fix(abacus): reject conflicting duplicate species definitions

get_frame_from_stru merges repeated atom labels and keeps the metadata
of the first matching row for masses, pp_files, and orb_files. A
conflicting duplicate would otherwise be silently discarded on dump.

Add validate_duplicate_species() to check duplicate rows for consistency
and raise a clear error on conflict. Add STRU-conflict-atomtype.ch4 test
data and regression tests covering the read path and all three metadata
fields.

Author: pxlxingliang

dpdata/formats/abacus/stru.py

@@ -441,6 +441,46 @@ def right_hand_rule(
     return cell, coord
 
 
+def validate_duplicate_species(atom_names, masses, pp_files, orb_files):
+    """Check that duplicate atom species definitions are consistent.
+
+    get_frame_from_stru merges repeated labels and keeps the metadata of the
+    first matching row for masses, pp_files, and orb_files. If a duplicate
+    row conflicts with the first one, raise an error instead of silently
+    discarding the conflicting metadata.
+
+    Args:
+        atom_names (list): list of atom names.
+        masses (list): list of atomic masses.
+        pp_files (list): list of pseudo potential files.
+        orb_files (list): list of orbital files.
+
+    Raises
+    ------
+        RuntimeError: if duplicate species have conflicting metadata.
+    """
+    for name in dict.fromkeys(atom_names):
+        indices = [j for j in range(len(atom_names)) if atom_names[j] == name]
+        if len(indices) < 2:
+            continue
+        ref_mass = masses[indices[0]]
+        ref_pp = pp_files[indices[0]]
+        ref_orb = orb_files[indices[0]] if orb_files else None
+        for j in indices[1:]:
+            if not np.isclose(masses[j], ref_mass):
+                raise RuntimeError(
+                    f"Conflicting duplicate species '{name}': mass {masses[j]} != {ref_mass}"
+                )
+            if pp_files[j] != ref_pp:
+                raise RuntimeError(
+                    f"Conflicting duplicate species '{name}': pp_file {pp_files[j]} != {ref_pp}"
+                )
+            if ref_orb is not None and orb_files[j] != ref_orb:
+                raise RuntimeError(
+                    f"Conflicting duplicate species '{name}': orb_file {orb_files[j]} != {ref_orb}"
+                )
+
+
 def get_frame_from_stru(stru):
     """Read the ABACUS STRU file and return the dpdata frame.
 
@@ -492,6 +532,8 @@ def get_frame_from_stru(stru):
         blocks["ATOMIC_POSITIONS"], atom_names, celldm, cell
     )
 
+    validate_duplicate_species(atom_names, masses, pp_files, orb_files)
+
     cell, coords = right_hand_rule(cell, coords)
     uniq_name = []
     uniq_atom_num = []

tests/abacus.scf/STRU-conflict-atomtype.ch4

@@ -0,0 +1,37 @@
+#This is the atom card containing all the information
+#about the lattice structure.
+
+ATOMIC_SPECIES
+C 1.000 C_ONCV_PBE-1.0.upf 	#Element, Mass, Pseudopotential
+H 1.000 H_ONCV_PBE-1.0.upf 
+H 2.000 H_ONCV_PBE-1.0.upf 
+
+NUMERICAL_ORBITAL
+c.orb
+h.orb
+h.orb
+
+LATTICE_CONSTANT
+10  					#Lattice constant
+
+LATTICE_VECTORS
+1 0.0 0.0 			#Lattice vector 1
+0.0 1 0.0 			#Lattice vector 2
+0.0 0.0 1 			#Lattice vector 3
+
+ATOMIC_POSITIONS
+Cartesian 				#Cartesian(Unit is LATTICE_CONSTANT)
+C 						#Name of element	
+0.0						#Magnetic for this element.
+1				#Number of atoms
+0.981274803    0.861285385    0.838442496 1 1 1
+H
+0.0
+2
+1.023557202    0.758025625    0.66351336 0 0 0
+0.78075702    0.889445935    0.837363468 1 0 1
+H
+0.0
+2
+1.064091613    1.043438905    0.840995502 1 0 1
+1.039321214    0.756530859    1.009609207 0 1 1

tests/test_abacus_stru_dump.py

@@ -287,6 +287,44 @@ def test_dump_stru_with_repeat_atomtype(self):
         self.assertTrue("H\n0.0\n4\n" in c)
         os.remove("STRU_tmp")
 
+    def test_read_stru_with_conflicting_duplicate_atomtype(self):
+        with self.assertRaisesRegex(
+            RuntimeError,
+            "Conflicting duplicate species 'H'.*mass 2.0 != 1.0",
+        ):
+            dpdata.System("abacus.scf/STRU-conflict-atomtype.ch4", fmt="stru")
+
+    def test_validate_duplicate_species(self):
+        from dpdata.formats.abacus.stru import validate_duplicate_species
+
+        validate_duplicate_species(
+            ["C", "H", "H"],
+            [1.0, 1.0, 1.0],
+            ["C.upf", "H.upf", "H.upf"],
+            ["c.orb", "h.orb", "h.orb"],
+        )
+        with self.assertRaisesRegex(RuntimeError, "mass"):
+            validate_duplicate_species(
+                ["C", "H", "H"],
+                [1.0, 1.0, 2.0],
+                ["C.upf", "H.upf", "H.upf"],
+                ["c.orb", "h.orb", "h.orb"],
+            )
+        with self.assertRaisesRegex(RuntimeError, "pp_file"):
+            validate_duplicate_species(
+                ["C", "H", "H"],
+                [1.0, 1.0, 1.0],
+                ["C.upf", "H.upf", "O.upf"],
+                ["c.orb", "h.orb", "h.orb"],
+            )
+        with self.assertRaisesRegex(RuntimeError, "orb_file"):
+            validate_duplicate_species(
+                ["C", "H", "H"],
+                [1.0, 1.0, 1.0],
+                ["C.upf", "H.upf", "H.upf"],
+                ["c.orb", "h.orb", "o.orb"],
+            )
+
 
 class TestABACUSParseStru(unittest.TestCase):
     def test_parse_pos_oneline(self):