Include trust_remote_code flag for code-based HF builders preparation.

The TensorFlow Datasets Authors · The TensorFlow Datasets Authors · commit 69c097c4361e · 2024-08-21T01:49:04.000-07:00
PiperOrigin-RevId: 661246526
diff --git a/tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder.py b/tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder.py
@@ -200,6 +200,7 @@ def __init__(
       tfds_num_proc: Optional[int] = None,
       ignore_hf_errors: bool = False,
       overwrite_version: str | None = None,
+      trust_remote_code: bool = False,
       **config_kwargs,
   ):
     self._hf_repo_id = hf_repo_id
@@ -210,7 +211,10 @@ def __init__(
     )
     try:
       self._hf_builder = hf_datasets.load_dataset_builder(
-          self._hf_repo_id, self._hf_config, **self.config_kwargs
+          self._hf_repo_id,
+          self._hf_config,
+          trust_remote_code=trust_remote_code,
+          **self.config_kwargs,
       )
     except Exception as e:
       raise RuntimeError(
diff --git a/tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder_test.py b/tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder_test.py
@@ -104,7 +104,10 @@ def mock_huggingface_dataset_builder(
       other_arg='this is another arg',
   )
   load_dataset_builder.assert_called_once_with(
-      'foo/bar', 'config', other_arg='this is another arg'
+      'foo/bar',
+      'config',
+      other_arg='this is another arg',
+      trust_remote_code=False,
   )
   login_to_hf.assert_called_once_with('SECRET_TOKEN')
   yield builder

Original file line number	Diff line number	Diff line change
`@@ -104,7 +104,10 @@ def mock_huggingface_dataset_builder(`
`104`	`104`	`other_arg='this is another arg',`
`105`	`105`	`)`
`106`	`106`	`load_dataset_builder.assert_called_once_with(`
`107`		`- 'foo/bar', 'config', other_arg='this is another arg'`
	`107`	`+ 'foo/bar',`
	`108`	`+ 'config',`
	`109`	`+ other_arg='this is another arg',`
	`110`	`+ trust_remote_code=False,`
`108`	`111`	`)`
`109`	`112`	`login_to_hf.assert_called_once_with('SECRET_TOKEN')`
`110`	`113`	`yield builder`