Merge branch 'master' into 5.0.X

apache · Dec 21, 2023 · 93fd895 · 93fd895
2 parents 0fbe027 + 4362127
commit 93fd895
Show file tree

Hide file tree

Showing 26 changed files with 460 additions and 275 deletions.
diff --git a/src/main/java/org/apache/datasketches/hll/Union.java b/src/main/java/org/apache/datasketches/hll/Union.java
@@ -306,8 +306,9 @@ public byte[] toUpdatableByteArray() {
   @Override
   public String toString(final boolean summary, final boolean hllDetail,
       final boolean auxDetail, final boolean all) {
-    checkRebuildCurMinNumKxQ(gadget);
-    return gadget.toString(summary, hllDetail, auxDetail, all);
+    final HllSketch clone = gadget.copy();
+    checkRebuildCurMinNumKxQ(clone);
+    return clone.toString(summary, hllDetail, auxDetail, all);
   }
 
   /**

diff --git a/src/main/java/org/apache/datasketches/kll/KllDirectCompactItemsSketch.java b/src/main/java/org/apache/datasketches/kll/KllDirectCompactItemsSketch.java
@@ -32,6 +32,7 @@
 import java.lang.reflect.Array;
 import java.util.Comparator;
 
+import org.apache.datasketches.common.ArrayOfBooleansSerDe;
 import org.apache.datasketches.common.ArrayOfItemsSerDe;
 import org.apache.datasketches.common.SketchesArgumentException;
 import org.apache.datasketches.memory.Memory;
@@ -65,6 +66,14 @@ final class KllDirectCompactItemsSketch<T> extends KllItemsSketch<T> {
     levelsArr = memVal.levelsArr; //always converted to writable form.
   }
 
+  //End of constructors
+
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString(getTotalItemsArray()[index]);
+  }
+
   @Override
   public int getK() {
     return getMemoryK(mem);
@@ -83,6 +92,12 @@ public T getMaxItem() {
     return serDe.deserializeFromMemory(mem, offset, 2)[1];
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString(getMaxItem());
+  }
+
   @Override
   public T getMinItem() {
     if (sketchStructure == COMPACT_EMPTY || isEmpty()) {
@@ -96,6 +111,12 @@ public T getMinItem() {
     return serDe.deserializeFromMemory(mem, offset, 1)[0];
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString(getMinItem());
+  }
+
   @Override
   public long getN() {
     if (sketchStructure == COMPACT_EMPTY) { return 0; }
@@ -134,6 +155,7 @@ byte[] getMinMaxByteArr() { //this is only used by COMPACT_FULL
   @Override
   int getMinMaxSizeBytes() { //this is only used by COMPACT_FULL
     final int offset = DATA_START_ADR + getNumLevels() * Integer.BYTES;
+    if (serDe instanceof ArrayOfBooleansSerDe) { return 2; }
     return serDe.sizeOf(mem, offset, 2);
   }
 

diff --git a/src/main/java/org/apache/datasketches/kll/KllDirectDoublesSketch.java b/src/main/java/org/apache/datasketches/kll/KllDirectDoublesSketch.java
@@ -116,7 +116,13 @@ static KllDirectDoublesSketch newDirectUpdatableInstance(
     return new KllDirectDoublesSketch(UPDATABLE, wMem, memReqSvr, memVal);
   }
 
-  //END of Constructors
+  //End of constructors
+
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(getDoubleItemsArray()[index]);
+  }
 
   @Override
   public int getK() {
@@ -137,6 +143,12 @@ else if (sketchStructure == COMPACT_FULL) {
     return wmem.getDouble(offset);
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(getMaxItem());
+  }
+
   @Override
   public double getMinItem() {
     int levelsArrBytes = 0;
@@ -151,6 +163,12 @@ else if (sketchStructure == COMPACT_FULL) {
     return wmem.getDouble(offset);
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(getMinItem());
+  }
+
   @Override
   public long getN() {
     if (sketchStructure == COMPACT_EMPTY) { return 0; }

diff --git a/src/main/java/org/apache/datasketches/kll/KllDirectFloatsSketch.java b/src/main/java/org/apache/datasketches/kll/KllDirectFloatsSketch.java
@@ -118,6 +118,12 @@ static KllDirectFloatsSketch newDirectUpdatableInstance(
 
