fix(deepmd): expose unlabeled mixed-system loading

dpdata/system.py

@@ -1481,9 +1481,32 @@ def __add__(self, others):
         raise RuntimeError("Unspported data structure")
 
     @classmethod
-    def from_file(cls, file_name, fmt: str, **kwargs: Any):
+    def from_file(
+        cls, file_name, fmt: str, *, labeled: bool = True, **kwargs: Any
+    ) -> MultiSystems:
+        """Load multiple systems from a file or directory.
+
+        Parameters
+        ----------
+        file_name
+            Source accepted by the selected format backend.
+        fmt : str
+            Format identifier, such as ``"deepmd/npy/mixed"``.
+        labeled : bool, default=True
+            Load :class:`LabeledSystem` objects when true. Set this to false for
+            coordinate-only data that does not contain energies or forces.
+        **kwargs
+            Additional arguments forwarded to the format backend.
+
+        Returns
+        -------
+        MultiSystems
+            Systems reconstructed from the source.
+        """
         multi_systems = cls()
-        multi_systems.load_systems_from_file(file_name=file_name, fmt=fmt, **kwargs)
+        multi_systems.load_systems_from_file(
+            file_name=file_name, fmt=fmt, labeled=labeled, **kwargs
+        )
         return multi_systems
 
     @classmethod
@@ -1506,10 +1529,31 @@ def from_dir(
             )
         return multi_systems
 
-    def load_systems_from_file(self, file_name=None, fmt: str | None = None, **kwargs):
+    def load_systems_from_file(
+        self,
+        file_name=None,
+        fmt: str | None = None,
+        *,
+        labeled: bool = True,
+        **kwargs,
+    ):
+        """Load systems into this collection.
+
+        ``labeled=False`` selects regular :class:`System` objects, which is
+        required for DeepMD datasets that omit label arrays.
+        """
         assert fmt is not None
         fmt = fmt.lower()
-        return self.from_fmt_obj(load_format(fmt), file_name, **kwargs)
+        try:
+            return self.from_fmt_obj(
+                load_format(fmt), file_name, labeled=labeled, **kwargs
+            )
+        except DataError as exc:
+            if labeled and fmt in {"deepmd/npy/mixed", "deepmd/hdf5/mixed"}:
+                raise DataError(
+                    f"{exc} For coordinate-only mixed datasets, pass labeled=False."
+                ) from exc
+            raise
 
     def get_nframes(self) -> int:
         """Returns number of frames in all systems."""

tests/test_deepmd_mixed.py

@@ -2,8 +2,10 @@
 
 import os
 import shutil
+import tempfile
 import unittest
 from glob import glob
+from inspect import Parameter, signature
 
 import numpy as np
 from comp_sys import (
@@ -17,10 +19,51 @@
 
 from dpdata.data_type import (
     Axis,
+    DataError,
     DataType,
 )
 
 
+class TestMixedMultiSystemsUnlabeled(unittest.TestCase):
+    """Regression coverage for loading coordinate-only mixed datasets."""
+
+    def test_from_file_unlabeled_round_trip(self):
+        labeled_parameter = signature(dpdata.MultiSystems.from_file).parameters[
+            "labeled"
+        ]
+        self.assertIs(labeled_parameter.kind, Parameter.KEYWORD_ONLY)
+        self.assertIs(labeled_parameter.default, True)
+
+        system = dpdata.System("poscars/POSCAR.h2o.md", fmt="vasp/poscar")
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            mixed_dir = os.path.join(tmpdir, "mixed")
+            dpdata.MultiSystems(system).to("deepmd/npy/mixed", mixed_dir)
+
+            # The default remains labeled for backward compatibility, but the
+            # error now points coordinate-only users to the public flag.
+            with self.assertRaisesRegex(DataError, "pass labeled=False"):
+                dpdata.MultiSystems.from_file(
+                    mixed_dir,
+                    fmt="deepmd/npy/mixed",
+                )
+
+            # The explicit, discoverable flag selects ordinary System objects.
+            systems = dpdata.MultiSystems.from_file(
+                mixed_dir, fmt="deepmd/npy/mixed", labeled=False
+            )
+
+        self.assertEqual(len(systems), 1)
+        loaded = next(iter(systems.systems.values()))
+        self.assertIs(type(loaded), dpdata.System)
+        self.assertNotIn("energies", loaded.data)
+        np.testing.assert_array_equal(
+            loaded.data["atom_types"], system.data["atom_types"]
+        )
+        np.testing.assert_allclose(loaded.data["cells"], system.data["cells"])
+        np.testing.assert_allclose(loaded.data["coords"], system.data["coords"])
+
+
 class TestMixedMultiSystemsDumpLoad(
     unittest.TestCase, CompLabeledMultiSys, MultiSystems, MSAllIsNoPBC
 ):