Fix Atoms.cluster_from_file() in io.feff.inputs giving wrong numb…

…er of atoms (#3426)

* Update test_sets.py

* add test file

* fix a bug in inputs.py

* fix AttributeError: 'list' object has no attribute 'startswith'

        for line in lines:
>           if line and not line.startswith("*"):

* gzip tests/files/ATOMS_test.inp

* improve test

---------

Signed-off-by: KaifengZheng <[email protected]>
Co-authored-by: Janosh Riebesell <[email protected]>

pymatgen/io/feff/inputs.py

@@ -482,13 +482,13 @@ def cluster_from_file(filename):
                 is the one at the origin.
         """
         atoms_string = Atoms.atoms_string_from_file(filename)
-        lines = [line.split() for line in atoms_string.splitlines()[3:]]
+        lines = [line.split() for line in atoms_string.splitlines()[1:]]
         coords = []
         symbols = []
-        for line in lines:
-            if line:
-                coords.append([float(val) for val in line[:3]])
-                symbols.append(line[4])
+        for tokens in lines:
+            if tokens and not tokens[0].startswith("*"):
+                coords.append([float(val) for val in tokens[:3]])
+                symbols.append(tokens[4])
         return Molecule(symbols, coords)
 
     def get_lines(self) -> list[list[str | int]]:

tests/io/feff/test_inputs.py

@@ -137,6 +137,12 @@ def test_cluster_from_file(self):
         assert len(mol_1) == len(mol_2)
         os.remove("ATOMS_test")
 
+    def test_atom_num(self):
+        filepath = f"{TEST_FILES_DIR}/Pt37_atoms.inp.gz"
+        atoms = Atoms.cluster_from_file(filepath)
+        assert len(atoms) == 37
+        assert atoms.formula == "Pt37"
+
 
 class TestFeffTags(unittest.TestCase):
     def test_init(self):