perf: add fast path for uniform fill values in array_resize (#20617)

lyne7-sc · alamb · web-flow · commit 1068686fc636 · 2026-04-16T18:39:49.000Z
## Which issue does this PR close?  - Closes #. ## Rationale for this change  `array_resize` currently does extra per-row work when resizing list arrays. This pr optimizes the common fast path where the fill value is uniform. ## What changes are included in this PR? - Add a fast path in `array_resize` for uniform fill values. - Precompute the maximum required growth and reuse a single fill buffer in the uniform-fill path. ### Benchmarks ``` group main optimized ----- ---- --------- array_resize_i64/grow_default_null_fill_10_to_500 11.24 3.2±0.07ms ? ?/sec 1.00 283.9±20.10µs ? ?/sec array_resize_i64/grow_uniform_fill_10_to_500 4.15 1648.0±38.12µs ? ?/sec 1.00 397.2±20.46µs ? ?/sec array_resize_i64/grow_variable_fill_10_to_500 1.00 1667.7±50.31µs ? ?/sec 1.02 1692.8±61.54µs ? ?/sec array_resize_i64/mixed_grow_shrink_1000x_100 4.83 373.0±8.91µs ? ?/sec 1.00 77.1±5.90µs ? ?/sec array_resize_i64/shrink_uniform_fill_500_to_10 1.00 8.1±0.51µs ? ?/sec 1.06 8.5±0.50µs ? ?/sec ```  ## Are these changes tested? Yes  ## Are there any user-facing changes? No   --------- Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>
diff --git a/datafusion/functions-nested/Cargo.toml b/datafusion/functions-nested/Cargo.toml
@@ -126,3 +126,7 @@ name = "array_sort"
 [[bench]]
 harness = false
 name = "string_to_array"
+
+[[bench]]
+harness = false
+name = "array_resize"
diff --git a/datafusion/functions-nested/benches/array_resize.rs b/datafusion/functions-nested/benches/array_resize.rs
@@ -0,0 +1,170 @@
+// 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::{ArrayRef, Int64Array, ListArray};
+use arrow::buffer::OffsetBuffer;
+use arrow::datatypes::{DataType, Field};
+use criterion::{
+    BenchmarkGroup, Criterion, criterion_group, criterion_main, measurement::WallTime,
+};
+use datafusion_common::config::ConfigOptions;
+use datafusion_expr::{ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl};
+use datafusion_functions_nested::resize::ArrayResize;
+use std::hint::black_box;
+use std::sync::Arc;
+
+const NUM_ROWS: usize = 1_000;
+
+fn criterion_benchmark(c: &mut Criterion) {
+    let mut group = c.benchmark_group("array_resize_i64");
+    let list_field: Arc<Field> = Field::new_list_field(DataType::Int64, true).into();
+    let list_data_type = DataType::List(Arc::clone(&list_field));
+    let arg_fields = vec![
+        Field::new("array", list_data_type.clone(), true).into(),
+        Field::new("size", DataType::Int64, false).into(),
+        Field::new("value", DataType::Int64, true).into(),
+    ];
+    let return_field: Arc<Field> = Field::new("result", list_data_type, true).into();
+    let config_options = Arc::new(ConfigOptions::default());
+    let two_arg_fields = arg_fields[..2].to_vec();
+
+    bench_case(
+        &mut group,
+        "grow_uniform_fill_10_to_500",
+        &[
+            ColumnarValue::Array(create_int64_list_array(NUM_ROWS, 10)),
+            ColumnarValue::Array(repeated_int64_array(500)),
+            ColumnarValue::Array(repeated_int64_array(7)),
+        ],
+        &arg_fields,
+        &return_field,
+        &config_options,
+    );
+
+    bench_case(
+        &mut group,
+        "shrink_uniform_fill_500_to_10",
+        &[
+            ColumnarValue::Array(create_int64_list_array(NUM_ROWS, 500)),
+            ColumnarValue::Array(repeated_int64_array(10)),
+            ColumnarValue::Array(repeated_int64_array(7)),
+        ],
+        &arg_fields,
+        &return_field,
+        &config_options,
+    );
+
+    bench_case(
+        &mut group,
+        "grow_default_null_fill_10_to_500",
+        &[
+            ColumnarValue::Array(create_int64_list_array(NUM_ROWS, 10)),
+            ColumnarValue::Array(repeated_int64_array(500)),
+        ],
+        &two_arg_fields,
+        &return_field,
+        &config_options,
+    );
+
+    bench_case(
+        &mut group,
+        "grow_variable_fill_10_to_500",
+        &[
+            ColumnarValue::Array(create_int64_list_array(NUM_ROWS, 10)),
+            ColumnarValue::Array(repeated_int64_array(500)),
+            ColumnarValue::Array(distinct_fill_array()),
+        ],
+        &arg_fields,
+        &return_field,
+        &config_options,
+    );
+
+    bench_case(
+        &mut group,
+        "mixed_grow_shrink_1000x_100",
+        &[
+            ColumnarValue::Array(create_int64_list_array(NUM_ROWS, 100)),
+            ColumnarValue::Array(mixed_size_array()),
+        ],
+        &arg_fields[..2],
+        &return_field,
+        &config_options,
+    );
+
+    group.finish();
+}
+
+fn bench_case(
+    group: &mut BenchmarkGroup<'_, WallTime>,
+    name: &str,
+    args: &[ColumnarValue],
+    arg_fields: &[Arc<Field>],
+    return_field: &Arc<Field>,
+    config_options: &Arc<ConfigOptions>,
+) {
+    let udf = ArrayResize::new();
+    group.bench_function(name, |b| {
+        b.iter(|| {
+            black_box(
+                udf.invoke_with_args(ScalarFunctionArgs {
+                    args: args.to_vec(),
+                    arg_fields: arg_fields.to_vec(),
+                    number_rows: NUM_ROWS,
+                    return_field: return_field.clone(),
+                    config_options: config_options.clone(),
+                })
+                .unwrap(),
+            )
+        })
+    });
+}
+
+fn create_int64_list_array(num_rows: usize, list_len: usize) -> ArrayRef {
+    let values = (0..(num_rows * list_len))
+        .map(|v| Some(v as i64))
+        .collect::<Int64Array>();
+    let offsets = (0..=num_rows)
+        .map(|i| (i * list_len) as i32)
+        .collect::<Vec<i32>>();
+
+    Arc::new(
+        ListArray::try_new(
+            Arc::new(Field::new_list_field(DataType::Int64, true)),
+            OffsetBuffer::new(offsets.into()),
+            Arc::new(values),
+            None,
+        )
+        .unwrap(),
+    )
+}
+
+fn repeated_int64_array(value: i64) -> ArrayRef {
+    Arc::new(Int64Array::from_value(value, NUM_ROWS))
+}
+
+fn distinct_fill_array() -> ArrayRef {
+    Arc::new(Int64Array::from_iter((0..NUM_ROWS).map(|i| Some(i as i64))))
+}
+
+fn mixed_size_array() -> ArrayRef {
+    Arc::new(Int64Array::from_iter(
+        (0..NUM_ROWS).map(|i| Some(if i % 2 == 0 { 200_i64 } else { 10_i64 })),
+    ))
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);
diff --git a/datafusion/functions-nested/src/resize.rs b/datafusion/functions-nested/src/resize.rs
@@ -198,24 +198,113 @@ fn general_list_resize<O: OffsetSizeTrait + TryInto<i64>>(
     let values = array.values();
     let original_data = values.to_data();
 
