Support Beam in Croissant preparation.

PiperOrigin-RevId: 679042687

tensorflow_datasets/scripts/cli/build.py

@@ -356,6 +356,7 @@ def _download_and_prepare(
       publish_dir=args.publish_dir,
       skip_if_published=args.skip_if_published,
       overwrite=args.overwrite,
+      beam_pipeline_options=args.beam_pipeline_options,
   )
 
 
@@ -384,7 +385,7 @@ def _make_download_config(
   if args.update_metadata_only:
     kwargs['download_mode'] = tfds.download.GenerateMode.UPDATE_DATASET_INFO
 
-  dl_config = tfds.download.DownloadConfig(
+  return tfds.download.DownloadConfig(
       extract_dir=args.extract_dir,
       manual_dir=manual_dir,
       max_examples_per_split=args.max_examples_per_split,
@@ -393,20 +394,6 @@ def _make_download_config(
       **kwargs,
   )
 
-  # Add Apache Beam options to download config
-  try:
-    import apache_beam as beam  # pylint: disable=g-import-not-at-top
-  except ImportError:
-    beam = None
-
-  if beam is not None:
-    if args.beam_pipeline_options:
-      dl_config.beam_options = beam.options.pipeline_options.PipelineOptions(
-          flags=[f'--{opt}' for opt in args.beam_pipeline_options.split(',')]
-      )
-
-  return dl_config
-
 
 def _get_config_name(
     builder_cls: Type[tfds.core.DatasetBuilder],

tensorflow_datasets/scripts/cli/cli_utils.py

@@ -299,6 +299,7 @@ def download_and_prepare(
     publish_dir: epath.Path | None,
     skip_if_published: bool,
     overwrite: bool,
+    beam_pipeline_options: str | None,
 ) -> None:
   """Generate a single builder."""
   dataset = builder.info.full_name
@@ -312,6 +313,22 @@ def download_and_prepare(
       )
       return
 
+  if not download_config:
+    download_config = download.DownloadConfig()
+
+  # Add Apache Beam options to download config
+  try:
+    import apache_beam as beam  # pylint: disable=g-import-not-at-top
+
+    if beam_pipeline_options:
+      download_config.beam_options = (
+          beam.options.pipeline_options.PipelineOptions(
+              flags=[f'--{opt}' for opt in beam_pipeline_options.split(',')]
+          )
+      )
+  except ImportError:
+    pass
+
   builder.download_and_prepare(
       download_dir=download_dir,
       download_config=download_config,

tensorflow_datasets/scripts/cli/croissant.py

@@ -155,6 +155,7 @@ def prepare_croissant_builder(
       publish_dir=args.publish_dir,
       skip_if_published=args.skip_if_published,
       overwrite=args.overwrite,
+      beam_pipeline_options=None,
   )
   return builder