ray 2.10.0 breaks import of awswrangler

[tleonhardt]

Describe the bug

ray 2.10 was released on March 21. Using this with awswrangler 3.7.1 causes awswrangler to fail to import due to use of a deprecated functionality.

How to Reproduce

import awswrangler as wr

Output:

---------------------------------------------------------------------------
DeprecationWarning                        Traceback (most recent call last)
Cell In[4], line 1
----> 1 import awswrangler as wr

File [/usr/local/lib/python3.9/site-packages/awswrangler/__init__.py:40](/usr/local/lib/python3.9/site-packages/awswrangler/__init__.py#line=39)
     37 from awswrangler._config import config
     38 from awswrangler._distributed import EngineEnum, MemoryFormatEnum, engine, memory_format
---> 40 engine.register()
     42 __all__ = [
     43     "athena",
     44     "catalog",
   (...)
     75     "MemoryFormatEnum",
     76 ]
     79 _logging.getLogger("awswrangler").addHandler(_logging.NullHandler())

File [/usr/local/lib/python3.9/site-packages/awswrangler/_distributed.py:121](/usr/local/lib/python3.9/site-packages/awswrangler/_distributed.py#line=120), in Engine.register(cls, name)
    118 if engine_name == EngineEnum.RAY.value:
    119     from awswrangler.distributed.ray._register import register_ray
--> 121     register_ray()

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/_register.py:73](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/_register.py#line=72), in register_ray()
     66 from awswrangler.distributed.ray.modin._data_types import pyarrow_types_from_pandas_distributed
     67 from awswrangler.distributed.ray.modin._utils import (
     68     _arrow_refs_to_df,
     69     _copy_modin_df_shallow,
     70     _is_pandas_or_modin_frame,
     71     _split_modin_frame,
     72 )
---> 73 from awswrangler.distributed.ray.modin.s3._read_orc import _read_orc_distributed
     74 from awswrangler.distributed.ray.modin.s3._read_parquet import _read_parquet_distributed
     75 from awswrangler.distributed.ray.modin.s3._read_text import _read_text_distributed

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/modin/s3/_read_orc.py:13](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/modin/s3/_read_orc.py#line=12)
     10 from ray.data.datasource import FastFileMetadataProvider
     12 from awswrangler import _data_types
---> 13 from awswrangler.distributed.ray.datasources import ArrowORCDatasource
     14 from awswrangler.distributed.ray.modin._utils import _to_modin
     16 if TYPE_CHECKING:

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/__init__.py:3](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/__init__.py#line=2)
      1 """Ray Datasources Module."""
----> 3 from awswrangler.distributed.ray.datasources.arrow_csv_datasink import ArrowCSVDatasink
      4 from awswrangler.distributed.ray.datasources.arrow_csv_datasource import ArrowCSVDatasource
      5 from awswrangler.distributed.ray.datasources.arrow_json_datasource import ArrowJSONDatasource

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/arrow_csv_datasink.py:12](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/arrow_csv_datasink.py#line=11)
      9 from ray.data.block import BlockAccessor
     10 from ray.data.datasource.block_path_provider import BlockWritePathProvider
---> 12 from awswrangler.distributed.ray.datasources.file_datasink import _BlockFileDatasink
     14 _logger: logging.Logger = logging.getLogger(__name__)
     17 class ArrowCSVDatasink(_BlockFileDatasink):

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/file_datasink.py:22](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/file_datasink.py#line=21)
     17 from awswrangler.s3._write import _COMPRESSION_2_EXT
     19 _logger: logging.Logger = logging.getLogger(__name__)
---> 22 class _BlockFileDatasink(Datasink):
     23     def __init__(
     24         self,
     25         path: str,
   (...)
     32         **write_args: Any,
     33     ):
     34         self.path = path

File [/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/file_datasink.py:28](/usr/local/lib/python3.9/site-packages/awswrangler/distributed/ray/datasources/file_datasink.py#line=27), in _BlockFileDatasink()
     22 class _BlockFileDatasink(Datasink):
     23     def __init__(
     24         self,
     25         path: str,
     26         file_format: str,
     27         *,
---> 28         block_path_provider: BlockWritePathProvider | None = DefaultBlockWritePathProvider(),
     29         dataset_uuid: str | None = None,
     30         open_s3_object_args: dict[str, Any] | None = None,
     31         pandas_kwargs: dict[str, Any] | None = None,
     32         **write_args: Any,
     33     ):
     34         self.path = path
     35         self.file_format = file_format

File [/usr/local/lib64/python3.9/site-packages/ray/data/datasource/block_path_provider.py:14](/usr/local/lib64/python3.9/site-packages/ray/data/datasource/block_path_provider.py#line=13), in BlockWritePathProvider.__init__(self)
     13 def __init__(self) -> None:
---> 14     raise DeprecationWarning(
     15         "`BlockWritePathProvider` has been deprecated in favor of "
     16         "`FilenameProvider`. For more information, see "
     17         "https://docs.ray.io/en/master/data/api/doc/ray.data.datasource.FilenameProvider.html",  # noqa: E501
     18     )

DeprecationWarning: `BlockWritePathProvider` has been deprecated in favor of `FilenameProvider`. For more information, see https://docs.ray.io/en/master/data/api/doc/ray.data.datasource.FilenameProvider.html

Click to add a cell.

Expected behavior

Importing awswrangler still works

Your project

No response

Screenshots

No response

OS

Linux

Python version

3.9

AWS SDK for pandas version

3.7.1

Additional context

No response

[kukushking]

Thanks @tleonhardt looks like BlockWritePathProvider was deprecated in favour of FilenameProvider in Ray 2.10. Working on it.