   //END of Constructors
 
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "NaN"; }
+    return Float.toString(getFloatItemsArray()[index]);
+  }
+
   @Override
   public int getK() {
     return getMemoryK(wmem);
@@ -137,6 +143,12 @@ else if (sketchStructure == COMPACT_FULL) {
     return wmem.getFloat(offset);
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Float.toString(getMaxItem());
+  }
+
   @Override
   public float getMinItem() {
     int levelsArrBytes = 0;
@@ -151,6 +163,12 @@ else if (sketchStructure == COMPACT_FULL) {
     return wmem.getFloat(offset);
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Float.toString(getMinItem());
+  }
+
   @Override
   public long getN() {
     if (sketchStructure == COMPACT_EMPTY) { return 0; }

diff --git a/src/main/java/org/apache/datasketches/kll/KllDoublesHelper.java b/src/main/java/org/apache/datasketches/kll/KllDoublesHelper.java
@@ -309,12 +309,16 @@ static void updateDouble(final KllDoublesSketch dblSk,
       dblSk.setMinItem(min(dblSk.getMinItem(), item));
       dblSk.setMaxItem(max(dblSk.getMaxItem(), item));
     }
-    if (dblSk.levelsArr[0] == 0) { compressWhileUpdatingSketch(dblSk); }
-    final int myLevelsArrAtZero = dblSk.levelsArr[0]; //LevelsArr could be expanded
+    int level0space = dblSk.levelsArr[0];
+    assert (level0space >= 0);
+    if (level0space == 0) {
+      compressWhileUpdatingSketch(dblSk);
+      level0space = dblSk.levelsArr[0];
+      assert (level0space > 0);
+    }
     dblSk.incN();
     dblSk.setLevelZeroSorted(false);
-    final int nextPos = myLevelsArrAtZero - 1;
-    assert myLevelsArrAtZero >= 0;
+    final int nextPos = level0space - 1;
     dblSk.setLevelsArrayAt(0, nextPos);
     dblSk.setDoubleItemsArrayAt(nextPos, item);
   }

diff --git a/src/main/java/org/apache/datasketches/kll/KllDoublesSketch.java b/src/main/java/org/apache/datasketches/kll/KllDoublesSketch.java
@@ -306,6 +306,17 @@ public byte[] toByteArray() {
     return KllHelper.toByteArray(this, false);
   }
 
+  @Override
+  public String toString(final boolean withSummary, final boolean withData) {
+    KllSketch sketch = this;
+    if (withData && sketchStructure != UPDATABLE) {
+      final Memory mem = getWritableMemory();
+      assert mem != null;
+      sketch = KllDoublesSketch.heapify(getWritableMemory());
+    }
+    return KllHelper.toStringImpl(sketch, withSummary, withData, getSerDe());
+  }
+
   @Override
   public void update(final double item) {
     if (readOnly) { throw new SketchesArgumentException(TGT_IS_READ_ONLY_MSG); }

diff --git a/src/main/java/org/apache/datasketches/kll/KllFloatsHelper.java b/src/main/java/org/apache/datasketches/kll/KllFloatsHelper.java
@@ -309,12 +309,16 @@ static void updateFloat(final KllFloatsSketch fltSk,
       fltSk.setMinItem(min(fltSk.getMinItem(), item));
       fltSk.setMaxItem(max(fltSk.getMaxItem(), item));
     }
-    if (fltSk.levelsArr[0] == 0) { compressWhileUpdatingSketch(fltSk); }
-    final int myLevelsArrAtZero = fltSk.levelsArr[0]; //LevelsArr could be expanded
+    int level0space = fltSk.levelsArr[0];
+    assert level0space >= 0;
+    if (level0space == 0) {
+      compressWhileUpdatingSketch(fltSk);
+      level0space = fltSk.levelsArr[0];
+      assert (level0space > 0);
+    }
     fltSk.incN();
     fltSk.setLevelZeroSorted(false);
-    final int nextPos = myLevelsArrAtZero - 1;
-    assert myLevelsArrAtZero >= 0;
+    final int nextPos = level0space - 1;
     fltSk.setLevelsArrayAt(0, nextPos);
     fltSk.setFloatItemsArrayAt(nextPos, item);
   }

diff --git a/src/main/java/org/apache/datasketches/kll/KllFloatsSketch.java b/src/main/java/org/apache/datasketches/kll/KllFloatsSketch.java
@@ -306,6 +306,17 @@ public byte[] toByteArray() {
     return KllHelper.toByteArray(this, false);
   }
 
