[Data] Rename ParquetBaseDatasource to ParquetBulkDatasource

python/ray/data/datasource/__init__.py

@@ -42,7 +42,7 @@
 from ray.data.datasource.mongo_datasource import MongoDatasource
 from ray.data.datasource.numpy_datasink import _NumpyDatasink
 from ray.data.datasource.numpy_datasource import NumpyDatasource
-from ray.data.datasource.parquet_base_datasource import ParquetBaseDatasource
+from ray.data.datasource.parquet_bulk_datasource import ParquetBulkDatasource
 from ray.data.datasource.parquet_datasink import _ParquetDatasink
 from ray.data.datasource.parquet_datasource import ParquetDatasource
 from ray.data.datasource.partitioning import (
@@ -95,7 +95,7 @@
     "LanceDatasource",
     "_NumpyDatasink",
     "NumpyDatasource",
-    "ParquetBaseDatasource",
+    "ParquetBulkDatasource",
     "_ParquetDatasink",
     "ParquetDatasource",
     "ParquetMetadataProvider",

python/ray/data/datasource/parquet_base_datasource.py → python/ray/data/datasource/parquet_bulk_datasource.py

@@ -12,7 +12,7 @@
 
 
 @PublicAPI
-class ParquetBaseDatasource(FileBasedDatasource):
+class ParquetBulkDatasource(FileBasedDatasource):
     """Minimal Parquet datasource, for reading and writing Parquet files."""
 
     _FILE_EXTENSIONS = ["parquet"]

python/ray/data/datasource/parquet_datasource.py

@@ -186,7 +186,7 @@ def _check_for_legacy_tensor_type(schema):
 class ParquetDatasource(Datasource):
     """Parquet datasource, for reading and writing Parquet files.
 
-    The primary difference from ParquetBaseDatasource is that this uses
+    The primary difference from ParquetBulkDatasource is that this uses
     PyArrow's `ParquetDataset` abstraction for dataset reads, and thus offers
     automatic Arrow dataset schema inference and row count collection at the
     cost of some potential performance and/or compatibility penalties.
@@ -527,8 +527,8 @@ def _estimate_files_encoding_ratio(self) -> float:
 
     def get_name(self):
         """Return a human-readable name for this datasource.
+
         This will be used as the names of the read tasks.
-        Note: overrides the base `ParquetBaseDatasource` method.
         """
         return "Parquet"
 

python/ray/data/read_api.py

@@ -60,7 +60,7 @@
     LanceDatasource,
     MongoDatasource,
     NumpyDatasource,
-    ParquetBaseDatasource,
+    ParquetBulkDatasource,
     ParquetDatasource,
     ParquetMetadataProvider,
     PathPartitionFilter,
@@ -967,7 +967,7 @@ def read_parquet_bulk(
     partition_filter: Optional[PathPartitionFilter] = None,
     shuffle: Union[Literal["files"], None] = None,
     include_paths: bool = False,
-    file_extensions: Optional[List[str]] = ParquetBaseDatasource._FILE_EXTENSIONS,
+    file_extensions: Optional[List[str]] = ParquetBulkDatasource._FILE_EXTENSIONS,
     concurrency: Optional[int] = None,
     override_num_blocks: Optional[int] = None,
     **arrow_parquet_args,
@@ -1062,7 +1062,7 @@ def read_parquet_bulk(
     if columns is not None:
         read_table_args["columns"] = columns
 
-    datasource = ParquetBaseDatasource(
+    datasource = ParquetBulkDatasource(
         paths,
         read_table_args=read_table_args,
         filesystem=filesystem,

python/ray/data/tests/test_parquet.py

@@ -17,7 +17,7 @@
     DefaultFileMetadataProvider,
     DefaultParquetMetadataProvider,
 )
-from ray.data.datasource.parquet_base_datasource import ParquetBaseDatasource
+from ray.data.datasource.parquet_bulk_datasource import ParquetBulkDatasource
 from ray.data.datasource.parquet_datasource import (
     NUM_CPUS_FOR_META_FETCH_TASK,
     PARALLELIZE_META_FETCH_THRESHOLD,
@@ -1092,7 +1092,7 @@ def test_parquet_datasource_names(ray_start_regular_shared, tmp_path):
     path = os.path.join(tmp_path, "data.parquet")
     df.to_parquet(path)
 
-    assert ParquetBaseDatasource(path).get_name() == "ParquetBulk"
+    assert ParquetBulkDatasource(path).get_name() == "ParquetBulk"
     assert ParquetDatasource(path).get_name() == "Parquet"