Merge branch 'main' of https://github.com/barak1412/polars into expos…

…e_regex_escape

crates/polars-arrow/src/array/list/mod.rs

@@ -130,6 +130,49 @@ impl<O: Offset> ListArray<O> {
     impl_sliced!();
     impl_mut_validity!();
     impl_into_array!();
+
+    pub fn trim_to_normalized_offsets_recursive(&self) -> Self {
+        let offsets = self.offsets();
+        let values = self.values();
+
+        let first_idx = *offsets.first();
+        let len = offsets.range().to_usize();
+
+        if first_idx.to_usize() == 0 && values.len() == len {
+            return self.clone();
+        }
+
+        let offsets = if first_idx.to_usize() == 0 {
+            offsets.clone()
+        } else {
+            let v = offsets.iter().map(|x| *x - first_idx).collect::<Vec<_>>();
+            unsafe { OffsetsBuffer::<O>::new_unchecked(v.into()) }
+        };
+
+        let values = values.sliced(first_idx.to_usize(), len);
+
+        let values = match values.dtype() {
+            ArrowDataType::List(_) => {
+                let inner: &ListArray<i32> = values.as_ref().as_any().downcast_ref().unwrap();
+                Box::new(inner.trim_to_normalized_offsets_recursive()) as Box<dyn Array>
+            },
+            ArrowDataType::LargeList(_) => {
+                let inner: &ListArray<i64> = values.as_ref().as_any().downcast_ref().unwrap();
+                Box::new(inner.trim_to_normalized_offsets_recursive()) as Box<dyn Array>
+            },
+            _ => values,
+        };
+
+        assert_eq!(offsets.first().to_usize(), 0);
+        assert_eq!(values.len(), offsets.range().to_usize());
+
+        Self::new(
+            self.dtype().clone(),
+            offsets,
+            values,
+            self.validity().cloned(),
+        )
+    }
 }
 
 // Accessors

crates/polars-arrow/src/compute/aggregate/sum.rs

@@ -6,7 +6,7 @@ use polars_error::PolarsResult;
 use crate::array::{Array, PrimitiveArray};
 use crate::bitmap::utils::{BitChunkIterExact, BitChunksExact};
 use crate::bitmap::Bitmap;
-use crate::datatypes::{ArrowDataType, PhysicalType, PrimitiveType};
+use crate::datatypes::PhysicalType;
 use crate::scalar::*;
 use crate::types::simd::*;
 use crate::types::NativeType;
@@ -102,19 +102,6 @@ where
     }
 }
 
-/// Whether [`sum`] supports `dtype`
-pub fn can_sum(dtype: &ArrowDataType) -> bool {
-    if let PhysicalType::Primitive(primitive) = dtype.to_physical_type() {
-        use PrimitiveType::*;
-        matches!(
-            primitive,
-            Int8 | Int16 | Int64 | Int128 | UInt8 | UInt16 | UInt32 | UInt64 | Float32 | Float64
-        )
-    } else {
-        false
-    }
-}
-
 /// Returns the sum of all elements in `array` as a [`Scalar`] of the same physical
 /// and logical types as `array`.
 /// # Error