diff --git a/py-polars/polars/dataframe/frame.py b/py-polars/polars/dataframe/frame.py
index 000649a969abe..fd2423bed8482 100644
--- a/py-polars/polars/dataframe/frame.py
+++ b/py-polars/polars/dataframe/frame.py
@@ -3315,7 +3315,11 @@ def write_delta(
         ... )  # doctest: +SKIP
 
         """
-        from polars.io.delta import _check_if_delta_available, _resolve_delta_lake_uri, _create_delta_compatible_schema
+        from polars.io.delta import (
+            _check_if_delta_available,
+            _create_delta_compatible_schema,
+            _resolve_delta_lake_uri,
+        )
 
         _check_if_delta_available()
 
diff --git a/py-polars/polars/io/delta.py b/py-polars/polars/io/delta.py
index d45f6c0fdb053..6b883372aa782 100644
--- a/py-polars/polars/io/delta.py
+++ b/py-polars/polars/io/delta.py
@@ -1,16 +1,16 @@
 from __future__ import annotations
 
-import pyarrow as pa
+from functools import reduce
 from pathlib import Path
 from typing import TYPE_CHECKING, Any
 from urllib.parse import urlparse
-from functools import reduce
+
+import pyarrow as pa
 
 from polars.convert import from_arrow
 from polars.dependencies import _DELTALAKE_AVAILABLE, deltalake
 from polars.io.pyarrow_dataset import scan_pyarrow_dataset
 
-
 if TYPE_CHECKING:
     from polars import DataFrame, LazyFrame
 
@@ -319,48 +319,62 @@ def _check_if_delta_available() -> None:
 
 
 def _reconstruct_field_type(
-    field: pa.Field, field_head: pa.Field, reconstructed_field=None
+    field: pa.Field,
+    field_head: pa.Field,
+    reconstructed_field: list | None = None,
 ) -> pa.Field:
-    """Recursive function that traverses through pyArrow fields to change timestamps to US and reconstructing  it later.
+    """
+    Recursive function that traverses through pyArrow fields.
 
-    Args:
-        field (pa.Field): _description_
-        field_head (pa.Field): field head to retain column names
-        reconstructed_field (List, optional): . Defaults to None.
+    Parameters
+    ----------
+    field : pa.Field
+        field to analyze
+    field_head : pa.Field
+        field head to retain column names
+    reconstructed_field : List, optional
+        Collection of dtypes to reconstruct a field, by default None
 
-    Returns:
-        pa.Field: Reconstructed pyArrow field
+    Returns
+    -------
+    pa.Field
+        Reconstructed field
     """
     if isinstance(field.type, pa.TimestampType):
         if reconstructed_field is None:
             return pa.field(
                 name=field.name,
-                type=pa.timestamp("us"
-                                #   , tz=field.type.tz
-                                  ),
+                type=pa.timestamp(
+                    "us"
+                    #   , tz=field.type.tz
+                ),
             )
         else:
-            reconstructed_field.append(pa.timestamp("us", 
-                                                    # tz=field.type.tz
-                                                    ))
+            reconstructed_field.append(
+                pa.timestamp(
+                    "us",
+                    # tz=field.type.tz
+                )
+            )
             return pa.field(
                 name=field_head.name,
                 type=reduce(lambda x, y: y(x), reversed(reconstructed_field)),
             )
-    elif 'uint' in str(field.type) and not any(isinstance(field.type, dtype) for dtype in [pa.StructType, pa.LargeListType]):
+    elif "uint" in str(field.type) and not any(
+        isinstance(field.type, dtype) for dtype in [pa.StructType, pa.LargeListType]
+    ):
         if reconstructed_field is None:
             return pa.field(
-                name=field.name,
-                type=getattr(pa, str(field.type).strip('u'))()
+                name=field.name, type=getattr(pa, str(field.type).strip("u"))()
             )
         else:
-            reconstructed_field.append(getattr(pa, str(field.type).strip('u'))())
+            reconstructed_field.append(getattr(pa, str(field.type).strip("u"))())
             return pa.field(
                 name=field_head.name,
                 type=reduce(lambda x, y: y(x), reversed(reconstructed_field)),
             )
     elif isinstance(field.type, pa.LargeListType) and (
-        "timestamp" in str(field.type) or 'uint' in str(field.type)
+        "timestamp" in str(field.type) or "uint" in str(field.type)
     ):  # some checks to just skip traversing non timestamps and US timestamps
         if reconstructed_field is None:
             reconstructed_field = [pa.large_list]
@@ -391,15 +405,20 @@ def _reconstruct_field_type(
 
 
 def _create_delta_compatible_schema(schema: pa.schema) -> pa.Schema:
-    """Makes the dataframe schema compatible with Delta
+    """
+    Makes the dataframe schema compatible with Delta lake protocol.
 
-    Args:
-        schema (pa.schema): input schema
+    Parameters
+    ----------
+    schema : pa.schema
+        input schema
 
-    Returns:
-        pa.Schema: delta compatible schema
+    Returns
+    -------
+    pa.Schema
+        delta compatible schema
     """
     schema_out = [*map(_reconstruct_field_type, schema, schema)]
     schema = pa.schema(schema_out, metadata=schema.metadata)
 
-    return schema
\ No newline at end of file
+    return schema