Fix column order issue in cast

src/datasets/arrow_dataset.py

@@ -600,18 +600,19 @@ def cast_(self, features: Features):
 
         Args:
             features (:class:`datasets.Features`): New features to cast the dataset to.
-                The name and order of the fields in the features must match the current column names.
+                The name of the fields in the features must match the current column names.
                 The type of the data must also be convertible from one type to the other.
                 For non-trivial conversion, e.g. string <-> ClassLabel you should use :func:`map` to update the Dataset.
         """
-        if list(features) != self._data.column_names:
+        if sorted(features) != sorted(self._data.column_names):
             raise ValueError(
-                f"The columns in features ({list(features)}) must be identical and in the same order "
+                f"The columns in features ({list(features)}) must be identical "
                 f"as the columns in the dataset: {self._data.column_names}"
             )
 
         self._info.features = features
-        schema = pa.schema(features.type)
+        type = features.type
+        schema = pa.schema({col_name: type[col_name].type for col_name in self._data.column_names})
         self._data = self._data.cast(schema)
 
     @fingerprint(inplace=True)

tests/test_arrow_dataset.py

@@ -49,7 +49,7 @@ def reduce_ex(self):
 
         datasets.arrow_dataset.logger.__reduce_ex__ = reduce_ex
 
-    def _create_dummy_dataset(self, in_memory: bool, tmp_dir: str, multiple_columns=False):
+    def _create_dummy_dataset(self, in_memory: bool, tmp_dir: str, multiple_columns=False) -> Dataset:
         if multiple_columns:
             data = {"col_1": [3, 2, 1, 0], "col_2": ["a", "b", "c", "d"]}
             dset = Dataset.from_dict(data)
@@ -289,6 +289,7 @@ def test_cast_(self, in_memory):
             dset = self._create_dummy_dataset(in_memory, tmp_dir, multiple_columns=True)
             features = dset.features
             features["col_1"] = Value("float64")
+            features = Features({k: features[k] for k in list(features)[::-1]})
             fingerprint = dset._fingerprint
             dset.cast_(features)
             self.assertEqual(dset.num_columns, 2)