884f6 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromDoublesSketchTest.java
@@ -30,6 +30,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetKFromDoublesSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketchTest.java
index 3a98e38..1ab5155 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetKFromStringsSketchTest.java
@@ -33,6 +33,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetKFromStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketchTest.java
index 7d02a48..9e1e499 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromDoublesSketchTest.java
@@ -32,6 +32,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetPmfFromDoublesSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
@@ -47,7 +48,9 @@
public void normalCase() throws Exception {
EvalFunc<Tuple> func = new GetPmfFromDoublesSketch();
UpdateDoublesSketch sketch = DoublesSketch.builder().build();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
Tuple resultTuple = func.exec(tupleFactory.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 2.0, 7.0)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketchTest.java
index 8e38f06..3c7cd6c 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetPmfFromStringsSketchTest.java
@@ -34,6 +34,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetPmfFromStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -52,7 +53,9 @@
public void normalCase() throws Exception {
EvalFunc<Tuple> func = new GetPmfFromStringsSketch();
ItemsSketch<String> sketch = ItemsSketch.getInstance(COMPARATOR);
- for (int i = 1; i <= 10; i++) sketch.update(String.format("%02d", i));
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(String.format("%02d", i));
+ }
Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray(SER_DE)), "02", "07")));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketchTest.java
index a28ee58..afa026d 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromDoublesSketchTest.java
@@ -31,6 +31,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantileFromDoublesSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketchTest.java
index f748a17..68417a2 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantileFromStringsSketchTest.java
@@ -33,6 +33,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantileFromStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketchTest.java
index 09ad47a..9cdb18d 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromDoublesSketchTest.java
@@ -32,6 +32,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantilesFromDoublesSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
@@ -75,7 +76,9 @@
public void oneFraction() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromDoublesSketch();
UpdateDoublesSketch sketch = DoublesSketch.builder().build();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
Tuple resultTuple = func.exec(tupleFactory.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 0.5)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 1);
@@ -86,7 +89,9 @@
public void severalFractions() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromDoublesSketch();
UpdateDoublesSketch sketch = DoublesSketch.builder().build();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
Tuple resultTuple = func.exec(tupleFactory.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 0.0, 0.5, 1.0)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
@@ -99,7 +104,9 @@
public void numberOfEvenlySpacedIntervals() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromDoublesSketch();
UpdateDoublesSketch sketch = DoublesSketch.builder().build();
- for (int i = 1; i <= 10; i++) sketch.update(i);
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(i);
+ }
Tuple resultTuple = func.exec(tupleFactory.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray()), 3)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketchTest.java
index 2a11a18..c3f3976 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/GetQuantilesFromStringsSketchTest.java
@@ -34,6 +34,7 @@
import org.testng.annotations.Test;
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class GetQuantilesFromStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
@@ -81,7 +82,9 @@
public void oneFraction() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromStringsSketch();
ItemsSketch<String> sketch = ItemsSketch.getInstance(COMPARATOR);
- for (int i = 1; i <= 10; i++) sketch.update(String.format("%02d", i));
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(String.format("%02d", i));
+ }
Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray(SER_DE)), 0.5)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 1);
@@ -92,7 +95,9 @@
public void severalFractions() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromStringsSketch();
ItemsSketch<String> sketch = ItemsSketch.getInstance(COMPARATOR);
- for (int i = 1; i <= 10; i++) sketch.update(String.format("%02d", i));
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(String.format("%02d", i));
+ }
Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray(SER_DE)), 0.0, 0.5, 1.0)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
@@ -105,7 +110,9 @@
public void numberOfEvenlySpacedIntervals() throws Exception {
EvalFunc<Tuple> func = new GetQuantilesFromStringsSketch();
ItemsSketch<String> sketch = ItemsSketch.getInstance(COMPARATOR);
- for (int i = 1; i <= 10; i++) sketch.update(String.format("%02d", i));
+ for (int i = 1; i <= 10; i++) {
+ sketch.update(String.format("%02d", i));
+ }
Tuple resultTuple = func.exec(TUPLE_FACTORY.newTuple(Arrays.asList(new DataByteArray(sketch.toByteArray(SER_DE)), 3)));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 3);
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/StringsSketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/StringsSketchToStringTest.java
index 0458449..2862bb8 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/StringsSketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/StringsSketchToStringTest.java
@@ -34,6 +34,7 @@
import org.testng.Assert;
+@SuppressWarnings("javadoc")
public class StringsSketchToStringTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketchTest.java
index d901af1..c5921b9 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/UnionDoublesSketchTest.java
@@ -35,6 +35,7 @@
import org.apache.datasketches.quantiles.DoublesSketch;
import org.apache.datasketches.quantiles.UpdateDoublesSketch;
+@SuppressWarnings("javadoc")
public class UnionDoublesSketchTest {
private static final TupleFactory tupleFactory = TupleFactory.getInstance();
private static final BagFactory bagFactory = BagFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/quantiles/UnionStringsSketchTest.java b/src/test/java/org/apache/datasketches/pig/quantiles/UnionStringsSketchTest.java
index 705a50f..a8e4755 100644
--- a/src/test/java/org/apache/datasketches/pig/quantiles/UnionStringsSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/quantiles/UnionStringsSketchTest.java
@@ -38,7 +38,7 @@
import org.apache.datasketches.ArrayOfStringsSerDe;
import org.apache.datasketches.quantiles.ItemsSketch;
-
+@SuppressWarnings("javadoc")
public class UnionStringsSketchTest {
private static final TupleFactory TUPLE_FACTORY = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/DataToVarOptSketchTest.java b/src/test/java/org/apache/datasketches/pig/sampling/DataToVarOptSketchTest.java
index 366db35..ca6a05d 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/DataToVarOptSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/DataToVarOptSketchTest.java
@@ -38,7 +38,9 @@
import org.apache.datasketches.memory.Memory;
import org.apache.datasketches.sampling.VarOptItemsSketch;
+@SuppressWarnings("javadoc")
public class DataToVarOptSketchTest {
+
@Test
@SuppressWarnings("unused")
public void checkConstructors() {
@@ -85,7 +87,7 @@
// calling accumulate() twice, but keep in exact mode so reference sketch has same values
