fix(dataset): raise a clear error on an empty dataset entry

swift/dataset/dataset_syntax.py

@@ -55,6 +55,11 @@ def _safe_split(s: str,
     @classmethod
     def parse(cls, dataset: str) -> 'DatasetSyntax':
         """Parse the dataset from the command line"""
+        if dataset:
+            dataset = dataset.strip()
+        if not dataset:
+            raise ValueError('Received an empty dataset entry. Check `--dataset`/`--val_dataset` for a stray empty '
+                             'or whitespace-only string (e.g. an unset shell variable wrapped in quotes).')
         # hf/ms::dataset_id or dataset_path:subset1/subset2/subset3#dataset_sample
         if os.path.exists(dataset):
             use_hf = None

tests/general/test_dataset.py

@@ -79,6 +79,20 @@ def test_cls():
     _test_dataset(['simpleai/HC3-Chinese:baike_cls'])
 
 
+def test_dataset_syntax_rejects_empty_entry():
+    import pytest
+
+    from swift.dataset.dataset_syntax import DatasetSyntax
+
+    # An empty or whitespace-only dataset entry (e.g. an unset shell variable
+    # wrapped in quotes in a `--dataset "$DS1" "$DS2"` invocation) must raise a
+    # clear ValueError instead of crashing deep inside with a cryptic
+    # `os.path.exists(None)` TypeError.
+    for entry in ('', '   ', '\t', '\n'):
+        with pytest.raises(ValueError, match='empty dataset entry'):
+            DatasetSyntax.parse(entry)
+
+
 if __name__ == '__main__':
     # test_sft()
     # test_agent()