Merge branch '2.0.X'

Author: leerho

pom.xml

@@ -32,7 +32,7 @@ under the License.
 
   <groupId>org.apache.datasketches</groupId>
   <artifactId>datasketches-java</artifactId>
-  <version>1.4.0-SNAPSHOT</version>
+  <version>2.0.0</version>
   <packaging>jar</packaging>
 
   <name>${project.artifactId}</name>

src/main/java/org/apache/datasketches/theta/AnotB.java

@@ -108,20 +108,21 @@ public Family getFamily() {
   public abstract void notB(Sketch skB);
 
   /**
-   * Gets the result of the mutistep, stateful operation AnotB that have been executed with calls
+   * Gets the result of the multistep, stateful operation AnotB that have been executed with calls
    * to {@link #setA(Sketch)} and ({@link #notB(Sketch)} or
    * {@link #notB(org.apache.datasketches.theta.Sketch)}).
    *
    * @param reset If <i>true</i>, clears this operator to the empty state after this result is
    * returned. Set this to <i>false</i> if you wish to obtain an intermediate result.
+   *
    * @return the result of this operation as an ordered, on-heap {@link CompactSketch}.
    */
   public abstract CompactSketch getResult(boolean reset);
 
   /**
-   * Gets the result of this stateful set operation as a CompactSketch of the form based on
-   * the input arguments.
-   * The stateful input operations are {@link #setA(Sketch)} and {@link #notB(Sketch)}.
+   * Gets the result of the multistep, stateful operation AnotB that have been executed with calls
+   * to {@link #setA(Sketch)} and ({@link #notB(Sketch)} or
+   * {@link #notB(org.apache.datasketches.theta.Sketch)}).
    *
    * @param dstOrdered If <i>true</i>, the result will be an ordered {@link CompactSketch}.
    * <a href="{@docRoot}/resources/dictionary.html#dstOrdered">See Destination Ordered</a>.
@@ -132,7 +133,7 @@ public Family getFamily() {
    * @param reset If <i>true</i>, clears this operator to the empty state after this result is
    * returned. Set this to <i>false</i> if you wish to obtain an intermediate result.
    *
-   * @return the result of this operation as a {@link CompactSketch} of the chosen form.
+   * @return the result of this operation as a {@link CompactSketch} in the given dstMem.
    */
   public abstract CompactSketch getResult(boolean dstOrdered, WritableMemory dstMem, boolean reset);
 

src/main/java/org/apache/datasketches/theta/Intersection.java

@@ -50,15 +50,19 @@ public Family getFamily() {
 
   /**
    * Gets the result of this operation as an ordered CompactSketch on the Java heap.
-   * The {@link #intersect(Sketch)} method must have been called at least once.
+   * This does not disturb the underlying data structure of this intersection.
+   * The {@link #intersect(Sketch)} method must have been called at least once, otherwise an
+   * exception will be thrown. This is because a virgin Intersection object represents the
+   * Universal Set, which has an infinite number of values.
    * @return the result of this operation as an ordered CompactSketch on the Java heap
    */
   public CompactSketch getResult() {
     return getResult(true, null);
   }
 
   /**
-   * Gets the result of this operation as a CompactSketch of the chosen form.
+   * Gets the result of this operation as a CompactSketch in the given dstMem.
+   * This does not disturb the underlying data structure of this intersection.
    * The {@link #intersect(Sketch)} method must have been called at least once, otherwise an
    * exception will be thrown. This is because a virgin Intersection object represents the
    * Universal Set, which has an infinite number of values.
@@ -70,14 +74,14 @@ public CompactSketch getResult() {
    *
    * <p>Presenting an intersection with a null argument will throw an exception.</p>
    *
-   *
    * @param dstOrdered
    * <a href="{@docRoot}/resources/dictionary.html#dstOrdered">See Destination Ordered</a>
    *
    * @param dstMem
    * <a href="{@docRoot}/resources/dictionary.html#dstMem">See Destination Memory</a>.
    *
-   * @return the result of this operation as a CompactSketch of the chosen form
+   * @return the result of this operation as a CompactSketch stored in the given dstMem,
+   * which can be either on or off-heap..
    */
   public abstract CompactSketch getResult(boolean dstOrdered, WritableMemory dstMem);
 