try {
final VarOptItemsSketch<Tuple> sketch = VarOptItemsSketch.newInstance(k);
- for (int i = 1; i < k / 2; ++i) {
+ for (int i = 1; i < (k / 2); ++i) {
final Tuple t = TupleFactory.getInstance().newTuple(3);
t.set(0, 1.0 * i);
t.set(1, i);
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/GetVarOptSamplesTest.java b/src/test/java/org/apache/datasketches/pig/sampling/GetVarOptSamplesTest.java
index a95a4b0..70c9847 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/GetVarOptSamplesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/GetVarOptSamplesTest.java
@@ -40,6 +40,7 @@
import org.apache.datasketches.sampling.VarOptItemsSketch;
+@SuppressWarnings("javadoc")
public class GetVarOptSamplesTest {
private static final ArrayOfTuplesSerDe serDe_ = new ArrayOfTuplesSerDe();
@@ -83,7 +84,7 @@
cumResultWt += (double) sample.get(0);
final Tuple record = (Tuple) sample.get(1);
final int id = (int) record.get(1);
- assertTrue(id >= 1 && id <= n);
+ assertTrue((id >= 1) && (id <= n));
}
assertEquals(cumResultWt, cumWt, EPS);
} catch (final IOException e) {
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/ReservoirSamplingTest.java b/src/test/java/org/apache/datasketches/pig/sampling/ReservoirSamplingTest.java
index de40050..7a391e0 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/ReservoirSamplingTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/ReservoirSamplingTest.java
@@ -40,6 +40,7 @@
import org.apache.pig.impl.logicalLayer.schema.Schema;
import org.testng.annotations.Test;
+@SuppressWarnings("javadoc")
public class ReservoirSamplingTest {
@SuppressWarnings("unused")
@@ -131,7 +132,7 @@
assertEquals(((DataBag) result.get(2)).size(), n);
// add another n to the bag and repeat
- for (long i = n; i < 2 * n; ++i) {
+ for (long i = n; i < (2 * n); ++i) {
final Tuple t = tf.newTuple(2);
t.set(0, i);
t.set(1, Long.toString(-i));
@@ -231,7 +232,7 @@
// check that they're all in the target range
for (Tuple sample : ((DataBag) result.get(2))) {
final int val = (int) sample.get(0);
- if (val < 0 || val >= tgtMaxVal) {
+ if ((val < 0) || (val >= tgtMaxVal)) {
fail("Found value (" + val + ") outside target range [0, " + tgtMaxVal + "]");
}
}
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/ReservoirUnionTest.java b/src/test/java/org/apache/datasketches/pig/sampling/ReservoirUnionTest.java
index 2800cfc..897cd9b 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/ReservoirUnionTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/ReservoirUnionTest.java
@@ -39,6 +39,7 @@
import org.apache.pig.impl.logicalLayer.schema.Schema;
import org.testng.annotations.Test;
+@SuppressWarnings("javadoc")
public class ReservoirUnionTest {
@SuppressWarnings("unused")
@@ -139,7 +140,7 @@
for (Tuple t : outputSamples) {
// expected format: (i:int, -i:chararray)
final int i = (int) t.get(0);
- assertTrue(i >= 0 && i < k);
+ assertTrue((i >= 0) && (i < k));
assertEquals((String) t.get(1), Integer.toString(-i));
}
} catch (final IOException e) {
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/VarOptCommonAlgebraicTest.java b/src/test/java/org/apache/datasketches/pig/sampling/VarOptCommonAlgebraicTest.java
index e6be5bc..68c9dd0 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/VarOptCommonAlgebraicTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/VarOptCommonAlgebraicTest.java
@@ -41,6 +41,7 @@
import org.apache.datasketches.sampling.VarOptItemsSketch;
import org.apache.datasketches.sampling.VarOptItemsUnion;
+@SuppressWarnings("javadoc")
public class VarOptCommonAlgebraicTest {
private static final ArrayOfTuplesSerDe serDe_ = new ArrayOfTuplesSerDe();
@@ -167,7 +168,7 @@
final DataBag inputBag = BagFactory.getInstance().newDefaultBag();
try {
- for (int i = 0; i < k + 1; ++i) {
+ for (int i = 0; i < (k + 1); ++i) {
final Tuple t = TupleFactory.getInstance().newTuple(2);
t.set(0, Character.toString(id));
t.set(1, wt);
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/VarOptSamplingTest.java b/src/test/java/org/apache/datasketches/pig/sampling/VarOptSamplingTest.java
index 9858396..c8ae5ed 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/VarOptSamplingTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/VarOptSamplingTest.java
@@ -41,6 +41,7 @@
import org.apache.datasketches.sampling.VarOptItemsSketch;
+@SuppressWarnings("javadoc")
public class VarOptSamplingTest {
static final double EPS = 1e-10;
private static final ArrayOfTuplesSerDe serDe_ = new ArrayOfTuplesSerDe();
@@ -149,7 +150,7 @@
assertEquals(sample.size(), 3);
final int id = (int) sample.get(1);
- assertTrue(id > 0 && id < k);
+ assertTrue((id > 0) && (id < k));
}
assertEquals(cumResultWeight, 2 * cumWeight, EPS); // called accumulate() twice
} catch (final IOException e) {
diff --git a/src/test/java/org/apache/datasketches/pig/sampling/VarOptUnionTest.java b/src/test/java/org/apache/datasketches/pig/sampling/VarOptUnionTest.java
index 8bde81b..cc380b3 100644
--- a/src/test/java/org/apache/datasketches/pig/sampling/VarOptUnionTest.java
+++ b/src/test/java/org/apache/datasketches/pig/sampling/VarOptUnionTest.java
@@ -39,6 +39,7 @@
import org.apache.datasketches.sampling.VarOptItemsSketch;
import org.apache.datasketches.sampling.VarOptItemsUnion;
+@SuppressWarnings("javadoc")
public class VarOptUnionTest {
@SuppressWarnings("unused")
@Test
@@ -69,7 +70,7 @@
try {
final VarOptItemsSketch<Tuple> sketch = VarOptItemsSketch.newInstance(k);
final VarOptItemsUnion<Tuple> union = VarOptItemsUnion.newInstance(k);
- for (int i = 1; i < k / 2; ++i) {
+ for (int i = 1; i < (k / 2); ++i) {
sketch.reset();
final Tuple t = TupleFactory.getInstance().newTuple(3);
t.set(0, 1.0 * i);
diff --git a/src/test/java/org/apache/datasketches/pig/theta/AexcludeBTest.java b/src/test/java/org/apache/datasketches/pig/theta/AexcludeBTest.java
index dbd38b5..e592166 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/AexcludeBTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/AexcludeBTest.java
@@ -37,11 +37,9 @@
import org.apache.datasketches.pig.theta.AexcludeB;
import org.apache.datasketches.pig.theta.Estimate;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class AexcludeBTest {
-
+
@SuppressWarnings("unused")
@Test
public void checkConstructors() {
@@ -49,12 +47,12 @@
aNOTb = new AexcludeB("9001");
aNOTb = new AexcludeB(9001);
}
-
+
@Test
public void checkNullCombinations() throws IOException {
EvalFunc<Tuple> aNbFunc = new AexcludeB();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple, resultTuple;
Double est;
//Two nulls
@@ -64,7 +62,7 @@
assertEquals(resultTuple.size(), 1);
est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
-
+
//A is null
inputTuple = TupleFactory.getInstance().newTuple(2);
inputTuple.set(1, createDbaFromQssRange(256, 0, 128));
@@ -73,7 +71,7 @@
assertEquals(resultTuple.size(), 1);
est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
-
+
//A is valid, B is null
inputTuple = TupleFactory.getInstance().newTuple(2);
inputTuple.set(0, createDbaFromQssRange(256, 0, 256));
@@ -82,7 +80,7 @@
assertEquals(resultTuple.size(), 1);
est = estFunc.exec(resultTuple);
assertEquals(est, 256.0, 0.0);
-
+
//Both valid
inputTuple = TupleFactory.getInstance().newTuple(2);
inputTuple.set(0, createDbaFromQssRange(256, 0, 256));
@@ -93,43 +91,43 @@
est = estFunc.exec(resultTuple);
assertEquals(est, 128.0, 0.0);
}
-
+
@Test
public void outputSchemaTest() throws IOException {
EvalFunc<Tuple> udf = new AexcludeB("512");
-
+
Schema inputSchema = null;
-
+
Schema nullOutputSchema = null;
-
+
Schema outputSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputInnerFs0 = null;
-
+
inputSchema = Schema.generateNestedSchema(DataType.BAG, DataType.BYTEARRAY);
-
+
nullOutputSchema = udf.outputSchema(null);
-
+
outputSchema = udf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
-
+
outputInnerSchema = outputOuterFs0.schema;
