feat: Sort kernel for RunArray (#3695)

* Handle sliced array in run array iterator

* sort_to_indices for RunArray

* better loop

* sort for run array

* improve docs

* some minor tweaks

* doc fix

* format fix

* fix sort run to return all logical indices

* pr comment

* rename test function, pull sort run logic into a separate function

---------

Co-authored-by: ask <ask@local>

Author: askoa

arrow-array/src/array/run_array.rs

@@ -112,17 +112,37 @@ impl<R: RunEndIndexType> RunArray<R> {
 
     /// Returns a reference to run_ends array
     ///
-    /// Note: any slicing of this array is not applied to the returned array
+    /// Note: any slicing of this [`RunArray`] array is not applied to the returned array
     /// and must be handled separately
     pub fn run_ends(&self) -> &PrimitiveArray<R> {
         &self.run_ends
     }
 
     /// Returns a reference to values array
+    ///
+    /// Note: any slicing of this [`RunArray`] array is not applied to the returned array
+    /// and must be handled separately
     pub fn values(&self) -> &ArrayRef {
         &self.values
     }
 
+    /// Returns the physical index at which the array slice starts.
+    pub fn get_start_physical_index(&self) -> usize {
+        if self.offset() == 0 {
+            return 0;
+        }
+        self.get_zero_offset_physical_index(self.offset()).unwrap()
+    }
+
+    /// Returns the physical index at which the array slice ends.
+    pub fn get_end_physical_index(&self) -> usize {
+        if self.offset() + self.len() == Self::logical_len(&self.run_ends) {
+            return self.run_ends.len() - 1;
+        }
+        self.get_zero_offset_physical_index(self.offset() + self.len() - 1)
+            .unwrap()
+    }
+
     /// Downcast this [`RunArray`] to a [`TypedRunArray`]
     ///
     /// ```
@@ -230,11 +250,7 @@ impl<R: RunEndIndexType> RunArray<R> {
         }
 
         // Skip some physical indices based on offset.
-        let skip_value = if self.offset() > 0 {
-            self.get_zero_offset_physical_index(self.offset()).unwrap()
-        } else {
-            0
-        };
+        let skip_value = self.get_start_physical_index();
 
         let mut physical_indices = vec![0; indices_len];
 

arrow-array/src/run_iterator.rs

@@ -57,13 +57,8 @@ where
 {
     /// create a new iterator
     pub fn new(array: TypedRunArray<'a, R, V>) -> Self {
-        let current_front_physical: usize =
-            array.run_array().get_physical_index(0).unwrap();
-        let current_back_physical: usize = array
-            .run_array()
-            .get_physical_index(array.len() - 1)
-            .unwrap()
-            + 1;
+        let current_front_physical = array.run_array().get_start_physical_index();
+        let current_back_physical = array.run_array().get_end_physical_index() + 1;
         RunArrayIter {
             array,
             current_front_logical: array.offset(),