Add methods to count the number null values in the vector

Author: zeapo

java/vector/src/main/java/org/apache/arrow/vector/BaseValueVector.java

@@ -72,6 +72,11 @@ protected BaseAccessor() { }
     public boolean isNull(int index) {
       return false;
     }
+
+    @Override
+    public int getNullCount() {
+      return 0;
+    }
   }
 
   public abstract static class BaseMutator implements ValueVector.Mutator {

java/vector/src/main/java/org/apache/arrow/vector/BitVector.java

@@ -379,6 +379,21 @@ public final void get(int index, NullableBitHolder holder) {
       holder.isSet = 1;
       holder.value = get(index);
     }
+
+    /**
+     * Get the number of bits set to 1
+     * @return the number of bits set to 1
+     */
+    public final int getNullCount() {
+      int count = 0;
+      for (int i = 0; i < allocationSizeInBytes; ++i) {
+        byte byteValue = data.getByte(i);
+        // Java uses two's complement binary representation, hence 11111111_b which is -1 when converted to Int
+        // will have 32bits set to 1. Masking the MSB and then adding it back solves the issue.
+        count += Integer.bitCount(byteValue & 0x7F) - (byteValue >> 7);
+      }
+      return count;
+    }
   }
 
   /**

java/vector/src/main/java/org/apache/arrow/vector/ValueVector.java

@@ -180,6 +180,11 @@ interface Accessor {
      * Returns true if the value at the given index is null, false otherwise.
      */
     boolean isNull(int index);
+
+    /**
+     * Returns the number of null values
+     */
+    int getNullCount();
   }
 
   /**

java/vector/src/main/java/org/apache/arrow/vector/ZeroVector.java

@@ -69,6 +69,11 @@ public int getValueCount() {
     public boolean isNull(int index) {
       return true;
     }
+
+    @Override
+    public int getNullCount() {
+      return 0;
+    }
   };
 
   private final Mutator defaultMutator = new Mutator() {

java/vector/src/main/java/org/apache/arrow/vector/complex/ListVector.java

@@ -310,6 +310,11 @@ public Object getObject(int index) {
     public boolean isNull(int index) {
       return bits.getAccessor().get(index) == 0;
     }
+
+    @Override
+    public int getNullCount() {
+      return bits.getAccessor().getNullCount();
+    }
   }
 
   public class Mutator extends BaseRepeatedMutator {

java/vector/src/test/java/org/apache/arrow/vector/TestValueVector.java

@@ -301,6 +301,8 @@ public void testBitVector() {
       assertEquals(0, accessor.get(100));
       assertEquals(1, accessor.get(1022));
 
+      assertEquals(2, accessor.getNullCount());
+
       // test setting the same value twice
       m.set(0, 1);
       m.set(0, 1);
@@ -315,8 +317,22 @@ public void testBitVector() {
       assertEquals(0, accessor.get(0));
       assertEquals(1, accessor.get(1));
 
+      // should not change
+      assertEquals(2, accessor.getNullCount());
+
       // Ensure unallocated space returns 0
       assertEquals(0, accessor.get(3));
+
+      m.set(1, 0);
+      m.set(1022, 0);
+
+      assertEquals(0, accessor.getNullCount());
+
+      for (int i = 0; i < 1024; ++i) {
+        m.set(i, 1);
+      }
+
+      assertEquals(1024, accessor.getNullCount());
     }
   }