outputInnerFs0 = outputInnerSchema.getField(0);
-
+
Assert.assertNull(nullOutputSchema, "Should be null");
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) schema may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "bytearray";
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) schema may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -144,17 +142,17 @@
//@formatter:on
//end print schemas
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/DataToSketchTest.java b/src/test/java/org/apache/datasketches/pig/theta/DataToSketchTest.java
index b765bf5..cb7511d 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/DataToSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/DataToSketchTest.java
@@ -48,31 +48,29 @@
import org.apache.datasketches.pig.theta.Estimate;
import org.apache.datasketches.theta.Sketch;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class DataToSketchTest {
private String udfName = "org.apache.datasketches.pig.theta.DataToSketch";
private long seed_ = Util.DEFAULT_UPDATE_SEED;
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = SketchesArgumentException.class)
public void testConstructorExceptions1() {
DataToSketch test = new DataToSketch("1023");
}
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = IllegalArgumentException.class)
public void testConstructorExceptions3() {
DataToSketch test = new DataToSketch("8");
}
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = SketchesArgumentException.class)
public void testConstructorExceptions4() {
DataToSketch test = new DataToSketch("1024", "2.0");
}
-
+
@Test
public void checkNotDBAExcep() throws IOException {
DataToSketch inter = new DataToSketch();
@@ -80,38 +78,38 @@
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
inter.accumulate(inputTuple); //add empty tuple
-
+
innerTuple.set(0, new Double(1.0)); //not a DBA
inter = new DataToSketch();
inter.accumulate(inputTuple); //add wrong type
}
-
+
@SuppressWarnings("unused")
@Test
public void checkConstructors() {
-
+
DataToSketch inter = new DataToSketch();
inter = new DataToSketch("1024");
inter = new DataToSketch("1024", "1.0");
inter = new DataToSketch("1024", "1.0", "9001");
inter = new DataToSketch(1024, (float) 1.0, 9001);
-
+
DataToSketch.Initial initial = new DataToSketch.Initial();
initial = new DataToSketch.Initial("1024");
initial = new DataToSketch.Initial("1024", "1.0");
initial = new DataToSketch.Initial("1024", "1.0", "9001");
-
+
DataToSketch.IntermediateFinal interFin = new DataToSketch.IntermediateFinal();
interFin = new DataToSketch.IntermediateFinal("1024");
interFin = new DataToSketch.IntermediateFinal("1024", "1.0");
interFin = new DataToSketch.IntermediateFinal("1024", "1.0", "9001");
interFin = new DataToSketch.IntermediateFinal(1024, (float) 1.0, 9001);
}
-
+
@Test
public void testTopExec() throws IOException {
EvalFunc<Tuple> func = new DataToSketch(); //empty constructor, size 4096
@@ -119,9 +117,9 @@
Tuple inputTuple = null;
Tuple resultTuple = func.exec(inputTuple);
Sketch sketch = tupleToSketch(resultTuple, seed_);
-
+
assertTrue(sketch.isEmpty());
-
+
inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
@@ -141,7 +139,7 @@
sketch = tupleToSketch(resultTuple, seed_);
assertEquals(sketch.getEstimate(), 64.0, 0.0);
}
-
+
/*
* DataToSketch <br>
* Tests all possible data types: NULL, BYTE, INTEGER, LONG, FLOAT, DOUBLE,
@@ -227,13 +225,13 @@
Sketch sketch = tupleToSketch(resultTuple, seed_);
assertEquals(sketch.getEstimate(), 8.0, 0.0);
}
-
+
@SuppressWarnings("unchecked") //still triggers unchecked warning
@Test(expectedExceptions = IllegalArgumentException.class)
public void testRejectionOfNonSimpleType() throws IOException {
TupleFactory mTupleFactory = TupleFactory.getInstance();
BagFactory bagFactory = BagFactory.getInstance();
-
+
Tuple outerTuple = mTupleFactory.newTuple(1);
DataBag outerBag = bagFactory.newDefaultBag();
Tuple innerTuple = mTupleFactory.newTuple(1);
@@ -241,13 +239,13 @@
innerTuple.set(0, innerBag);
outerBag.add(innerTuple);
outerTuple.set(0, outerBag);
-
+
String[] ctorArgs = { "128" };
EvalFunc<Tuple> dataUdf =
(EvalFunc<Tuple>) PigContext.instantiateFuncFromSpec(new FuncSpec(udfName, ctorArgs));
dataUdf.exec(outerTuple);
}
-
+
@Test
public void testAccumulate() throws IOException {
Accumulator<Tuple> func = new DataToSketch("128");
@@ -295,12 +293,12 @@
sketch = tupleToSketch(resultTuple, seed_);
assertEquals(sketch.getEstimate(), 0.0, 0.0);
}
-
+
@Test
public void testInitial() throws IOException {
EvalFunc<Tuple> func = new DataToSketch.Initial("128");
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
@@ -318,22 +316,22 @@
DataBag resultBag = (DataBag) resultTuple.get(0);
assertEquals(resultBag.size(), 64);
}
-
+
@Test
public void testIntermediateFinal() throws IOException {
EvalFunc<Tuple> func = new DataToSketch.IntermediateFinal("128");
-
+
Tuple inputTuple = null;
Tuple resultTuple = func.exec(inputTuple);
Sketch sketch = tupleToSketch(resultTuple, seed_);
assertTrue(sketch.isEmpty());
-
+
inputTuple = TupleFactory.getInstance().newTuple(0);
resultTuple = func.exec(inputTuple);
sketch = tupleToSketch(resultTuple, seed_);
assertTrue(sketch.isEmpty());
-
-
+
+
inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
@@ -364,109 +362,109 @@
sketch = tupleToSketch(resultTuple, seed_);
assertEquals(sketch.getEstimate(), 100.0, 0.0);
}
-
+
@Test
public void checkAlgFinalOuterBagEmptyTuples() throws IOException {
EvalFunc<Tuple> interFuncFinal = new DataToSketch.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test
public void checkAlgFinalInnerBagEmpty() throws IOException {
EvalFunc<Tuple> interFuncFinal = new DataToSketch.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
DataBag bag2 = BagFactory.getInstance().newDefaultBag();
innerTuple.set(0, bag2);
-
+
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkAlgFinalInnerNotDBA() throws IOException {
EvalFunc<Tuple> interFuncFinal = new DataToSketch.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
innerTuple.set(0, new Double(1.0)); //not a DBA
-
+
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test
public void outputSchemaTest() throws IOException {
EvalFunc<Tuple> udf = new DataToSketch("512");
-
+
Schema inputSchema = null;
-
+
Schema nullOutputSchema = null;
-
+
Schema outputSchema = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
Schema.FieldSchema outputInnerFs0 = null;
//CHARARRAY is one of several possible inner types
inputSchema = Schema.generateNestedSchema(DataType.BAG, DataType.CHARARRAY);
-
+
nullOutputSchema = udf.outputSchema(null);
outputSchema = udf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
-
+
outputInnerSchema = outputOuterFs0.schema;
outputInnerFs0 = outputInnerSchema.getField(0);
Assert.assertNull(nullOutputSchema, "Should be null");
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "bytearray";
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -481,7 +479,7 @@
//@formatter:on
//end print schemas
}
-
+
@Test
@SuppressWarnings("unused")
public void checkMisc() throws IOException {
@@ -493,9 +491,9 @@
dts.accumulate(inputTuple);
Tuple resultTuple = dts.getValue();
Sketch sketch = tupleToSketch(resultTuple, seed_);
- assertTrue(sketch.isEmpty());
+ assertTrue(sketch.isEmpty());
}
-
+
@Test
public void checkSmall() throws IOException {
EvalFunc<Tuple> func = new DataToSketch("32");
@@ -503,9 +501,9 @@
Tuple inputTuple = null;
Tuple resultTuple = func.exec(inputTuple);
Sketch sketch = tupleToSketch(resultTuple, seed_);
-
+
assertTrue(sketch.isEmpty());
-
+
inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
@@ -526,17 +524,17 @@
sketch = tupleToSketch(resultTuple, seed_);
assertEquals(sketch.getEstimate(), u, 0.0);
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/ErrorBoundsTest.java b/src/test/java/org/apache/datasketches/pig/theta/ErrorBoundsTest.java
