Merge pull request #144 from DataSketches/tuple-reset

added reset method

sketches/src/main/java/com/yahoo/sketches/tuple/ArrayOfDoublesUpdatableSketch.java

@@ -100,6 +100,11 @@ public void update(final long[] key, final double[] values) {
    */
   public abstract void trim();
 
+  /**
+   * Resets this sketch an empty state.
+   */
+  public abstract void reset();
+
   /**
    * Gets an on-heap compact representation of the sketch
    * @return compact sketch

sketches/src/main/java/com/yahoo/sketches/tuple/DirectArrayOfDoublesQuickSelectSketch.java

@@ -5,18 +5,13 @@
 
 package com.yahoo.sketches.tuple;
 
-import static com.yahoo.sketches.Util.MIN_LG_ARR_LONGS;
-import static com.yahoo.sketches.Util.ceilingPowerOf2;
-import static com.yahoo.sketches.Util.startingSubMultiple;
-
 import java.nio.ByteOrder;
 import java.util.Arrays;
 
 import com.yahoo.memory.Memory;
 import com.yahoo.memory.NativeMemory;
 import com.yahoo.sketches.Family;
 import com.yahoo.sketches.HashOperations;
-import com.yahoo.sketches.ResizeFactor;
 import com.yahoo.sketches.SketchesArgumentException;
 
 /**
@@ -53,12 +48,7 @@ final class DirectArrayOfDoublesQuickSelectSketch extends ArrayOfDoublesQuickSel
       final float samplingProbability, final int numValues, final long seed, final Memory dstMem) {
     super(numValues, seed);
     mem_ = dstMem;
-    final int startingCapacity = 1 << startingSubMultiple(
-      // target table size is twice the number of nominal entries
-      Integer.numberOfTrailingZeros(ceilingPowerOf2(nomEntries) * 2),
-      ResizeFactor.getRF(lgResizeFactor),
-      MIN_LG_ARR_LONGS
-    );
+    final int startingCapacity = Util.getStartingCapacity(nomEntries, lgResizeFactor);
     checkIfEnoughMemory(dstMem, startingCapacity, numValues);
     mem_.putByte(PREAMBLE_LONGS_BYTE, (byte) 1);
     mem_.putByte(SERIAL_VERSION_BYTE, serialVersionUID);
@@ -159,6 +149,26 @@ public byte[] toByteArray() {
     return byteArray;
   }
 
+  @Override
+  public void reset() {
+    if (!isEmpty_) {
+      isEmpty_ = true;
+      mem_.setBits(FLAGS_BYTE, (byte) (1 << Flags.IS_EMPTY.ordinal()));
+    }
+    final int lgResizeFactor = mem_.getByte(LG_RESIZE_FACTOR_BYTE);
+    final float samplingProbability = mem_.getFloat(SAMPLING_P_FLOAT);
+    final int startingCapacity = Util.getStartingCapacity(getNominalEntries(), lgResizeFactor);
+    theta_ = (long) (Long.MAX_VALUE * (double) samplingProbability);
+    mem_.putLong(THETA_LONG, theta_);
+    mem_.putByte(LG_CUR_CAPACITY_BYTE, (byte) Integer.numberOfTrailingZeros(startingCapacity));
+    mem_.putInt(RETAINED_ENTRIES_INT, 0);
+    keysOffset_ = ENTRIES_START;
+    valuesOffset_ = keysOffset_ + SIZE_OF_KEY_BYTES * startingCapacity;
+    mem_.clear(keysOffset_, SIZE_OF_KEY_BYTES * startingCapacity); // clear keys only
+    lgCurrentCapacity_ = Integer.numberOfTrailingZeros(startingCapacity);
+    setRebuildThreshold();
+  }
+
   @Override
   protected long getKey(final int index) {
     return mem_.getLong(keysOffset_ + SIZE_OF_KEY_BYTES * index);

sketches/src/main/java/com/yahoo/sketches/tuple/HeapArrayOfDoublesQuickSelectSketch.java

@@ -5,9 +5,7 @@
 
 package com.yahoo.sketches.tuple;
 
-import static com.yahoo.sketches.Util.MIN_LG_ARR_LONGS;
 import static com.yahoo.sketches.Util.ceilingPowerOf2;
-import static com.yahoo.sketches.Util.startingSubMultiple;
 
 import java.nio.ByteOrder;
 import java.util.Arrays;
@@ -16,7 +14,6 @@
 import com.yahoo.memory.NativeMemory;
 import com.yahoo.sketches.Family;
 import com.yahoo.sketches.HashOperations;
-import com.yahoo.sketches.ResizeFactor;
 import com.yahoo.sketches.SketchesArgumentException;
 