@@ -90,7 +94,7 @@ public CompactSketch getResult() {
   /**
    * Resets this Intersection for stateful operations only.
    * The seed remains intact, otherwise reverts to
-   * the Universal Set, theta of 1.0 and empty = false.
+   * the Universal Set: theta = 1.0, no retained data and empty = false.
    */
   public abstract void reset();
 

src/main/java/org/apache/datasketches/theta/IntersectionImpl.java

@@ -219,7 +219,7 @@ static IntersectionImpl wrapInstance(
   @Override
   public CompactSketch intersect(final Sketch a, final Sketch b, final boolean dstOrdered,
      final WritableMemory dstMem) {
-    if ((wmem_ != null) && readOnly_) { throw new SketchesReadOnlyException(); }
+    if (wmem_ != null && readOnly_) { throw new SketchesReadOnlyException(); }
     hardReset();
     intersect(a);
     intersect(b);
@@ -233,7 +233,7 @@ public void intersect(final Sketch sketchIn) {
     if (sketchIn == null) {
       throw new SketchesArgumentException("Intersection argument must not be null.");
     }
-    if ((wmem_ != null) && readOnly_) { throw new SketchesReadOnlyException(); }
+    if (wmem_ != null && readOnly_) { throw new SketchesReadOnlyException(); }
     if (empty_ || sketchIn.isEmpty()) { //empty rule
       //Because of the def of null above and the Empty Rule (which is OR), empty_ must be true.
       //Whatever the current internal state, we make our local empty.
@@ -262,14 +262,14 @@ public void intersect(final Sketch sketchIn) {
     final int sketchInEntries = sketchIn.getRetainedEntries(true);
 
     //states 1,2,3,6
-    if ((curCount_ == 0) || (sketchInEntries == 0)) {
+    if (curCount_ == 0 || sketchInEntries == 0) {
       curCount_ = 0;
       if (wmem_ != null) { insertCurCount(wmem_, 0); }
       hashTable_ = null; //No need for a HT. Don't bother clearing mem if valid
     } //end of states 1,2,3,6
 
     // state 5
-    else if ((curCount_ < 0) && (sketchInEntries > 0)) {
+    else if (curCount_ < 0 && sketchInEntries > 0) {
       curCount_ = sketchIn.getRetainedEntries(true);
       final int requiredLgArrLongs = minLgHashTableSize(curCount_, REBUILD_THRESHOLD);
       final int priorLgArrLongs = lgArrLongs_; //prior only used in error message
@@ -295,7 +295,7 @@ else if ((curCount_ < 0) && (sketchInEntries > 0)) {
     } //end of state 5
 
     //state 7
-    else if ((curCount_ > 0) && (sketchInEntries > 0)) {
+    else if (curCount_ > 0 && sketchInEntries > 0) {
       //Sets resulting hashTable, curCount and adjusts lgArrLongs
       performIntersect(sketchIn);
     } //end of state 7
@@ -339,6 +339,16 @@ public CompactSketch getResult(final boolean dstOrdered, final WritableMemory ds
         dstMem, compactCache);
   }
 
+  @Override
+  public boolean hasResult() {
+    return wmem_ != null ? wmem_.getInt(RETAINED_ENTRIES_INT) >= 0 : curCount_ >= 0;
+  }
+
+  @Override
+  public boolean isSameResource(final Memory that) {
+    return wmem_ != null ? wmem_.isSameResource(that) : false;
+  }
+
   @Override
   public void reset() {
     hardReset();
@@ -347,7 +357,7 @@ public void reset() {
   @Override
   public byte[] toByteArray() {
     final int preBytes = CONST_PREAMBLE_LONGS << 3;
-    final int dataBytes = (curCount_ > 0) ? 8 << lgArrLongs_ : 0;
+    final int dataBytes = curCount_ > 0 ? 8 << lgArrLongs_ : 0;
     final byte[] byteArrOut = new byte[preBytes + dataBytes];
     if (wmem_ != null) {
       wmem_.getByteArray(0, byteArrOut, 0, preBytes + dataBytes);
@@ -376,16 +386,6 @@ public byte[] toByteArray() {
     return byteArrOut;
   }
 
