chore: import upstream snapshot with attribution

2026-07-13 13:17:40 +08:00
commit f1825c8ceb
10096 changed files with 2364182 additions and 0 deletions
@@ -0,0 +1,78 @@
+from ray.data._internal.datasource.delta_sharing_datasource import (
+    DeltaSharingDatasource,
+)
+from ray.data._internal.datasource.mcap_datasource import (
+    MCAPDatasource,
+    TimeRange,
+)
+from ray.data._internal.datasource.sql_datasource import Connection
+from ray.data._internal.datasource.turbopuffer_datasink import (
+    TurbopufferDatasink,
+)
+from ray.data._internal.savemode import SaveMode
+from ray.data.datasource.datasink import (
+    Datasink,
+    DummyOutputDatasink,
+    WriteResult,
+    WriteReturnType,
+)
+from ray.data.datasource.datasource import (
+    Datasource,
+    RandomIntRowDatasource,
+    Reader,
+    ReadTask,
+)
+from ray.data.datasource.file_based_datasource import (
+    FileBasedDatasource,
+    FileShuffleConfig,
+    _S3FileSystemWrapper,
+)
+from ray.data.datasource.file_datasink import (
+    BlockBasedFileDatasink,
+    RowBasedFileDatasink,
+)
+from ray.data.datasource.file_meta_provider import (
+    BaseFileMetadataProvider,
+    DefaultFileMetadataProvider,
+    FileMetadataProvider,
+)
+from ray.data.datasource.filename_provider import FilenameProvider
+from ray.data.datasource.partitioning import (
+    Partitioning,
+    PartitionStyle,
+    PathPartitionFilter,
+    PathPartitionParser,
+)
+
+# Note: HuggingFaceDatasource should NOT be imported here, because
+# we want to only import the Hugging Face datasets library when we use
+# ray.data.from_huggingface() or HuggingFaceDatasource() directly.
+__all__ = [
+    "BaseFileMetadataProvider",
+    "Connection",
+    "Datasink",
+    "Datasource",
+    "DefaultFileMetadataProvider",
+    "DeltaSharingDatasource",
+    "DummyOutputDatasink",
+    "FileBasedDatasource",
+    "FileShuffleConfig",
+    "FileMetadataProvider",
+    "FilenameProvider",
+    "MCAPDatasource",
+    "PartitionStyle",
+    "PathPartitionFilter",
+    "PathPartitionParser",
+    "Partitioning",
+    "RandomIntRowDatasource",
+    "ReadTask",
+    "Reader",
+    "RowBasedFileDatasink",
+    "TurbopufferDatasink",
+    "BlockBasedFileDatasink",
+    "_S3FileSystemWrapper",
+    "TimeRange",
+    "WriteResult",
+    "WriteReturnType",
+    "SaveMode",
+]