 /**
@@ -55,12 +52,7 @@ final class HeapArrayOfDoublesQuickSelectSketch extends ArrayOfDoublesQuickSelec
     lgResizeFactor_ = lgResizeFactor;
     samplingProbability_ = samplingProbability;
     theta_ = (long) (Long.MAX_VALUE * (double) samplingProbability);
-    final int startingCapacity = 1 << startingSubMultiple(
-      // target table size is twice the number of nominal entries
-      Integer.numberOfTrailingZeros(ceilingPowerOf2(nomEntries) * 2),
-      ResizeFactor.getRF(lgResizeFactor),
-      MIN_LG_ARR_LONGS
-    );
+    final int startingCapacity = Util.getStartingCapacity(nomEntries, lgResizeFactor);
     keys_ = new long[startingCapacity];
     values_ = new double[startingCapacity * numValues];
     lgCurrentCapacity_ = Integer.numberOfTrailingZeros(startingCapacity);
@@ -167,6 +159,18 @@ public byte[] toByteArray() {
     return byteArray;
   }
 
+  @Override
+  public void reset() {
+    isEmpty_ = true;
+    count_ = 0;
+    theta_ = (long) (Long.MAX_VALUE * (double) samplingProbability_);
+    final int startingCapacity = Util.getStartingCapacity(nomEntries_, lgResizeFactor_);
+    keys_ = new long[startingCapacity];
+    values_ = new double[startingCapacity * numValues_];
+    lgCurrentCapacity_ = Integer.numberOfTrailingZeros(startingCapacity);
+    setRebuildThreshold();
+  }
+
   @Override
   protected long getKey(final int index) {
     return keys_[index];

sketches/src/main/java/com/yahoo/sketches/tuple/QuickSelectSketch.java

@@ -5,11 +5,9 @@
 
 package com.yahoo.sketches.tuple;
 
-import static com.yahoo.sketches.Util.MIN_LG_ARR_LONGS;
 import static com.yahoo.sketches.Util.REBUILD_THRESHOLD;
 import static com.yahoo.sketches.Util.RESIZE_THRESHOLD;
 import static com.yahoo.sketches.Util.ceilingPowerOf2;
-import static com.yahoo.sketches.Util.startingSubMultiple;
 
 import java.lang.reflect.Array;
 import java.nio.ByteOrder;
@@ -20,7 +18,6 @@
 import com.yahoo.sketches.Family;
 import com.yahoo.sketches.HashOperations;
 import com.yahoo.sketches.QuickSelect;
-import com.yahoo.sketches.ResizeFactor;
 import com.yahoo.sketches.SketchesArgumentException;
 
 /**
@@ -93,12 +90,7 @@ private enum Flags { IS_BIG_ENDIAN, IS_IN_SAMPLING_MODE, IS_EMPTY, HAS_ENTRIES,
       lgResizeFactor,
       samplingProbability,
       summaryFactory,
-      1 << startingSubMultiple(
-        // target table size is twice the number of nominal entries
-        Integer.numberOfTrailingZeros(ceilingPowerOf2(nomEntries) * 2),
-        ResizeFactor.getRF(lgResizeFactor),
-        MIN_LG_ARR_LONGS
-      )
+      Util.getStartingCapacity(nomEntries, lgResizeFactor)
     );
   }
 
@@ -211,6 +203,21 @@ public void trim() {
     }
   }
 
+  /**
+   * Resets this sketch an empty state.
+   */
+  @SuppressWarnings("unchecked")
+  public void reset() {
+    isEmpty_ = true;
+    count_ = 0;
+    theta_ = (long) (Long.MAX_VALUE * (double) samplingProbability_);
+    final int startingCapacity = Util.getStartingCapacity(nomEntries_, lgResizeFactor_);
+    lgCurrentCapacity_ = Integer.numberOfTrailingZeros(startingCapacity);
+    keys_ = new long[startingCapacity];
+    summaries_ = (S[]) Array.newInstance(summaryFactory_.newSummary().getClass(), startingCapacity);
+    setRebuildThreshold();
+  }
+
   /**
    * Converts the current state of the sketch into a compact sketch
    * @return compact sketch
@@ -344,11 +351,11 @@ boolean isInSamplingMode() {
   }
 
   void setThetaLong(final long theta) {
-    this.theta_ = theta;
+    theta_ = theta;
   }
 
   void setNotEmpty() {
-    this.isEmpty_ = false;
+    isEmpty_ = false;
   }
 
   SummaryFactory<S> getSummaryFactory() {

sketches/src/main/java/com/yahoo/sketches/tuple/Util.java

@@ -5,9 +5,13 @@
 
 package com.yahoo.sketches.tuple;
 