-  @Override
-  public boolean hasResult() {
-    return (wmem_ != null) ? wmem_.getInt(RETAINED_ENTRIES_INT) >= 0 : curCount_ >= 0;
-  }
-
-  @Override
-  public boolean isSameResource(final Memory that) {
-    return (wmem_ != null) ? wmem_.isSameResource(that) : false;
-  }
-
   //restricted
 
   /**
@@ -405,7 +405,7 @@ boolean isEmpty() {
   @Override
   long[] getCache() {
     if (wmem_ == null) {
-      return (hashTable_ != null) ? hashTable_ : new long[0];
+      return hashTable_ != null ? hashTable_ : new long[0];
     }
     //Direct
     final int arrLongs = 1 << lgArrLongs_;
@@ -426,7 +426,7 @@ long getThetaLong() {
 
   private void performIntersect(final Sketch sketchIn) {
     // curCount and input data are nonzero, match against HT
-    assert ((curCount_ > 0) && (!empty_));
+    assert curCount_ > 0 && !empty_;
     final long[] cacheIn = sketchIn.getCache();
     final int arrLongsIn = cacheIn.length;
     final long[] hashTable;
@@ -458,7 +458,7 @@ private void performIntersect(final Sketch sketchIn) {
       //either unordered compact or hash table
       for (int i = 0; i < arrLongsIn; i++ ) {
         final long hashIn = cacheIn[i];
-        if ((hashIn <= 0L) || (hashIn >= thetaLong_)) { continue; }
+        if (hashIn <= 0L || hashIn >= thetaLong_) { continue; }
         final int foundIdx = hashSearch(hashTable, lgArrLongs_, hashIn);
         if (foundIdx == -1) { continue; }
         matchSet[matchSetCount++] = hashIn;
@@ -505,7 +505,7 @@ private void moveDataToTgt(final long[] arr, final int count) {
         tmpCnt++;
       }
     }
-    assert (tmpCnt == count) : "Intersection Count Check: got: " + tmpCnt + ", expected: " + count;
+    assert tmpCnt == count : "Intersection Count Check: got: " + tmpCnt + ", expected: " + count;
   }
 
   private void hardReset() {

src/main/java/org/apache/datasketches/theta/JaccardSimilarity.java

@@ -54,7 +54,7 @@ public final class JaccardSimilarity {
    */
   public static double[] jaccard(final Sketch sketchA, final Sketch sketchB) {
     //Corner case checks
-    if ((sketchA == null) || (sketchB == null)) { return ZEROS.clone(); }
+    if (sketchA == null || sketchB == null) { return ZEROS.clone(); }
     if (sketchA == sketchB) { return ONES.clone(); }
     if (sketchA.isEmpty() && sketchB.isEmpty()) { return ONES.clone(); }
     if (sketchA.isEmpty() || sketchB.isEmpty()) { return ZEROS.clone(); }
@@ -68,17 +68,17 @@ public static double[] jaccard(final Sketch sketchA, final Sketch sketchB) {
     final int newK = max(min(ceilingPowerOf2(countA + countB), maxK), minK);
     final Union union =
         SetOperation.builder().setNominalEntries(newK).buildUnion();
-    union.update(sketchA);
-    union.update(sketchB);
+    union.union(sketchA);
+    union.union(sketchB);
     final Sketch unionAB = union.getResult(false, null);
     final long thetaLongUAB = unionAB.getThetaLong();
     final long thetaLongA = sketchA.getThetaLong();
     final long thetaLongB = sketchB.getThetaLong();
     final int countUAB = unionAB.getRetainedEntries(true);
 
     //Check for identical data
-    if ((countUAB == countA) && (countUAB == countB)
-        && (thetaLongUAB == thetaLongA) && (thetaLongUAB == thetaLongB)) {
+    if (countUAB == countA && countUAB == countB
+        && thetaLongUAB == thetaLongA && thetaLongUAB == thetaLongB) {
       return ONES.clone();
     }
 
