Use reference rather than owned ArrayRef

Author: alamb

datafusion/physical-expr/src/aggregate/count_distinct/strings.rs

@@ -48,7 +48,7 @@ impl<O: OffsetSizeTrait> Accumulator for StringDistinctCountAccumulator<O> {
             return Ok(());
         }
 
-        self.0.insert(values[0].clone());
+        self.0.insert(&values[0]);
 
         Ok(())
     }
@@ -66,7 +66,7 @@ impl<O: OffsetSizeTrait> Accumulator for StringDistinctCountAccumulator<O> {
         let arr = as_list_array(&states[0])?;
         arr.iter().try_for_each(|maybe_list| {
             if let Some(list) = maybe_list {
-                self.0.insert(list);
+                self.0.insert(&list);
             };
             Ok(())
         })

datafusion/physical-expr/src/string_map.rs

@@ -42,7 +42,7 @@ impl<O: OffsetSizeTrait> ArrowStringSet<O> {
     }
 
     /// Inserts each string from `values` into the set
-    pub fn insert(&mut self, values: ArrayRef) {
+    pub fn insert(&mut self, values: &ArrayRef) {
         fn make_payload_fn(_value: Option<&[u8]>) {}
         fn observe_payload_fn(_payload: ()) {}
         self.0
@@ -173,7 +173,7 @@ where
     /// Safety: F is only ever invoked with valid values from `values`
     pub fn insert_if_new<MP, OP>(
         &mut self,
-        values: ArrayRef,
+        values: &ArrayRef,
         mut make_payload_fn: MP,
         mut observe_payload_fn: OP,
     ) where
@@ -440,7 +440,8 @@ mod tests {
     #[test]
     fn string_set_empty() {
         let mut set = ArrowStringSet::<i32>::new();
-        set.insert(Arc::new(StringArray::new_null(0)));
+        let array: ArrayRef = Arc::new(StringArray::new_null(0));
+        set.insert(&array);
         assert_eq!(set.len(), 0);
         assert_eq!(set.non_null_len(), 0);
         assert_set(set, &[]);
@@ -449,7 +450,8 @@ mod tests {
     #[test]
     fn string_set_one_null() {
         let mut set = ArrowStringSet::<i32>::new();
-        set.insert(Arc::new(StringArray::new_null(1)));
+        let array: ArrayRef = Arc::new(StringArray::new_null(1));
+        set.insert(&array);
         assert_eq!(set.len(), 1);
         assert_eq!(set.non_null_len(), 0);
         assert_set(set, &[None]);
@@ -458,7 +460,8 @@ mod tests {
     #[test]
     fn string_set_many_null() {
         let mut set = ArrowStringSet::<i32>::new();
-        set.insert(Arc::new(StringArray::new_null(11)));
+        let array: ArrayRef = Arc::new(StringArray::new_null(11));
+        set.insert(&array);
         assert_eq!(set.len(), 1);
         assert_eq!(set.non_null_len(), 0);
         assert_set(set, &[None]);
@@ -496,7 +499,8 @@ mod tests {
         ]);
 
         let mut set = ArrowStringSet::<O>::new();
-        set.insert(Arc::new(values));
+        let array: ArrayRef = Arc::new(values);
+        set.insert(&array);
         // values mut appear be in the order they were inserted
         assert_set(
             set,
@@ -534,7 +538,8 @@ mod tests {
         ]);
 
         let mut set = ArrowStringSet::<O>::new();
-        set.insert(Arc::new(values));
+        let array: ArrayRef = Arc::new(values);
+        set.insert(&array);
         // strings mut appear be in the order they were inserted
         assert_set(
             set,
@@ -587,7 +592,7 @@ mod tests {
         let mut set = ArrowStringSet::<i32>::new();
         let size_empty = set.size();
 
-        set.insert(values1.clone());
+        set.insert(&values1);
         let size_after_values1 = set.size();
         assert!(size_empty < size_after_values1);
         assert!(
@@ -597,11 +602,11 @@ mod tests {
         assert!(size_after_values1 < total_strings1_len + total_strings2_len);
 
         // inserting the same strings should not affect the size
-        set.insert(values1.clone());
+        set.insert(&values1);
         assert_eq!(set.size(), size_after_values1);
 
         // inserting the large strings should increase the reported size
-        set.insert(values2);
+        set.insert(&values2);
         let size_after_values2 = set.size();
         assert!(size_after_values2 > size_after_values1);
         assert!(size_after_values2 > total_strings1_len + total_strings2_len);

datafusion/physical-plan/src/aggregates/group_values/binary.rs

@@ -57,7 +57,7 @@ impl GroupValues for GroupValuesBinary {
 
         groups.clear();
         self.map.insert_if_new(
-            arr.clone(),
+            arr,
             // called for each new group
             |_value| {
                 // assign new group index on each insert