-    // create default element array
-    let default_element = if let Some(default_element) = default_element {
-        default_element
+    // Track the largest per-row growth so the uniform-fill fast path can
+    // materialize one reusable fill buffer of the required size.
+    let mut max_extra: usize = 0;
+    let mut output_values_len: usize = 0;
+    for (row_index, offset_window) in array.offsets().windows(2).enumerate() {
+        if array.is_null(row_index) {
+            continue;
+        }
+        let target_count = count_array.value(row_index).to_usize().ok_or_else(|| {
+            internal_datafusion_err!("array_resize: failed to convert size to usize")
+        })?;
+        output_values_len =
+            output_values_len.checked_add(target_count).ok_or_else(|| {
+                internal_datafusion_err!("array_resize: output size overflow")
+            })?;
+        let current_len = (offset_window[1] - offset_window[0]).to_usize().unwrap();
+        if target_count > current_len {
+            max_extra = max_extra.max(target_count - current_len);
+        }
+    }
+
+    // The fast path is valid when at least one row grows and every row would
+    // use the same fill value.
+    let use_bulk_fill = max_extra > 0
+        && match &default_element {
+            None => true,
+            Some(fill_array) => {
+                let len = fill_array.len();
+                let null_count = fill_array.logical_null_count();
+
+                len <= 1
+                    || null_count == len
+                    || (null_count == 0 && {
+                        let first = fill_array.slice(0, 1);
+                        (1..len)
+                            .all(|i| fill_array.slice(i, 1).as_ref() == first.as_ref())
+                    })
+            }
+        };
+
+    if use_bulk_fill {
+        // Fast path: materialize one reusable fill buffer for all grown rows.
+        let fill_scalar = match &default_element {
+            None => ScalarValue::try_from(&data_type)?,
+            Some(fill_array) if fill_array.logical_null_count() == fill_array.len() => {
+                ScalarValue::try_from(&data_type)?
+            }
+            Some(fill_array) => ScalarValue::try_from_array(fill_array.as_ref(), 0)?,
+        };
+        let fill_values = fill_scalar.to_array_of_size(max_extra)?;
+        let default_value_data = fill_values.to_data();
+        build_resized_list(
+            array,
+            count_array,
+            field,
+            &original_data,
+            &default_value_data,
+            output_values_len,
+            |mutable, _, extra_count| mutable.extend(1, 0, extra_count),
+        )
     } else {
-        let null_scalar = ScalarValue::try_from(&data_type)?;
-        null_scalar.to_array_of_size(original_data.len())?
-    };
-    let default_value_data = default_element.to_data();
+        // Slow path: rows may need different fill values, so append from the
+        // corresponding slot in the input fill array for each grown element.
+        let fill_values = match default_element {
+            Some(fill_values) => fill_values,
+            None => {
+                let null_scalar = ScalarValue::try_from(&data_type)?;
+                null_scalar.to_array_of_size(original_data.len())?
+            }
+        };
+        let default_value_data = fill_values.to_data();
+        build_resized_list(
+            array,
+            count_array,
+            field,
+            &original_data,
+            &default_value_data,
+            output_values_len,
+            |mutable, row_index, extra_count| {
+                for _ in 0..extra_count {
+                    mutable.extend(1, row_index, row_index + 1);
+                }
+            },
+        )
+    }
+}
 