index 38b9acb..3df5436 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/ErrorBoundsTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/ErrorBoundsTest.java
@@ -40,9 +40,7 @@
import org.apache.datasketches.pig.theta.ErrorBounds;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class ErrorBoundsTest {
@Test
@@ -51,24 +49,24 @@
Tuple dataTuple = null;
Tuple result = func.exec(dataTuple);
assertNull(result);
-
+
dataTuple = TupleFactory.getInstance().newTuple(0);
result = func.exec(dataTuple);
assertNull(result);
}
-
+
@Test
public void testExactModeBounds() throws IOException {
EvalFunc<Tuple> func = new ErrorBounds();
int nomEntries = 128;
-
+
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(nomEntries, 0, 64));
Tuple result = func.exec(dataTuple);
assertNotNull(result);
assertEquals(result.size(), 3);
-
+
assertEquals(((Double) result.get(0)).doubleValue(), 64.0, 0.0);
assertEquals(((Double) result.get(1)).doubleValue(), 64.0, 0.0);
assertEquals(((Double) result.get(2)).doubleValue(), 64.0, 0.0);
@@ -97,7 +95,7 @@
assertTrue(Math.abs((ub / (est + (epsilon2SD * est))) - 1) < .01);
assertTrue(Math.abs((lb / (est - (epsilon2SD * est))) - 1) < .01);
}
-
+
@Test
public void testEstModeBoundsWithSeed() throws IOException {
EvalFunc<Tuple> func = new ErrorBounds(Long.toString(DEFAULT_UPDATE_SEED));
@@ -121,11 +119,11 @@
assertTrue(Math.abs((ub / (est + (epsilon2SD * est))) - 1) < .01);
assertTrue(Math.abs((lb / (est - (epsilon2SD * est))) - 1) < .01);
}
-
+
@Test
public void outputSchemaTest() throws IOException {
EvalFunc<Tuple> udf = new ErrorBounds();
-
+
Schema inputSchema = null;
Schema.FieldSchema inputFieldSchema = new Schema.FieldSchema("Sketch", DataType.BYTEARRAY);
@@ -133,16 +131,16 @@
Schema outputSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputInnerFs0 = null;
Schema.FieldSchema outputInnerFs1 = null;
Schema.FieldSchema outputInnerFs2 = null;
inputSchema = new Schema(inputFieldSchema);
-
+
nullOutputSchema = udf.outputSchema(null);
-
+
outputSchema = udf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
outputInnerSchema = outputOuterFs0.schema;
@@ -151,27 +149,27 @@
outputInnerFs2 = outputInnerSchema.getField(2);
Assert.assertNull(nullOutputSchema, "Should be null");
-
+
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) schema may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "double";
-
+
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) schema may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
Assert.assertNotNull(outputInnerFs1, "innerSchema.getField(1) schema may not be null");
result = DataType.findTypeName(outputInnerFs1.type);
Assert.assertEquals(result, expected);
-
+
Assert.assertNotNull(outputInnerFs2, "innerSchema.getField(2) schema may not be null");
result = DataType.findTypeName(outputInnerFs2.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -190,17 +188,17 @@
//@formatter:on
//end print schemas
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/EstimateTest.java b/src/test/java/org/apache/datasketches/pig/theta/EstimateTest.java
index 6a430f1..760db9f 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/EstimateTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/EstimateTest.java
@@ -34,9 +34,7 @@
import org.apache.datasketches.pig.theta.Estimate;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class EstimateTest {
@Test
@@ -45,12 +43,12 @@
Tuple inputTuple = null;
Double returnValue = func.exec(inputTuple);
assertNull(returnValue);
-
+
inputTuple = TupleFactory.getInstance().newTuple(0);
returnValue = func.exec(inputTuple);
assertNull(returnValue);
}
-
+
@Test
public void testExact() throws IOException {
EvalFunc<Double> func = new Estimate();
@@ -62,7 +60,7 @@
assertNotNull(result);
assertEquals(result, 64.0, 0.0);
}
-
+
@Test
public void testExactWithSeed() throws IOException {
EvalFunc<Double> func = new Estimate(Long.toString(DEFAULT_UPDATE_SEED));
@@ -74,17 +72,17 @@
assertNotNull(result);
assertEquals(result, 64.0, 0.0);
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/IntersectTest.java b/src/test/java/org/apache/datasketches/pig/theta/IntersectTest.java
index ef00ee7..41e4620 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/IntersectTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/IntersectTest.java
@@ -43,17 +43,15 @@
import org.apache.datasketches.pig.theta.Estimate;
import org.apache.datasketches.pig.theta.Intersect;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class IntersectTest {
-
+
@Test(expectedExceptions = IllegalStateException.class)
public void checkGetValueExcep() {
Intersect inter = new Intersect();
inter.getValue();
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkNotDBAExcep() throws IOException {
Intersect inter = new Intersect();
@@ -61,356 +59,356 @@
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
inter.accumulate(inputTuple); //add empty tuple
-
+
innerTuple.set(0, new Double(1.0)); //not a DBA
inter = new Intersect();
inter.accumulate(inputTuple); //add wrong type
}
-
-
+
+
@SuppressWarnings("unused")
@Test
public void checkConstructors() {
Intersect inter = new Intersect();
inter = new Intersect(9001);
-
+
Intersect.Initial initial = new Intersect.Initial();
initial = new Intersect.Initial("9001");
-
+
Intersect.IntermediateFinal interFin = new Intersect.IntermediateFinal();
interFin = new Intersect.IntermediateFinal("9001");
interFin = new Intersect.IntermediateFinal(9001);
inter.cleanup();
}
-
+
@Test
public void checkNullInput() throws IOException {
EvalFunc<Tuple> interFunc = new Intersect();
EvalFunc<Double> estFunc = new Estimate();
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
//null bag
-
+
Tuple resultTuple = interFunc.exec(inputTuple);
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
}
-
+
@Test
public void checkExactTopExec() throws IOException {
EvalFunc<Tuple> interFunc = new Intersect();
EvalFunc<Double> estFunc = new Estimate();
-
+
//create inputTuple and a bag, add bag to inputTuple
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
//create 4 overlapping sketches of 64 in a bag
for (int i = 0; i < 4; i++ ) {
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(256, i*64, 256));
-
+
bag.add(dataTuple);
}
-
+
Tuple resultTuple = interFunc.exec(inputTuple);
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 64.0, 0.0);
}
-
+
@Test(expectedExceptions = ClassCastException.class)
public void checkBadClassCast() throws IOException {
Accumulator<Tuple> interFunc = new Intersect();
Tuple inputTuple = TupleFactory.getInstance().newTuple(1); //valid size, but null
-
+
inputTuple.set(0, new Double(1.0)); //wrong type. Cannot intersect datums.
interFunc.accumulate(inputTuple); //throws ClassCastException
}
-
+
@Test
public void checkNullEmptyAccumulator() throws IOException {
Accumulator<Tuple> interFunc = new Intersect();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = null;
interFunc.accumulate(inputTuple); //does nothing
-
+
inputTuple = TupleFactory.getInstance().newTuple(0); //invalid size
interFunc.accumulate(inputTuple); //does nothing
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size, but null bag
interFunc.accumulate(inputTuple); //does nothing
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //correct type, but empty
interFunc.accumulate(inputTuple); //does nothing
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(0); //empty
bag.add(innerTuple);
interFunc.accumulate(inputTuple); //does nothing
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size
bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //correct type
innerTuple = TupleFactory.getInstance().newTuple(1); //correct size
bag.add(innerTuple); //but innerTuple(0) is null
interFunc.accumulate(inputTuple); //does nothing
-
+
//Must call accumulate at least once before calling getValue.
//To prove that all the above stuff truely did nothing,
// we call accumulate once with a valid sketch and affirm that
// getValue() returns it unaltered.
