HIVE-29368: Sonar Qube feedback + one more test

Author: konstantinb

ql/src/java/org/apache/hadoop/hive/ql/stats/StatsUtils.java

@@ -2088,19 +2088,7 @@ private static List<Long> extractNDVGroupingColumns(List<ColStatistics> colStats
     // compute product of distinct values of grouping columns
     for (ColStatistics cs : colStats) {
       if (cs != null) {
-        long ndv = cs.getCountDistint();
-
-        if (ndv == 0L) {
-          // Typically, ndv == 0 means "NDV unknown", and no safe GROUPBY adjustments are possible
-          // However, there is a special exception for "constant NULL" columns. They are intentionally generated
-          // with NDV values of 0 and numNulls == numRows, while their actual NDV is 1
-          if (cs.getNumNulls() >= parentStats.getNumRows()) {
-            ndv = 1L;
-          }
-        } else if (cs.getNumNulls() > 0L) {
-          ndv = StatsUtils.safeAdd(ndv, 1L);
-        }
-        ndvValues.add(ndv);
+        ndvValues.add(getGroupingColumnNdv(cs, parentStats));
       } else {
         if (parentStats.getColumnStatsState().equals(Statistics.State.COMPLETE)) {
           // the column must be an aggregate column inserted by GBY. We
@@ -2119,4 +2107,20 @@ private static List<Long> extractNDVGroupingColumns(List<ColStatistics> colStats
 
     return ndvValues;
   }
+
+  private static long getGroupingColumnNdv(ColStatistics cs, Statistics parentStats) {
+    long ndv = cs.getCountDistint();
+
+    if (ndv == 0L) {
+      // Typically, ndv == 0 means "NDV unknown", and no safe GROUPBY adjustments are possible
+      // However, there is a special exception for "constant NULL" columns. They are intentionally generated
+      // with NDV values of 0 and numNulls == numRows, while their actual NDV is 1
+      if (cs.getNumNulls() >= parentStats.getNumRows()) {
+        ndv = 1L;
+      }
+    } else if (cs.getNumNulls() > 0L) {
+      ndv = StatsUtils.safeAdd(ndv, 1L);
+    }
+    return ndv;
+  }
 }

ql/src/test/org/apache/hadoop/hive/ql/stats/TestStatsUtils.java

@@ -321,4 +321,16 @@ void testComputeNDVGroupingColumnsPartialStats() {
     assertEquals(0, ndv, "Partial stats (ndv=0, numNulls<numRows) should return 0, not inflate to 1");
   }
 
+  @Test
+  void testComputeNDVGroupingColumnsAllNulls() {
+    // When ndv=0 and numNulls >= numRows, it's a "constant NULL" column, so NDV should be 1
+    ColStatistics cs = createColStats("all_nulls_col", 0, 1000);
+    Statistics parentStats = createParentStats(1000);
+    List<ColStatistics> colStats = Collections.singletonList(cs);
+
+    long ndv = StatsUtils.computeNDVGroupingColumns(colStats, parentStats, false);
+
+    assertEquals(1, ndv, "All-null column (ndv=0, numNulls==numRows) should have NDV inflated to 1");
+  }
+
 }