fix(qe): honor angstrom cells without cel files

dpdata/formats/qe/traj.py

@@ -66,11 +66,30 @@ def convert_celldm(ibrav, celldm):
 
 
 def load_cell_parameters(lines):
-    blk = load_block(lines, "CELL_PARAMETERS", 3)
-    ret = []
-    for ii in blk:
-        ret.append([float(jj) for jj in ii.split()[0:3]])
-    return np.array(ret)
+    """Load ``CELL_PARAMETERS`` and convert its vectors to angstrom.
+
+    CP trajectory ``.cel`` files always use atomic units, but the fallback
+    cell in the QE input file follows the unit declared on the
+    ``CELL_PARAMETERS`` card.  Keeping that distinction here prevents an
+    angstrom input cell from being converted a second time when no ``.cel``
+    file is available.
+    """
+    for idx, line in enumerate(lines):
+        # Ignore commented examples such as ``!CELL_PARAMETERS {bohr}``, which
+        # are common in QE inputs and must not shadow the active card below.
+        card = line.split("!", 1)[0].strip()
+        if card.upper().startswith("CELL_PARAMETERS"):
+            blk = lines[idx + 1 : idx + 4]
+            break
+    else:
+        raise ValueError("CELL_PARAMETERS is required when ibrav is 0")
+
+    cell = np.array([[float(value) for value in row.split()[:3]] for row in blk])
+    # Preserve the historical Bohr conversion for the default/Bohr spelling.
+    # An explicit angstrom card, however, is already in dpdata's length unit.
+    if "angstrom" not in card.lower():
+        cell *= length_convert
+    return cell
 
 
 def load_atom_names(lines, ntypes):
@@ -111,8 +130,8 @@ def load_param_file(fname: FileType):
     if ibrav == 0:
         cell = load_cell_parameters(lines)
     else:
-        cell = convert_celldm(ibrav, celldm)
-    cell = cell * length_convert
+        # celldm and cells reconstructed from it are expressed in Bohr.
+        cell = convert_celldm(ibrav, celldm) * length_convert
     # print(atom_names)
     # print(atom_numbs)
     # print(atom_types)

tests/qe.traj/angstrom_no_cel/cp.in

@@ -0,0 +1,20 @@
+&CONTROL
+  calculation = 'cp',
+  prefix = 'cp',
+/
+&SYSTEM
+  ibrav = 0,
+  nat = 1,
+  ntyp = 1,
+/
+
+CELL_PARAMETERS { angstrom }
+  19.7299995422   0.0000000000   0.0000000000
+   0.0000000000  19.7299995422   0.0000000000
+   0.0000000000   0.0000000000  19.7299995422
+
+ATOMIC_SPECIES
+H  1.00794  H.UPF
+
+ATOMIC_POSITIONS { bohr }
+H  0.0  0.0  0.0

tests/qe.traj/angstrom_no_cel/cp.pos

@@ -0,0 +1,2 @@
+0 0.0
+0.0 0.0 0.0

tests/qe.traj/bohr_no_cel/cp.in

@@ -0,0 +1,20 @@
+&CONTROL
+  calculation = 'cp',
+  prefix = 'cp',
+/
+&SYSTEM
+  ibrav = 0,
+  nat = 1,
+  ntyp = 1,
+/
+
+CELL_PARAMETERS { bohr }
+  2.0  0.0  0.0
+  0.0  2.0  0.0
+  0.0  0.0  2.0
+
+ATOMIC_SPECIES
+H  1.00794  H.UPF
+
+ATOMIC_POSITIONS { bohr }
+H  0.0  0.0  0.0

tests/qe.traj/bohr_no_cel/cp.pos

@@ -0,0 +1,2 @@
+0 0.0
+0.0 0.0 0.0

tests/qe.traj/missing_cell_no_cel/cp.in

@@ -0,0 +1,15 @@
+&CONTROL
+  calculation = 'cp',
+  prefix = 'cp',
+/
+&SYSTEM
+  ibrav = 0,
+  nat = 1,
+  ntyp = 1,
+/
+
+ATOMIC_SPECIES
+H  1.00794  H.UPF
+
+ATOMIC_POSITIONS { bohr }
+H  0.0  0.0  0.0

tests/test_qe_cp_traj.py

@@ -61,6 +61,43 @@ def setUp(self):
         self.system = dpdata.LabeledSystem("qe.traj/oh-md", fmt="qe/cp/traj")
 
 
+class TestCPTRAJInputCellUnits(unittest.TestCase):
+    def test_angstrom_cell_without_cel_trajectory(self):
+        # Earlier tests did not expose this regression: the only missing-.cel
+        # fixture used ibrav/celldm, whose cell is correctly expressed in Bohr.
+        # Exercise the distinct fallback path where CELL_PARAMETERS already
+        # stores angstrom values and therefore must not be converted again.
+        system = dpdata.System(
+            "qe.traj/angstrom_no_cel/cp",
+            fmt="qe/cp/traj",
+        )
+
+        np.testing.assert_allclose(
+            system["cells"][0],
+            np.eye(3) * 19.7299995422,
+        )
+
+    def test_bohr_cell_without_cel_trajectory(self):
+        system = dpdata.System(
+            "qe.traj/bohr_no_cel/cp",
+            fmt="qe/cp/traj",
+        )
+
+        np.testing.assert_allclose(
+            system["cells"][0],
+            np.eye(3) * 2.0 * bohr2ang,
+        )
+
+    def test_missing_cell_parameters_for_ibrav_zero_raises(self):
+        with self.assertRaisesRegex(
+            ValueError, "CELL_PARAMETERS is required when ibrav is 0"
+        ):
+            dpdata.System(
+                "qe.traj/missing_cell_no_cel/cp",
+                fmt="qe/cp/traj",
+            )
+
+
 class TestConverCellDim(unittest.TestCase):
     def test_case_null(self):
         cell = convert_celldm(8, [1, 1, 1])