-
+
//create inputTuple and a bag, add bag to inputTuple
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size
bag = BagFactory.getInstance().newDefaultBag();
- inputTuple.set(0, bag);
-
+ inputTuple.set(0, bag);
+
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(256, 0, 64));
bag.add(dataTuple);
-
+
interFunc.accumulate(inputTuple);
-
+
Tuple resultTuple = interFunc.getValue();
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 64.0, 0.0);
}
-
+
@Test
public void checkExactAccumulator() throws IOException {
Accumulator<Tuple> interFunc = new Intersect();
EvalFunc<Double> estFunc = new Estimate();
-
+
//create inputTuple and a bag, add bag to inputTuple
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
//create 4 distinct sketches of 32 in a bag
for (int i = 0; i < 4; i++ ) {
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(256, i*64, 256));
-
+
bag.add(dataTuple);
}
-
+
interFunc.accumulate(inputTuple); //A tuple, bag with 4 sketches
-
+
Tuple resultTuple = interFunc.getValue();
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray dba = (DataByteArray) resultTuple.get(0);
assertTrue(dba.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 64.0, 0.0);
}
-
+
@Test
public void checkExactAlgebraicInitial() throws IOException {
EvalFunc<Tuple> interFuncInit = new Intersect.Initial();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
for (int i = 0; i < 4; i++ ) { //4 sketches with one value each
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(16, i, 1));
-
+
bag.add(dataTuple);
}
-
+
Tuple resultTuple = interFuncInit.exec(inputTuple);
assertTrue(resultTuple == inputTuple); //returns the inputTuple
}
-
+
@Test
public void checkAlgFinalFromPriorIntermed() throws IOException {
EvalFunc<Tuple> interFuncIFinal = new Intersect.IntermediateFinal();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
-
+
for (int i = 0; i < 4; i++ ) {
Tuple sketchTuple = TupleFactory.getInstance().newTuple(1);
sketchTuple.set(0, createDbaFromQssRange(256, i*64, 256));
-
- bag.add(sketchTuple);
+
+ bag.add(sketchTuple);
//inputTuple.bag0:sketchTuple0.DBA0
//inputTuple.bag0:sketchTuple1.DBA1
//inputTuple.bag0:sketchTuple2.DBA2
//inputTuple.bag0:sketchTuple3.DBA3
}
-
+
Tuple resultTuple = interFuncIFinal.exec(inputTuple);
-
+
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
assertTrue(bytes.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 64.0, 0.0);
}
-
+
@Test
public void checkAlgFinalFromPriorInitial() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Intersect.IntermediateFinal();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
DataBag innerBag = BagFactory.getInstance().newDefaultBag();
innerTuple.set(0, innerBag); //innerTuple.innerBag0:null
bag.add(innerTuple); //inputTuple.bag0.innerTuple0.innerBag0:null
-
+
for (int i = 0; i < 4; i++ ) {
Tuple sketchTuple = TupleFactory.getInstance().newTuple(1);
sketchTuple.set(0, createDbaFromQssRange(256, i*64, 256));
-
+
innerBag.add(sketchTuple);
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple0.DBA0
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple1.DBA1
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple2.DBA2
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple3.DBA3
}
-
+
Tuple resultTuple = interFuncFinal.exec(inputTuple);
-
+
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
assertTrue(bytes.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 64.0, 0.0);
}
-
+
@Test(expectedExceptions = SketchesStateException.class)
public void checkAlgFinalOuterBagEmptyTuples() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Intersect.IntermediateFinal();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
resultTuple = interFuncFinal.exec(inputTuple); //Throws Illegal Result from HeapIntersection
//assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test(expectedExceptions = SketchesStateException.class)
public void checkAlgFinalInnerBagEmpty() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Intersect.IntermediateFinal();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
DataBag bag2 = BagFactory.getInstance().newDefaultBag();
innerTuple.set(0, bag2);
-
+
resultTuple = interFuncFinal.exec(inputTuple); //Throws Illegal Result from HeapIntersection
//assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkAlgFinalInnerNotDBA() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Intersect.IntermediateFinal();
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
innerTuple.set(0, new Double(1.0)); //not a DBA
-
+
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test
public void outputSchemaTest() throws IOException {
EvalFunc<Tuple> udf = new Intersect();
-
+
Schema inputSchema = null;
-
+
Schema nullOutputSchema = null;
-
+
Schema outputSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputInnerFs0 = null;
-
+
inputSchema = Schema.generateNestedSchema(DataType.BAG, DataType.BYTEARRAY);
-
+
nullOutputSchema = udf.outputSchema(null);
-
+
outputSchema = udf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
-
+
outputInnerSchema = outputOuterFs0.schema;
outputInnerFs0 = outputInnerSchema.getField(0);
-
+
Assert.assertNull(nullOutputSchema, "Should be null");
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) schema may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "bytearray";
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) schema may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -425,17 +423,17 @@
//@formatter:on
//end print schemas
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/JaccardSimilarityTest.java b/src/test/java/org/apache/datasketches/pig/theta/JaccardSimilarityTest.java
index fd34f83..6f12e7e 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/JaccardSimilarityTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/JaccardSimilarityTest.java
@@ -30,9 +30,7 @@
import org.apache.pig.data.TupleFactory;
import org.testng.annotations.Test;
-/**
- * @author eshcar
- */
+@SuppressWarnings("javadoc")
public class JaccardSimilarityTest {
@Test
diff --git a/src/test/java/org/apache/datasketches/pig/theta/PigUtilTest.java b/src/test/java/org/apache/datasketches/pig/theta/PigUtilTest.java
index 471c6d6..fddaa05 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/PigUtilTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/PigUtilTest.java
@@ -30,35 +30,35 @@
import org.apache.datasketches.theta.CompactSketch;
import org.apache.datasketches.theta.UpdateSketch;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class PigUtilTest {
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkCompOrdSketchToTuple() {
UpdateSketch usk = UpdateSketch.builder().setNominalEntries(16).build();
- for (int i=0; i<16; i++) usk.update(i);
+ for (int i=0; i<16; i++) {
+ usk.update(i);
+ }
CompactSketch csk = usk.compact(false, null);
compactOrderedSketchToTuple(csk);
}
-
+
@Test
public void checkExtractTypeAtIndex() {
Tuple tuple = TupleFactory.getInstance().newTuple(0);
assertNull(extractTypeAtIndex(tuple, 0));
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/SketchToStringTest.java b/src/test/java/org/apache/datasketches/pig/theta/SketchToStringTest.java
index 91e3356..4324b8f 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/SketchToStringTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/SketchToStringTest.java
@@ -35,72 +35,70 @@
import org.apache.datasketches.pig.theta.SketchToString;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class SketchToStringTest {
-
+
@Test
public void testNullEmpty() throws IOException {
EvalFunc<String> func = new SketchToString("false");
Tuple dataTuple = null;
String result = func.exec(dataTuple);
assertNull(result);
-
+
dataTuple = TupleFactory.getInstance().newTuple(0);
result = func.exec(dataTuple);
assertNull(result);
}
-
+
@Test
public void testExactNoDetail() throws IOException {
EvalFunc<String> func = new SketchToString("false");
-
+
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(64, 0, 64));
-
+
String result = func.exec(dataTuple);
assertNotNull(result);
assertTrue(result.contains("SUMMARY"));
assertFalse(result.contains("SKETCH DATA DETAIL"));
}
-
+
@Test
public void testExactNoDetailWithSeed() throws IOException {
EvalFunc<String> func = new SketchToString("false", Long.toString(DEFAULT_UPDATE_SEED));
-
+
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(64, 0, 64));
-
+
String result = func.exec(dataTuple);
assertNotNull(result);
assertTrue(result.contains("SUMMARY"));
assertFalse(result.contains("SKETCH DATA DETAIL"));
}
-
+
@Test
public void testExactWithDetail() throws IOException {
EvalFunc<String> func = new SketchToString("true");
-
+
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(64, 0, 64));
-
+
String result = func.exec(dataTuple);
assertNotNull(result);
assertTrue(result.contains("SUMMARY"));
assertTrue(result.contains("SKETCH DATA DETAIL"));
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/theta/UnionTest.java b/src/test/java/org/apache/datasketches/pig/theta/UnionTest.java
index 761b25b..7f80a1a 100644
--- a/src/test/java/org/apache/datasketches/pig/theta/UnionTest.java
+++ b/src/test/java/org/apache/datasketches/pig/theta/UnionTest.java
@@ -43,31 +43,29 @@
import org.apache.datasketches.pig.theta.Estimate;
import org.apache.datasketches.pig.theta.Union;
-/**
- * @author Lee Rhodes
- */
+@SuppressWarnings("javadoc")
public class UnionTest {
// private String udfName = "org.apache.datasketches.pig.theta.SketchUnions";
// private long seed_ = Util.DEFAULT_UPDATE_SEED;
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = SketchesArgumentException.class)
public void checkConstructorExceptions1() {
Union test = new Union("1023");
}
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkConstructorExceptions3() {
Union test = new Union("8");
}
-
+
@SuppressWarnings("unused")
@Test(expectedExceptions = SketchesArgumentException.class)
public void testConstructorExceptions4() {
Union test = new Union("1024", "2.0");
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkNotDBAExcep() throws IOException {
Union inter = new Union();
@@ -75,58 +73,58 @@
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
inter.accumulate(inputTuple); //add empty tuple
-
+
innerTuple.set(0, new Double(1.0)); //not a DBA
inter = new Union();
inter.accumulate(inputTuple); //add wrong type
}
-
+
@SuppressWarnings("unused")
@Test
public void checkConstructors() {
-
+
Union inter = new Union();
inter = new Union("1024");
inter = new Union("1024", "1.0");
inter = new Union("1024", "1.0", "9001");
inter = new Union(1024, (float) 1.0, 9001);
-
+
Union.Initial initial = new Union.Initial();
initial = new Union.Initial("1024");
initial = new Union.Initial("1024", "1.0");
initial = new Union.Initial("1024", "1.0", "9001");
-
+
Union.IntermediateFinal interFin = new Union.IntermediateFinal();
interFin = new Union.IntermediateFinal("1024");
interFin = new Union.IntermediateFinal("1024", "1.0");
interFin = new Union.IntermediateFinal("1024", "1.0", "9001");
interFin = new Union.IntermediateFinal(1024, (float) 1.0, 9001);
}
-
+
@Test
public void checkNullInput() throws IOException {
EvalFunc<Tuple> unionFunc = new Union(); //default 4096
EvalFunc<Double> estFunc = new Estimate();
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
//null bag
-
+
Tuple resultTuple = unionFunc.exec(inputTuple);
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
}
-
+
@Test
public void checkExactTopExec() throws IOException {
EvalFunc<Tuple> unionFunc = new Union(); //default 4096
EvalFunc<Double> estFunc = new Estimate();
-
+
//create inputTuple and a bag, add bag to inputTuple
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
@@ -143,262 +141,262 @@
Tuple resultTuple = unionFunc.exec(inputTuple);
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 256.0, 0.0);
}
-
+
@Test(expectedExceptions = ClassCastException.class)
public void checkBadClassCast() throws IOException {
Accumulator<Tuple> unionFunc = new Union("256");
Tuple inputTuple = TupleFactory.getInstance().newTuple(1); //valid size, but null
-
+
inputTuple.set(0, new Double(1.0)); //wrong type. Cannot Union datums.