-    // create a mutable array to store the original data
-    let capacity = Capacities::Array(original_data.len() + default_value_data.len());
+fn build_resized_list<O, F>(
+    array: &GenericListArray<O>,
+    count_array: &Int64Array,
+    field: &FieldRef,
+    original_data: &arrow::array::ArrayData,
+    default_value_data: &arrow::array::ArrayData,
+    output_values_len: usize,
+    mut append_fill_values: F,
+) -> Result<ArrayRef>
+where
+    O: OffsetSizeTrait + TryInto<i64>,
+    F: FnMut(&mut MutableArrayData, usize, usize),
+{
+    let capacity = Capacities::Array(output_values_len);
     let mut offsets = vec![O::usize_as(0)];
     let mut mutable = MutableArrayData::with_capacities(
-        vec![&original_data, &default_value_data],
+        vec![original_data, default_value_data],
         false,
         capacity,
     );
-
     let mut null_builder = NullBufferBuilder::new(array.len());
 
     for (row_index, offset_window) in array.offsets().windows(2).enumerate() {
@@ -232,21 +321,13 @@ fn general_list_resize<O: OffsetSizeTrait + TryInto<i64>>(
         let count = O::usize_as(count);
         let start = offset_window[0];
         if start + count > offset_window[1] {
-            let extra_count =
-                (start + count - offset_window[1]).try_into().map_err(|_| {
-                    internal_datafusion_err!(
-                        "array_resize: failed to convert size to i64"
-                    )
-                })?;
+            let extra_count = (start + count - offset_window[1]).to_usize().unwrap();
             let end = offset_window[1];
-            mutable.extend(0, (start).to_usize().unwrap(), (end).to_usize().unwrap());
-            // append default element
-            for _ in 0..extra_count {
-                mutable.extend(1, row_index, row_index + 1);
-            }
+            mutable.extend(0, start.to_usize().unwrap(), end.to_usize().unwrap());
+            append_fill_values(&mut mutable, row_index, extra_count);
         } else {
             let end = start + count;
-            mutable.extend(0, (start).to_usize().unwrap(), (end).to_usize().unwrap());
+            mutable.extend(0, start.to_usize().unwrap(), end.to_usize().unwrap());
         };
         offsets.push(offsets[row_index] + count);
     }
diff --git a/datafusion/sqllogictest/test_files/array/array_resize.slt b/datafusion/sqllogictest/test_files/array/array_resize.slt
@@ -156,5 +156,18 @@ NULL
 [51, 52, 53, 54, 55, NULL, 57, 58, 59, 60, NULL, NULL, NULL]
 [61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 7, 7, 7, 7, 7]
 
+# array_resize columnar test #3
+query ?
+select array_resize(column1, column2, 9) from array_resize_values;
+----
+[1, NULL]
+[11, 12, NULL, 14, 15]
+[21, 22, 23, 24, NULL, 26, 27, 28]
+[31, 32, 33, 34, 35, 36, NULL, 38, 39, 40, 9, 9]
+NULL
+[]
+[51, 52, 53, 54, 55, NULL, 57, 58, 59, 60, 9, 9, 9]
+[61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 9, 9, 9, 9, 9]
+
 
 include ./cleanup.slt.part