+  @Override
+  public String toString(final boolean withSummary, final boolean withData) {
+    KllSketch sketch = this;
+    if (withData && sketchStructure != UPDATABLE) {
+      final Memory mem = getWritableMemory();
+      assert mem != null;
+      sketch = KllFloatsSketch.heapify(getWritableMemory());
+    }
+    return KllHelper.toStringImpl(sketch, withSummary, withData, getSerDe());
+  }
+
   @Override
   public void update(final float item) {
     if (readOnly) { throw new SketchesArgumentException(TGT_IS_READ_ONLY_MSG); }

diff --git a/src/main/java/org/apache/datasketches/kll/KllHeapDoublesSketch.java b/src/main/java/org/apache/datasketches/kll/KllHeapDoublesSketch.java
@@ -141,6 +141,14 @@ static KllHeapDoublesSketch heapifyImpl(final Memory srcMem) {
     return new KllHeapDoublesSketch(srcMem, memVal);
   }
 
+  //End of constructors
+
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(doubleItems[index]);
+  }
+
   @Override
   public int getK() { return k; }
 
@@ -150,12 +158,24 @@ public double getMaxItem() {
     return maxDoubleItem;
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(maxDoubleItem);
+  }
+
   @Override
   public double getMinItem() {
     if (isEmpty()) { throw new SketchesArgumentException(EMPTY_MSG); }
     return minDoubleItem;
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(minDoubleItem);
+  }
+
   @Override
   public long getN() { return n; }
 

diff --git a/src/main/java/org/apache/datasketches/kll/KllHeapFloatsSketch.java b/src/main/java/org/apache/datasketches/kll/KllHeapFloatsSketch.java
@@ -141,6 +141,14 @@ static KllHeapFloatsSketch heapifyImpl(final Memory srcMem) {
     return new KllHeapFloatsSketch(srcMem, memVal);
   }
 
+  //End of constructors
+
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "NaN"; }
+    return Double.toString(floatItems[index]);
+  }
+
   @Override
   public int getK() { return k; }
 
@@ -150,12 +158,24 @@ public float getMaxItem() {
     return maxFloatItem;
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Float.toString(maxFloatItem);
+  }
+
   @Override
   public float getMinItem() {
     if (isEmpty()) { throw new SketchesArgumentException(EMPTY_MSG); }
     return minFloatItem;
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "NaN"; }
+    return Float.toString(minFloatItem);
+  }
+
   @Override
   public long getN() { return n; }
 

diff --git a/src/main/java/org/apache/datasketches/kll/KllHeapItemsSketch.java b/src/main/java/org/apache/datasketches/kll/KllHeapItemsSketch.java
@@ -117,6 +117,14 @@ final class KllHeapItemsSketch<T> extends KllItemsSketch<T> {
     }
   }
 
+  //End of constructors
+
+  @Override
+  String getItemAsString(final int index) {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString((T)(itemsArr[index]));
+  }
+
   @Override
   public int getK() {
     return k;
@@ -128,12 +136,24 @@ public T getMaxItem() {
     return maxItem;
   }
 
+  @Override
+  String getMaxItemAsString() {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString(maxItem);
+  }
+
   @Override
   public T getMinItem() {
     if (isEmpty()) { throw new SketchesArgumentException(EMPTY_MSG); }
     return minItem;
   }
 
+  @Override
+  String getMinItemAsString() {
+    if (isEmpty()) { return "Null"; }
+    return serDe.toString(minItem);
+  }
+
   @Override
   public long getN() {
     return n;