unionFunc.accumulate(inputTuple); //throws ClassCastException
}
-
+
@Test
public void checkNullEmptyAccumulator() throws IOException {
Accumulator<Tuple> unionFunc = new Union("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = null;
unionFunc.accumulate(inputTuple); //does nothing, just returns
-
+
inputTuple = TupleFactory.getInstance().newTuple(0); //invalid size
unionFunc.accumulate(inputTuple); //does nothing, just returns
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size, but null
unionFunc.accumulate(inputTuple); //does nothing, just returns
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //correct type, but empty
unionFunc.accumulate(inputTuple); //does nothing, just returns
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(0); //empty
bag.add(innerTuple);
unionFunc.accumulate(inputTuple); //does nothing, just returns
-
+
inputTuple = TupleFactory.getInstance().newTuple(1); //valid size
bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //correct type
innerTuple = TupleFactory.getInstance().newTuple(1); //correct size
bag.add(innerTuple); //but innerTuple(0) is null
unionFunc.accumulate(inputTuple); //ignores
-
+
Tuple resultTuple = unionFunc.getValue();
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
}
-
+
@Test
public void checkEmptyGetValue() throws IOException {
Accumulator<Tuple> unionFunc = new Union("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple resultTuple = unionFunc.getValue();
DataByteArray dba = (DataByteArray) resultTuple.get(0);
assertEquals(dba.size(), 8);
Double est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
}
-
+
@Test
public void checkExactAccumulator() throws IOException {
Accumulator<Tuple> unionFunc = new Union("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
//create inputTuple and a bag, add bag to inputTuple
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
//create 4 distinct sketches of 32 in a bag
for (int i = 0; i < 4; i++ ) {
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(256, i*64, 64));
-
+
bag.add(dataTuple);
}
-
+
unionFunc.accumulate(inputTuple); //A tuple, bag with 4 sketches
-
+
Tuple resultTuple = unionFunc.getValue();
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray dba = (DataByteArray) resultTuple.get(0);
assertTrue(dba.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 256.0, 0.0);
-
+
unionFunc.cleanup();
-
+
resultTuple = unionFunc.getValue();
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
dba = (DataByteArray) resultTuple.get(0);
assertTrue(dba.size() > 0);
-
+
est = estFunc.exec(resultTuple);
assertEquals(est, 0.0, 0.0);
}
-
+
@Test
public void checkExactAlgebraicInitial() throws IOException {
EvalFunc<Tuple> unionFuncInit = new Union.Initial("256");
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
-
+
for (int i = 0; i < 4; i++ ) { //4 sketches with one value each
Tuple dataTuple = TupleFactory.getInstance().newTuple(1);
dataTuple.set(0, createDbaFromQssRange(16, i, 1));
-
+
bag.add(dataTuple);
}
-
+
Tuple resultTuple = unionFuncInit.exec(inputTuple);
assertTrue(resultTuple == inputTuple); //returns the inputTuple
}
-
+
@Test
public void checkAlgFinalFromPriorIntermed() throws IOException {
EvalFunc<Tuple> unionFuncIFinal = new Union.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
-
+
for (int i = 0; i < 4; i++ ) {
Tuple sketchTuple = TupleFactory.getInstance().newTuple(1);
sketchTuple.set(0, createDbaFromQssRange(64, i*64, 64));
-
- bag.add(sketchTuple);
+
+ bag.add(sketchTuple);
//inputTuple.bag0:sketchTuple0.DBA0
//inputTuple.bag0:sketchTuple1.DBA1
//inputTuple.bag0:sketchTuple2.DBA2
//inputTuple.bag0:sketchTuple3.DBA3
}
-
+
Tuple resultTuple = unionFuncIFinal.exec(inputTuple);
-
+
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
assertTrue(bytes.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 256.0, 0.0);
}
-
+
@Test
public void checkAlgFinalFromPriorInitial() throws IOException {
EvalFunc<Tuple> unionFuncFinal = new Union.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
DataBag innerBag = BagFactory.getInstance().newDefaultBag();
innerTuple.set(0, innerBag); //innerTuple.innerBag0:null
bag.add(innerTuple); //inputTuple.bag0.innerTuple0.innerBag0:null
-
+
for (int i = 0; i < 4; i++ ) {
Tuple sketchTuple = TupleFactory.getInstance().newTuple(1);
sketchTuple.set(0, createDbaFromQssRange(64, i*64, 64));
-
+
innerBag.add(sketchTuple);
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple0.DBA0
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple1.DBA1
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple2.DBA2
//inputTuple.bag0.innerTuple0.innerBag0.sketchTuple3.DBA3
}
-
+
Tuple resultTuple = unionFuncFinal.exec(inputTuple);
-
+
assertNotNull(resultTuple);
assertEquals(resultTuple.size(), 1);
-
+
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
assertTrue(bytes.size() > 0);
-
+
Double est = estFunc.exec(resultTuple);
assertEquals(est, 256.0, 0.0);
}
-
+
@Test
public void checkAlgFinalOuterBagEmptyTuples() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Union.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test
public void checkAlgFinalInnerBagEmpty() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Union.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
DataBag bag2 = BagFactory.getInstance().newDefaultBag();
innerTuple.set(0, bag2);
-
+
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
-
+
@Test(expectedExceptions = IllegalArgumentException.class)
public void checkAlgFinalInnerNotDBA() throws IOException {
EvalFunc<Tuple> interFuncFinal = new Union.IntermediateFinal("256");
EvalFunc<Double> estFunc = new Estimate();
-
+
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
Tuple resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag); //inputTuple.bag0:null
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
-
+
Tuple innerTuple = TupleFactory.getInstance().newTuple(1);
bag.add(innerTuple);
innerTuple.set(0, new Double(1.0)); //not a DBA
-
+
resultTuple = interFuncFinal.exec(inputTuple);
assertEquals(estFunc.exec(resultTuple), 0.0, 0.0);
}
@@ -406,39 +404,39 @@
@Test
public void outputSchemaTest() throws IOException {
EvalFunc<Tuple> udf = new Union("512");
-
+
Schema inputSchema = null;
-
+
Schema nullOutputSchema = null;
-
+
Schema outputSchema = null;
Schema.FieldSchema outputOuterFs0 = null;
-
+
Schema outputInnerSchema = null;
Schema.FieldSchema outputInnerFs0 = null;
-
+
inputSchema = Schema.generateNestedSchema(DataType.BAG, DataType.BYTEARRAY);
-
+
nullOutputSchema = udf.outputSchema(null);
-
+
outputSchema = udf.outputSchema(inputSchema);
outputOuterFs0 = outputSchema.getField(0);
-
+
outputInnerSchema = outputOuterFs0.schema;
outputInnerFs0 = outputInnerSchema.getField(0);
-
+
Assert.assertNull(nullOutputSchema, "Should be null");
Assert.assertNotNull(outputOuterFs0, "outputSchema.getField(0) schema may not be null");
-
+
String expected = "tuple";
String result = DataType.findTypeName(outputOuterFs0.type);
Assert.assertEquals(result, expected);
-
+
expected = "bytearray";
Assert.assertNotNull(outputInnerFs0, "innerSchema.getField(0) schema may not be null");
result = DataType.findTypeName(outputInnerFs0.type);
Assert.assertEquals(result, expected);
-
+
//print schemas
//@formatter:off
StringBuilder sb = new StringBuilder();
@@ -453,17 +451,17 @@
//@formatter:on
//end print schemas
}
-
+
@Test
public void printlnTest() {
println(this.getClass().getSimpleName());
}
-
+
/**
- * @param s value to print
+ * @param s value to print
*/
static void println(String s) {
//System.out.println(s); //disable here
}
-
+
}
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBoundsTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBoundsTest.java
index 9d86884..eb83f9d 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBoundsTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimateAndErrorBoundsTest.java
@@ -29,6 +29,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToEstimateAndErrorBoundsTest {
static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimatesTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimatesTest.java
index ff57445..4119f2d 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimatesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToEstimatesTest.java
@@ -29,7 +29,9 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToEstimatesTest {
+
@Test
public void nullInput() throws Exception {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchToEstimates();
@@ -61,8 +63,12 @@
EvalFunc<Tuple> func = new ArrayOfDoublesSketchToEstimates();
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().build();
int iterations = 100000;
- for (int i = 0; i < iterations; i++) sketch.update(i, new double[] {1});
- for (int i = 0; i < iterations; i++) sketch.update(i, new double[] {1});
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, new double[] {1});
+ }
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, new double[] {1});
+ }
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()));
Tuple resultTuple = func.exec(inputTuple);
Assert.assertNotNull(resultTuple);
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeansTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeansTest.java
