[BugFix] fix concat expr with multiple args nullsFraction (StarRocks#…

…52683) Signed-off-by: stephen <[email protected]> Signed-off-by: zhiminr.ren <[email protected]>
renzhimin7 · Nov 7, 2024 · 5bde6f1 · 5bde6f1
1 parent 6e1f8dd
commit 5bde6f1
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 2 deletions.
diff --git a/...e/src/main/java/com/starrocks/sql/optimizer/statistics/ExpressionStatisticCalculator.java b/...e/src/main/java/com/starrocks/sql/optimizer/statistics/ExpressionStatisticCalculator.java
@@ -643,8 +643,8 @@ private ColumnStatistic multiaryExpressionCalculate(CallOperator callOperator,
                     distinctValues = Math.min(rowCount,
                             childColumnStatisticList.stream().mapToDouble(ColumnStatistic::getDistinctValuesCount).sum());
                     averageRowSize = childColumnStatisticList.stream().mapToDouble(ColumnStatistic::getAverageRowSize).sum();
-                    nullsFraction = 1 - childColumnStatisticList.stream().mapToDouble(ColumnStatistic::getAverageRowSize)
-                            .reduce(1, (a, b) -> (1 - a) * (1 - b));
+                    nullsFraction = 1 - childColumnStatisticList.stream().mapToDouble(ColumnStatistic::getNullsFraction)
+                            .reduce(1.0, (accumulator, nullFraction) -> accumulator * (1 - nullFraction));
                     return new ColumnStatistic(Double.NEGATIVE_INFINITY, Double.POSITIVE_INFINITY,
                             nullsFraction, averageRowSize, distinctValues);
                 default:

diff --git a/...c/test/java/com/starrocks/sql/optimizer/statistics/PredicateStatisticsCalculatorTest.java b/...c/test/java/com/starrocks/sql/optimizer/statistics/PredicateStatisticsCalculatorTest.java
@@ -15,10 +15,12 @@
 
 package com.starrocks.sql.optimizer.statistics;
 
+import com.google.common.collect.Lists;
 import com.starrocks.analysis.BinaryType;
 import com.starrocks.catalog.Type;
 import com.starrocks.sql.optimizer.Utils;
 import com.starrocks.sql.optimizer.operator.scalar.BinaryPredicateOperator;
+import com.starrocks.sql.optimizer.operator.scalar.CallOperator;
 import com.starrocks.sql.optimizer.operator.scalar.ColumnRefOperator;
 import com.starrocks.sql.optimizer.operator.scalar.CompoundPredicateOperator;
 import com.starrocks.sql.optimizer.operator.scalar.ConstantOperator;
@@ -111,4 +113,18 @@ public void testNullEqStatistic() throws Exception {
         Assert.assertEquals(5000, estimatedStatistics.getOutputRowCount(), 0.001);
         Assert.assertEquals(1, estimatedStatistics.getColumnStatistic(c1).getNullsFraction(), 0.001);
     }
+
+    @Test
+    public void testConcatExpressionCalculate() {
+        ColumnRefOperator c1 = new ColumnRefOperator(0, Type.VARCHAR, "c1", true);
+        ConstantOperator c2 = new ConstantOperator("-", Type.VARCHAR);
+        ColumnRefOperator c3 = new ColumnRefOperator(1, Type.VARCHAR, "c3", true);
+        CallOperator concat = new CallOperator("concat", Type.VARCHAR, Lists.newArrayList(c1, c2, c3));
+        Statistics statistics = Statistics.builder()
+                .addColumnStatistic(c1, ColumnStatistic.builder().setNullsFraction(0.2).setDistinctValuesCount(10).build())
+                .addColumnStatistic(c3, ColumnStatistic.builder().setNullsFraction(0.4).setDistinctValuesCount(10).build())
+                .setOutputRowCount(10000).build();
+        ColumnStatistic estimatedStatistics = ExpressionStatisticCalculator.calculate(concat, statistics);
+        Assert.assertEquals(0.52, estimatedStatistics.getNullsFraction(), 0.001);
+    }
 }