@@ -105,7 +105,7 @@ public static double[] jaccard(final Sketch sketchA, final Sketch sketchB) {
    */
   public static boolean exactlyEqual(final Sketch sketchA, final Sketch sketchB) {
     //Corner case checks
-    if ((sketchA == null) || (sketchB == null)) { return false; }
+    if (sketchA == null || sketchB == null) { return false; }
     if (sketchA == sketchB) { return true; }
     if (sketchA.isEmpty() && sketchB.isEmpty()) { return true; }
     if (sketchA.isEmpty() || sketchB.isEmpty()) { return false; }
@@ -116,17 +116,17 @@ public static boolean exactlyEqual(final Sketch sketchA, final Sketch sketchB) {
     //Create the Union
     final Union union =
         SetOperation.builder().setNominalEntries(ceilingPowerOf2(countA + countB)).buildUnion();
-    union.update(sketchA);
-    union.update(sketchB);
+    union.union(sketchA);
+    union.union(sketchB);
     final Sketch unionAB = union.getResult();
     final long thetaLongUAB = unionAB.getThetaLong();
     final long thetaLongA = sketchA.getThetaLong();
     final long thetaLongB = sketchB.getThetaLong();
     final int countUAB = unionAB.getRetainedEntries(true);
 
     //Check for identical counts and thetas
-    if ((countUAB == countA) && (countUAB == countB)
-        && (thetaLongUAB == thetaLongA) && (thetaLongUAB == thetaLongB)) {
+    if (countUAB == countA && countUAB == countB
+        && thetaLongUAB == thetaLongA && thetaLongUAB == thetaLongB) {
       return true;
     }
     return false;

src/main/java/org/apache/datasketches/theta/Union.java

@@ -35,6 +35,14 @@ public Family getFamily() {
     return Family.UNION;
   }
 
+  /**
+   * Gets the result of this operation as an ordered CompactSketch on the Java heap.
+   * This does not disturb the underlying data structure of the union.
+   * Therefore, it is OK to continue updating the union after this operation.
+   * @return the result of this operation as an ordered CompactSketch on the Java heap
+   */
+  public abstract CompactSketch getResult();
+
   /**
    * Gets the result of this operation as a CompactSketch of the chosen form.
    * This does not disturb the underlying data structure of the union.
@@ -50,14 +58,6 @@ public Family getFamily() {
    */
   public abstract CompactSketch getResult(boolean dstOrdered, WritableMemory dstMem);
 
-  /**
-   * Gets the result of this operation as an ordered CompactSketch on the Java heap.
-   * This does not disturb the underlying data structure of the union.
-   * Therefore, it is OK to continue updating the union after this operation.
-   * @return the result of this operation as an ordered CompactSketch on the Java heap
-   */
-  public abstract CompactSketch getResult();
-
   /**
    * Resets this Union. The seed remains intact, otherwise reverts back to its virgin state.
    */
@@ -108,6 +108,20 @@ public abstract CompactSketch union(Sketch sketchA, Sketch sketchB, boolean dstO
    *
    * @param sketchIn The incoming sketch.
    */
+  public abstract void union(Sketch sketchIn);
+
+
+  /**
+   * Perform a Union operation with <i>this</i> union and the given on-heap sketch of the Theta Family.
+   * This method is not valid for the older SetSketch, which was prior to Open Source (August, 2015).
+   *
+   * <p>This method can be repeatedly called.
+   * If the given sketch is null it is interpreted as an empty sketch.</p>
+   *
+   * @param sketchIn The incoming sketch.
+   * @deprecated 2.0.0.  Use {@link #union(Sketch)} instead.
+   */
+  @Deprecated
   public abstract void update(Sketch sketchIn);
 
   /**
@@ -120,6 +134,20 @@ public abstract CompactSketch union(Sketch sketchA, Sketch sketchB, boolean dstO
    *
    * @param mem Memory image of sketch to be merged
    */
+  public abstract void union(Memory mem);
+
+  /**
+   * Perform a Union operation with <i>this</i> union and the given Memory image of any sketch of the
+   * Theta Family. The input image may be from earlier versions of the Theta Compact Sketch,
+   * called the SetSketch (circa 2012), which was prior to Open Source and are compact and ordered.
+   *
+   * <p>This method can be repeatedly called.
+   * If the given sketch is null it is interpreted as an empty sketch.</p>
+   *
+   * @param mem Memory image of sketch to be merged
+   * @deprecated 2.0.0. Use {@link #union(Memory)} instead.
+   */
+  @Deprecated
   public abstract void update(Memory mem);
 
   /**