index 5100512..6e4b2fa 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeansTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToMeansTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToMeansTest {
@Test
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntriesTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntriesTest.java
index e0a993b..b22bd1a 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntriesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToNumberOfRetainedEntriesTest.java
@@ -28,6 +28,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToNumberOfRetainedEntriesTest {
static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketchTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketchTest.java
index 8089844..bdcd6ed 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToQuantilesSketchTest.java
@@ -32,6 +32,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToQuantilesSketchTest {
static final TupleFactory tupleFactory = TupleFactory.getInstance();
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariancesTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariancesTest.java
index 2772eaa..ca19415 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariancesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchToVariancesTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class ArrayOfDoublesSketchToVariancesTest {
@Test
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimatesTest.java b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimatesTest.java
index d62b791..ac2854c 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimatesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/ArrayOfDoublesSketchesToPValueEstimatesTest.java
@@ -31,6 +31,7 @@
import org.apache.commons.math3.stat.inference.TTest;
+import java.io.IOException;
import java.util.Random;
/**
@@ -39,10 +40,10 @@
public class ArrayOfDoublesSketchesToPValueEstimatesTest {
/**
* Check null input to UDF.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void nullInput() throws Exception {
+ public void nullInput() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
Tuple resultTuple = func.exec(null);
@@ -52,10 +53,10 @@
/**
* Check input of empty tuple.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void emptyInput() throws Exception {
+ public void emptyInput() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
Tuple resultTuple = func.exec(TupleFactory.getInstance().newTuple());
@@ -65,10 +66,10 @@
/**
* Check input of single empty sketch.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void oneEmptySketch() throws Exception {
+ public void oneEmptySketch() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().build();
@@ -82,10 +83,10 @@
/**
* Check input of two empty sketches.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void twoEmptySketches() throws Exception {
+ public void twoEmptySketches() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
ArrayOfDoublesUpdatableSketch sketchA = new ArrayOfDoublesUpdatableSketchBuilder().build();
@@ -101,10 +102,10 @@
/**
* Check p-value for the smoker data set. Single metric.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void smokerDatasetSingleMetric() throws Exception {
+ public void smokerDatasetSingleMetric() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
// Create the two sketches
@@ -146,10 +147,10 @@
/**
* Check p-value for a large data set.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void largeDataSet() throws Exception {
+ public void largeDataSet() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
// Create the two sketches
@@ -203,10 +204,10 @@
/**
* Check p-value for two metrics at the same time.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void twoMetrics() throws Exception {
+ public void twoMetrics() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
// Create the two sketches
@@ -249,10 +250,10 @@
/**
* Check with sketch having only one input.
- * @throws Exception
+ * @throws IOException from EvalFunc<Tuple>.exec(...)
*/
@Test
- public void sketchWithSingleValue() throws Exception {
+ public void sketchWithSingleValue() throws IOException {
EvalFunc<Tuple> func = new ArrayOfDoublesSketchesToPValueEstimates();
// Create the two sketches
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchTest.java
index a0629d8..65f397a 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/DataToArrayOfDoublesSketchTest.java
@@ -35,6 +35,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class DataToArrayOfDoublesSketchTest {
@Test
@@ -117,7 +118,9 @@
EvalFunc<Tuple> func = new DataToArrayOfDoublesSketch("1024", "0.5", "1");
DataBag bag = BagFactory.getInstance().newDefaultBag();
int uniques = 10000;
- for (int i = 0; i < uniques; i++) bag.add(PigUtil.objectsToTuple(i, 1.0));
+ for (int i = 0; i < uniques; i++) {
+ bag.add(PigUtil.objectsToTuple(i, 1.0));
+ }
Tuple resultTuple = func.exec(PigUtil.objectsToTuple(bag));
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 1);
@@ -225,7 +228,9 @@
DataBag bag = BagFactory.getInstance().newDefaultBag();
int uniques = 10000;
- for (int i = 0; i < uniques; i++) bag.add(PigUtil.objectsToTuple(i, 1.0));
+ for (int i = 0; i < uniques; i++) {
+ bag.add(PigUtil.objectsToTuple(i, 1.0));
+ }
Tuple resultTuple = func.exec(PigUtil.objectsToTuple(PigUtil.tuplesToBag(PigUtil.objectsToTuple(bag))));
Assert.assertNotNull(resultTuple);
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketchTest.java b/src/test/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketchTest.java
index 36a9e1f..bfbff78 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/DataToDoubleSummarySketchTest.java
@@ -39,6 +39,7 @@
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummaryFactory;
+@SuppressWarnings("javadoc")
public class DataToDoubleSummarySketchTest {
@Test
@@ -64,7 +65,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 0.0);
}
@@ -166,7 +168,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
SketchIterator<DoubleSummary> it = sketch.iterator();
@@ -181,7 +184,8 @@
Assert.assertEquals(resultTuple.size(), 1);
bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch2 = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch2 =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch2.getEstimate(), 0.0, 0.0);
}
@@ -230,7 +234,8 @@
bag.add(PigUtil.objectsToTuple(PigUtil.tuplesToBag(PigUtil.objectsToTuple("a", 1.0))));
// this is to simulate the output from a prior call of IntermediateFinal
- UpdatableSketch<Double, DoubleSummary> ts = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> ts =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
ts.update("b", 1.0);
ts.update("a", 2.0);
ts.update("b", 2.0);
@@ -242,7 +247,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
SketchIterator<DoubleSummary> it = sketch.iterator();
@@ -269,7 +275,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
SketchIterator<DoubleSummary> it = sketch.iterator();
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimatesTest.java b/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimatesTest.java
index 3c82f72..89bf2df 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimatesTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToEstimatesTest.java
@@ -31,6 +31,7 @@
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryFactory;
+@SuppressWarnings("javadoc")
public class DoubleSummarySketchToEstimatesTest {
@Test
public void nullInput() throws Exception {
@@ -49,7 +50,8 @@
@Test
public void emptySketch() throws Exception {
EvalFunc<Tuple> func = new DoubleSummarySketchToEstimates();
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()));
Tuple resultTuple = func.exec(inputTuple);
Assert.assertNotNull(resultTuple);
@@ -61,10 +63,15 @@
@Test
public void normalCase() throws Exception {
EvalFunc<Tuple> func = new DoubleSummarySketchToEstimates();
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
int iterations = 100000;
- for (int i = 0; i < iterations; i++) sketch.update(i, 1.0);
- for (int i = 0; i < iterations; i++) sketch.update(i, 1.0);
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, 1.0);
+ }
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, 1.0);
+ }
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()));
Tuple resultTuple = func.exec(inputTuple);
Assert.assertNotNull(resultTuple);
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentileTest.java b/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentileTest.java
index a94585d..d9414c1 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentileTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/DoubleSummarySketchToPercentileTest.java
@@ -34,12 +34,16 @@
import org.apache.datasketches.tuple.adouble.DoubleSummary;
import org.apache.datasketches.tuple.adouble.DoubleSummaryFactory;
+@SuppressWarnings("javadoc")
public class DoubleSummarySketchToPercentileTest {
+
@Test
public void emptySketch() throws Exception {
EvalFunc<Double> func = new DoubleSummarySketchToPercentile();
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
- Tuple inputTuple = TupleFactory.getInstance().newTuple(Arrays.asList(new DataByteArray(sketch.compact().toByteArray()), 0.0));
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
+ Tuple inputTuple = TupleFactory.getInstance().
