ArraySort: support structs (apache#15527)

* ArraySort: support structs

* fix

* fix

* fix

* Update datafusion/functions-nested/src/sort.rs

---------

Co-authored-by: Andrew Lamb <[email protected]>

Author: cht42

datafusion/functions-nested/src/sort.rs

@@ -20,6 +20,7 @@
 use crate::utils::make_scalar_function;
 use arrow::array::{new_null_array, Array, ArrayRef, ListArray, NullBufferBuilder};
 use arrow::buffer::OffsetBuffer;
+use arrow::compute::SortColumn;
 use arrow::datatypes::DataType::{FixedSizeList, LargeList, List};
 use arrow::datatypes::{DataType, Field};
 use arrow::{compute, compute::SortOptions};
@@ -207,9 +208,24 @@ pub fn array_sort_inner(args: &[ArrayRef]) -> Result<ArrayRef> {
             valid.append_null();
         } else {
             let arr_ref = list_array.value(i);
-            let arr_ref = arr_ref.as_ref();
 
-            let sorted_array = compute::sort(arr_ref, sort_option)?;
+            // arrow sort kernel does not support Structs, so use
+            // lexsort_to_indices instead:
+            // https://github.com/apache/arrow-rs/issues/6911#issuecomment-2562928843
+            let sorted_array = match arr_ref.data_type() {
+                DataType::Struct(_) => {
+                    let sort_columns: Vec<SortColumn> = vec![SortColumn {
+                        values: Arc::clone(&arr_ref),
+                        options: sort_option,
+                    }];
+                    let indices = compute::lexsort_to_indices(&sort_columns, None)?;
+                    compute::take(arr_ref.as_ref(), &indices, None)?
+                }
+                _ => {
+                    let arr_ref = arr_ref.as_ref();
+                    compute::sort(arr_ref, sort_option)?
+                }
+            };
             array_lengths.push(sorted_array.len());
             arrays.push(sorted_array);
             valid.append_non_null();

datafusion/sqllogictest/test_files/array.slt

@@ -2396,6 +2396,11 @@ NULL NULL
 NULL NULL
 NULL NULL
 
+query ?
+select array_sort([struct('foo', 3), struct('foo', 1), struct('bar', 1)])
+----
+[{c0: bar, c1: 1}, {c0: foo, c1: 1}, {c0: foo, c1: 3}]
+
 ## test with argument of incorrect types
 query error DataFusion error: Execution error: the second parameter of array_sort expects DESC or ASC
 select array_sort([1, 3, null, 5, NULL, -5], 1), array_sort([1, 3, null, 5, NULL, -5], 'DESC', 1), array_sort([1, 3, null, 5, NULL, -5], 1, 1);