feat(data-types): Support PyArrow ExtensionTypes in pyarrow2athena (#3351)

* Update _data_types.py

* Update test_s3_parquet.py

---------

Co-authored-by: Anton Kukushkin <kukushkin.anton@gmail.com>

Author: richacode007-byte

awswrangler/_data_types.py

@@ -59,6 +59,8 @@ def pyarrow2athena(  # noqa: PLR0911,PLR0912
         )
     if pa.types.is_map(dtype):
         return f"map<{pyarrow2athena(dtype=dtype.key_type, ignore_null=ignore_null)},{pyarrow2athena(dtype=dtype.item_type, ignore_null=ignore_null)}>"
+    if isinstance(dtype, getattr(pa, "BaseExtensionType", pa.ExtensionType)):
+        return pyarrow2athena(dtype=dtype.storage_type, ignore_null=ignore_null)
     if dtype == pa.null():
         if ignore_null:
             return ""

tests/unit/test_s3_parquet.py

@@ -84,6 +84,42 @@ def test_read_parquet_metadata_large_dtype(path):
     assert columns_types.get("c1") == "string"
 
 
+def test_pyarrow2athena_uuid_extension_type():
+    from awswrangler._data_types import pyarrow2athena
+
+    assert pyarrow2athena(pa.uuid()) == "binary"
+
+
+def test_pyarrow2athena_custom_extension_type():
+    from awswrangler._data_types import pyarrow2athena
+
+    class _TestExtType(pa.ExtensionType):
+        def __init__(self):
+            super().__init__(pa.int64(), "test.custom_ext")
+
+        def __arrow_ext_serialize__(self):
+            return b""
+
+        @classmethod
+        def __arrow_ext_deserialize__(cls, storage_type, serialized):
+            return cls()
+
+    assert pyarrow2athena(_TestExtType()) == "bigint"
+
+
+def test_athena_types_from_pyarrow_schema_with_extension():
+    from awswrangler._data_types import athena_types_from_pyarrow_schema
+
+    schema = pa.schema(
+        [
+            pa.field("id", pa.uuid()),
+            pa.field("value", pa.int64()),
+        ]
+    )
+    result = athena_types_from_pyarrow_schema(schema)
+    assert result == {"id": "binary", "value": "bigint"}
+
+
 @pytest.mark.parametrize(
     "partition_cols",
     [