+import static com.yahoo.sketches.Util.MIN_LG_ARR_LONGS;
+import static com.yahoo.sketches.Util.ceilingPowerOf2;
+import static com.yahoo.sketches.Util.startingSubMultiple;
 import static com.yahoo.sketches.hash.MurmurHash3.hash;
 import static java.nio.charset.StandardCharsets.UTF_8;
 
+import com.yahoo.sketches.ResizeFactor;
 import com.yahoo.sketches.SketchesArgumentException;
 
 final class Util {
@@ -49,4 +53,13 @@ static final void checkSeedHashes(final short seedHashA, final short seedHashB)
 
   }
 
+  static int getStartingCapacity(final int nomEntries, final int lgResizeFactor) {
+    return 1 << startingSubMultiple(
+      // target table size is twice the number of nominal entries
+      Integer.numberOfTrailingZeros(ceilingPowerOf2(nomEntries) * 2),
+      ResizeFactor.getRF(lgResizeFactor),
+      MIN_LG_ARR_LONGS
+    );
+  }
+
 }

sketches/src/test/java/com/yahoo/sketches/tuple/DirectArrayOfDoublesQuickSelectSketchTest.java

@@ -92,6 +92,19 @@ public void exactMode() {
     for (int i = 0; i < values.length; i++) if (values[i] != null) count++;
     Assert.assertEquals(count, 4096);
     for (int i = 0; i < 4096; i++) Assert.assertEquals(values[i][0], 1.0);
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
+    ArrayOfDoublesSketchIterator it = sketch.iterator();
+    while (it.next()) {
+      Assert.fail("empty sketch expected");
+    }
   }
 
   @Test
@@ -120,6 +133,19 @@ public void estimationMode() {
       }
     }
     Assert.assertEquals(count, values.length);
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
+    ArrayOfDoublesSketchIterator it = sketch.iterator();
+    while (it.next()) {
+      Assert.fail("empty sketch expected");
+    }
   }
 
   @Test

sketches/src/test/java/com/yahoo/sketches/tuple/HeapArrayOfDoublesQuickSelectSketchTest.java

@@ -74,6 +74,19 @@ public void exactMode() {
     for (int i = 0; i < values.length; i++) if (values[i] != null) count++;
     Assert.assertEquals(count, 4096);
     for (int i = 0; i < 4096; i++) Assert.assertEquals(values[i][0], 1.0);
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
+    ArrayOfDoublesSketchIterator it = sketch.iterator();
+    while (it.next()) {
+      Assert.fail("empty sketch expected");
+    }
   }
 
   @Test
@@ -101,6 +114,19 @@ public void estimationMode() {
       }
     }
     Assert.assertEquals(count, values.length);
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
+    ArrayOfDoublesSketchIterator it = sketch.iterator();
+    while (it.next()) {
+      Assert.fail("empty sketch expected");
+    }
   }
 
   @Test

sketches/src/test/java/com/yahoo/sketches/tuple/UpdatableSketchWithDoubleSummaryTest.java

@@ -71,6 +71,15 @@ public void exactMode() {
     for (int i = 0; i < summaries.length; i++) if (summaries[i] != null) count++;
     Assert.assertEquals(count, 4096);
     Assert.assertEquals(summaries[0].getValue(), 1.0);
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
   }
 
   @Test
@@ -95,7 +104,16 @@ public void estimationMode() {
       }
     }
     Assert.assertEquals(count, summaries.length);
-  }
+
+    sketch.reset();
+    Assert.assertTrue(sketch.isEmpty());
+    Assert.assertFalse(sketch.isEstimationMode());
+    Assert.assertEquals(sketch.getEstimate(), 0.0);
+    Assert.assertEquals(sketch.getUpperBound(1), 0.0);
+    Assert.assertEquals(sketch.getLowerBound(1), 0.0);
+    Assert.assertEquals(sketch.getThetaLong(), Long.MAX_VALUE);
+    Assert.assertEquals(sketch.getTheta(), 1.0);
+}
 
   @Test
   public void estimationModeWithSamplingNoResizing() {