Fix count on null values

Before the change, the `ValuesExec` containing `NullArray` would
incorrectly report column statistics as being non-null, which would
misinform `AggregateStatistics` optimizer and fold `count(always_null)`
into row count instead of 0.

This commit fixes the column statistics derivation for values with
`NullArray` and therefore fixes execution of logical plans with count
over such values.

Note that the bug was not reproducible using DataFusion SQL frontend,
because in DataFusion SQL the `VALUES (NULL)` doesn't have type
`DataType:Null` (it has some apparently arbitrarily picked type
instead).

As a follow-up, all usages of `Array:null_count` should be inspected.
The function can easily be misused (it returns "physical nulls", which
do not exist for null type).

datafusion/core/tests/core_integration.rs

@@ -24,6 +24,9 @@ mod dataframe;
 /// Run all tests that are found in the `macro_hygiene` directory
 mod macro_hygiene;
 
+/// Run all tests that are found in the `execution` directory
+mod execution;
+
 /// Run all tests that are found in the `expr_api` directory
 mod expr_api;
 

datafusion/core/tests/execution/logical_plan.rs

@@ -0,0 +1,84 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+use arrow_array::{RecordBatch, UInt64Array};
+use arrow_schema::{DataType, Field, Schema};
+use datafusion::execution::session_state::SessionStateBuilder;
+use datafusion_common::{Column, DFSchema, Result, ScalarValue};
+use datafusion_execution::TaskContext;
+use datafusion_expr::expr::AggregateFunction;
+use datafusion_expr::logical_plan::{LogicalPlan, Values};
+use datafusion_expr::{Aggregate, AggregateUDF, Expr};
+use datafusion_functions_aggregate::count::Count;
+use datafusion_physical_plan::collect;
+///! Logical plans need to provide stable semantics, as downstream projects
+///! create them and depend on them. Test executable semantics of logical plans.
+use std::collections::HashMap;
+use std::sync::Arc;
+
+#[tokio::test]
+async fn count_only_nulls() -> Result<()> {
+    // Input: VALUES (NULL), (NULL), (NULL) AS _(col)
+    let input_schema = Arc::new(DFSchema::from_unqualified_fields(
+        vec![Field::new("col", DataType::Null, true)].into(),
+        HashMap::new(),
+    )?);
+    let input = Arc::new(LogicalPlan::Values(Values {
+        schema: input_schema,
+        values: vec![
+            vec![Expr::Literal(ScalarValue::Null)],
+            vec![Expr::Literal(ScalarValue::Null)],
+            vec![Expr::Literal(ScalarValue::Null)],
+        ],
+    }));
+    let input_col_ref = Expr::Column(Column {
+        relation: None,
+        name: "col".to_string(),
+    });
+
+    // Aggregation: count(col) AS count
+    let aggregate = LogicalPlan::Aggregate(Aggregate::try_new(
+        input,
+        vec![],
+        vec![Expr::AggregateFunction(AggregateFunction {
+            func: Arc::new(AggregateUDF::new_from_impl(Count::new())),
+            args: vec![input_col_ref],
+            distinct: false,
+            filter: None,
+            order_by: None,
+            null_treatment: None,
+        })],
+    )?);
+
+    // Execute and verify results
+    let session_state = SessionStateBuilder::new().build();
+    let physical_plan = session_state.create_physical_plan(&aggregate).await?;
+    let result =
+        collect(physical_plan, Arc::new(TaskContext::from(&session_state))).await?;
+
+    let expected_schema = Arc::new(Schema::new(vec![Field::new(
+        "count",
+        DataType::UInt64,
+        false,
+    )]));
+    let expected = RecordBatch::try_new(
+        expected_schema,
+        vec![Arc::new(UInt64Array::from(vec![0]))],
+    )?;
+    assert_eq!(result, vec![expected]);
+    Ok(())
+}

datafusion/core/tests/execution/mod.rs

@@ -0,0 +1,18 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+mod logical_plan;

datafusion/physical-plan/src/common.rs

@@ -156,7 +156,11 @@ pub fn compute_record_batch_statistics(
     for partition in batches.iter() {
         for batch in partition {
             for (stat_index, col_index) in projection.iter().enumerate() {
-                null_counts[stat_index] += batch.column(*col_index).null_count();
+                null_counts[stat_index] += batch
+                    .column(*col_index)
+                    .logical_nulls()
+                    .map(|nulls| nulls.null_count())
+                    .unwrap_or(Default::default());
             }
         }
     }

datafusion/physical-plan/src/values.rs

@@ -219,6 +219,7 @@ mod tests {
     use crate::test::{self, make_partition};
 
     use arrow_schema::{DataType, Field};
+    use datafusion_common::stats::{ColumnStatistics, Precision};
 
     #[tokio::test]
     async fn values_empty_case() -> Result<()> {
@@ -269,4 +270,34 @@ mod tests {
         let _ = ValuesExec::try_new(schema, vec![vec![lit(ScalarValue::UInt32(None))]])
             .unwrap_err();
     }
+
+    #[test]
+    fn values_stats_with_nulls_only() -> Result<()> {
+        let data = vec![
+            vec![lit(ScalarValue::Null)],
+            vec![lit(ScalarValue::Null)],
+            vec![lit(ScalarValue::Null)],
+        ];
+        let rows = data.len();
+        let values = ValuesExec::try_new(
+            Arc::new(Schema::new(vec![Field::new("col0", DataType::Null, true)])),
+            data,
+        )?;
+
+        assert_eq!(
+            values.statistics()?,
+            Statistics {
+                num_rows: Precision::Exact(rows),
+                total_byte_size: Precision::Exact(8), // not important
+                column_statistics: vec![ColumnStatistics {
+                    null_count: Precision::Exact(rows), // there are only nulls
+                    distinct_count: Precision::Absent,
+                    max_value: Precision::Absent,
+                    min_value: Precision::Absent,
+                },],
+            }
+        );
+
+        Ok(())
+    }
 }