+ newTuple(Arrays.asList(new DataByteArray(sketch.compact().toByteArray()), 0.0));
double result = func.exec(inputTuple);
Assert.assertEquals(result, Double.NaN);
}
@@ -47,10 +51,15 @@
@Test
public void normalCase() throws Exception {
EvalFunc<Double> func = new DoubleSummarySketchToPercentile();
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
int iterations = 100000;
- for (int i = 0; i < iterations; i++) sketch.update(i, (double) i);
- for (int i = 0; i < iterations; i++) sketch.update(i, (double) i);
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, (double) i);
+ }
+ for (int i = 0; i < iterations; i++) {
+ sketch.update(i, (double) i);
+ }
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()), 50.0);
double result = func.exec(inputTuple);
Assert.assertEquals(result, iterations, iterations * 0.02);
@@ -65,7 +74,8 @@
@Test(expectedExceptions = IllegalArgumentException.class)
public void percentileOutOfRange() throws Exception {
EvalFunc<Double> func = new DoubleSummarySketchToPercentile();
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
func.exec(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()), 200.0));
}
}
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/PigUtil.java b/src/test/java/org/apache/datasketches/pig/tuple/PigUtil.java
index efab328..045613b 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/PigUtil.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/PigUtil.java
@@ -26,6 +26,7 @@
import org.apache.pig.data.Tuple;
import org.apache.pig.data.TupleFactory;
+@SuppressWarnings("javadoc")
public class PigUtil {
/**
* Wraps input Objects in a Tuple
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchTest.java b/src/test/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchTest.java
index 4f4a0ae..ca93a98 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/UnionArrayOfDoublesSketchTest.java
@@ -37,6 +37,7 @@
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketch;
import org.apache.datasketches.tuple.ArrayOfDoublesUpdatableSketchBuilder;
+@SuppressWarnings("javadoc")
public class UnionArrayOfDoublesSketchTest {
@Test
@@ -268,7 +269,9 @@
// this is to simulate the output from Initial
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 20000; i++) sketch.update(value++, new double[] {1.0});
+ for (int i = 0; i < 20000; i++) {
+ sketch.update(value++, new double[] {1.0});
+ }
DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
bag.add(PigUtil.objectsToTuple(innerBag));
}
@@ -276,7 +279,9 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 20000; i++) sketch.update(value++, new double[] {1.0});
+ for (int i = 0; i < 20000; i++) {
+ sketch.update(value++, new double[] {1.0});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
@@ -301,7 +306,9 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(1024).build();
- for (int i = 0; i < 10000; i++) sketch.update(value++, new double[] {1.0});
+ for (int i = 0; i < 10000; i++) {
+ sketch.update(value++, new double[] {1.0});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
@@ -329,7 +336,9 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
@@ -337,7 +346,9 @@
key -= 20000; // overlap
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
@@ -345,7 +356,9 @@
key -= 20000; // overlap
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
@@ -353,21 +366,27 @@
key -= 20000; // overlap
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
}
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
}
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
@@ -375,7 +394,9 @@
key -= 20000; // overlap
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, new double[] {rnd.nextDouble() * 20});
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
diff --git a/src/test/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketchTest.java b/src/test/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketchTest.java
index 550a4aa..2d05b17 100644
--- a/src/test/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketchTest.java
+++ b/src/test/java/org/apache/datasketches/pig/tuple/UnionDoubleSummarySketchTest.java
@@ -41,6 +41,7 @@
import org.apache.datasketches.tuple.adouble.DoubleSummaryDeserializer;
import org.apache.datasketches.tuple.adouble.DoubleSummaryFactory;
+@SuppressWarnings("javadoc")
public class UnionDoubleSummarySketchTest {
@Test
@@ -62,13 +63,15 @@
EvalFunc<Tuple> func = new UnionDoubleSummarySketch("4096");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -78,7 +81,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
SketchIterator<DoubleSummary> it = sketch.iterator();
while (it.next()) {
@@ -91,13 +95,15 @@
EvalFunc<Tuple> func = new UnionDoubleSummarySketch("4096", "Max");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 3.0);
sketch.update(2, 3.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -198,7 +204,8 @@
Accumulator<Tuple> func = new UnionDoubleSummarySketch("4096");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
func.accumulate(PigUtil.objectsToTuple(bag));
@@ -207,7 +214,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 0.0);
}
@@ -216,7 +224,8 @@
Accumulator<Tuple> func = new UnionDoubleSummarySketch("4096");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -225,7 +234,8 @@
bag = BagFactory.getInstance().newDefaultBag();
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -237,7 +247,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
SketchIterator<DoubleSummary> it = sketch.iterator();
while (it.next()) {
@@ -267,7 +278,8 @@
// this is to simulate the output from Initial
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 1.0);
sketch.update(2, 1.0);
DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -276,7 +288,8 @@
// this is to simulate the output from a prior call of IntermediateFinal
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).build();
sketch.update(1, 3.0);
sketch.update(2, 3.0);
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
@@ -303,16 +316,24 @@
long value = 1;
// this is to simulate the output from Initial
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 20000; i++) sketch.update(value++, 1.0);
- DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 20000; i++) {
+ sketch.update(value++, 1.0);
+ }
+ DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(
+ new DataByteArray(sketch.compact().toByteArray())));
bag.add(PigUtil.objectsToTuple(innerBag));
}
// this is to simulate the output from a prior call of IntermediateFinal
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 20000; i++) sketch.update(value++, 1.0);
+ UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 20000; i++) {
+ sketch.update(value++, 1.0);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
@@ -321,7 +342,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), 40000.0, 40000.0 * 0.01);
SketchIterator<DoubleSummary> it = sketch.iterator();
while (it.next()) {
@@ -337,8 +359,12 @@
long value = 1;
// this is to simulate the output from a prior call of IntermediateFinal
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(1024).build();
- for (int i = 0; i < 10000; i++) sketch.update(value++, 1.0);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(1024).build();
+ for (int i = 0; i < 10000; i++) {
+ sketch.update(value++, 1.0);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
@@ -366,54 +392,81 @@
long updates = 0;
// this is to simulate the output from a prior call of IntermediateFinal
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
}
key -= 20000; // overlap
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
}
key -= 20000; // overlap
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
}
key -= 20000; // overlap
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
}
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
}
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 40000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 40000;
}
key -= 20000; // overlap
{
- UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
- for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
+ UpdatableSketch<Double, DoubleSummary> sketch =
+ new UpdatableSketchBuilder<>(
+ new DoubleSummaryFactory(DoubleSummary.Mode.Sum)).setNominalEntries(16384).build();
+ for (int i = 0; i < 60000; i++) {
+ sketch.update(key++, rnd.nextDouble() * 20);
+ }
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
uniques += 40000;
updates += 60000;
@@ -424,7 +477,8 @@
Assert.assertEquals(resultTuple.size(), 1);
DataByteArray bytes = (DataByteArray) resultTuple.get(0);
Assert.assertTrue(bytes.size() > 0);
- Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
+ Sketch<DoubleSummary> sketch =
+ Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
Assert.assertEquals(sketch.getEstimate(), uniques, uniques * 0.01);
double sum = 0;
SketchIterator<DoubleSummary> it = sketch.iterator();
@@ -432,6 +486,7 @@
sum += it.getSummary().getValue();
}
// each update added 10 to the total on average
- Assert.assertEquals(sum / sketch.getTheta(), updates * 10.0, updates * 10.0 * 0.02); // there is a slight chance of failing here
+ // there is a slight chance of failing here
+ Assert.assertEquals(sum / sketch.getTheta(), updates * 10.0, updates * 10.0 * 0.02);
}
}
