feat: add strictly_order_preserving for more expr

datafusion/functions-nested/src/make_array.rs

@@ -37,6 +37,7 @@ use datafusion_expr::{
     ColumnarValue, Documentation, ScalarFunctionArgs, ScalarUDFImpl, Signature,
     Volatility,
 };
+use datafusion_expr_common::sort_properties::ExprProperties;
 use datafusion_macros::user_doc;
 use itertools::Itertools as _;
 
@@ -113,6 +114,11 @@ impl ScalarUDFImpl for MakeArray {
         &self.aliases
     }
 
+    fn strictly_order_preserving(&self, _inputs: &[ExprProperties]) -> Result<bool> {
+        // Not strictly order preserving since null input does not correspond to null output
+        Ok(false)
+    }
+
     fn coerce_types(&self, arg_types: &[DataType]) -> Result<Vec<DataType>> {
         if arg_types.is_empty() {
             Ok(vec![])

datafusion/functions-nested/src/repeat.rs

@@ -37,6 +37,7 @@ use datafusion_expr::{
     Volatility,
 };
 use datafusion_expr_common::signature::{Coercion, TypeSignatureClass};
+use datafusion_expr_common::sort_properties::ExprProperties;
 use datafusion_macros::user_doc;
 use std::mem::size_of;
 use std::sync::Arc;
@@ -137,6 +138,12 @@ impl ScalarUDFImpl for ArrayRepeat {
         make_scalar_function(array_repeat_inner)(&args.args)
     }
 
+    fn strictly_order_preserving(&self, _inputs: &[ExprProperties]) -> Result<bool> {
+        // this is not strictly order preserving
+        // since array_repeat(NULL, 2) = [NULL, NULL] and not NULL
+        Ok(false)
+    }
+
     fn aliases(&self) -> &[String] {
         &self.aliases
     }

datafusion/functions/src/math/factorial.rs

@@ -28,6 +28,7 @@ use datafusion_expr::{
     ColumnarValue, Documentation, ScalarFunctionArgs, ScalarUDFImpl, Signature,
     Volatility,
 };
+use datafusion_expr_common::sort_properties::{ExprProperties, SortProperties};
 use datafusion_macros::user_doc;
 
 #[user_doc(
@@ -116,6 +117,17 @@ impl ScalarUDFImpl for FactorialFunc {
         }
     }
 
+    fn output_ordering(&self, inputs: &[ExprProperties]) -> Result<SortProperties> {
+        // Keep the same ordering as the input
+        Ok(inputs[0].sort_properties)
+    }
+
+    fn strictly_order_preserving(&self, _inputs: &[ExprProperties]) -> Result<bool> {
+        // There is one-to-one mapping between input and output and nulls maps to nulls
+        // and because overflow will be resulted in an error and not wrapping or saturating we are ok
+        Ok(true)
+    }
+
     fn documentation(&self) -> Option<&Documentation> {
         self.doc()
     }

datafusion/functions/src/math/floor.rs

@@ -234,6 +234,11 @@ impl ScalarUDFImpl for FloorFunc {
         Ok(input[0].sort_properties)
     }
 
+    fn strictly_order_preserving(&self, _inputs: &[ExprProperties]) -> Result<bool> {
+        // floor allow distinct inputs to collapse into equal outputs
+        Ok(false)
+    }
+
     fn evaluate_bounds(&self, inputs: &[&Interval]) -> Result<Interval> {
         let data_type = inputs[0].data_type();
         Interval::make_unbounded(&data_type)

datafusion/functions/src/string/repeat.rs

@@ -32,6 +32,7 @@ use datafusion_common::{
 use datafusion_expr::{ColumnarValue, Documentation, Volatility};
 use datafusion_expr::{ScalarFunctionArgs, ScalarUDFImpl, Signature};
 use datafusion_expr_common::signature::{Coercion, TypeSignatureClass};
+use datafusion_expr_common::sort_properties::ExprProperties;
 use datafusion_macros::user_doc;
 
 #[user_doc(
@@ -160,6 +161,12 @@ impl ScalarUDFImpl for RepeatFunc {
     fn documentation(&self) -> Option<&Documentation> {
         self.doc()
     }
+
+    fn strictly_order_preserving(&self, _inputs: &[ExprProperties]) -> Result<bool> {
+        // repeat does not strictly preserve ordering since for the following 2 rows:
+        // "bc" and "bca" and repeat(2) the result will be "bcbc" and "bcabca" which now the order is different
+        Ok(false)
+    }
 }
 
 /// Computes repeat for a single string value with max size check