Fix hill_formula Composition property (#3086)

* Added a test showing unexpected Hill formula behavior in a case with both C and H present, but not one after another (mp-1228185).

* Adjusted the hill_formula property to fit the specification / docstring.

Author: amkrajewski

pymatgen/core/composition.py

@@ -440,10 +440,16 @@ def hill_formula(self) -> str:
         """
         c = self.element_composition
         elements = sorted(el.symbol for el in c)
+        hill_elements = []
         if "C" in elements:
-            elements = ["C"] + [el for el in elements if el != "C"]
-
-        formula = [f"{el}{formula_double_format(c[el]) if c[el] != 1 else ''}" for el in elements]
+            hill_elements.append("C")
+            elements.remove("C")
+            if "H" in elements:
+                hill_elements.append("H")
+                elements.remove("H")
+        hill_elements += elements
+
+        formula = [f"{el}{formula_double_format(c[el]) if c[el] != 1 else ''}" for el in hill_elements]
         return " ".join(formula)
 
     @property

pymatgen/core/tests/test_composition.py

@@ -67,6 +67,12 @@ def test_hill_formula(self):
         assert c.hill_formula == "C Ca O3"
         c = Composition("C2H5OH")
         assert c.hill_formula == "C2 H6 O"
+        # A test case with both C and H, but not one after another (mp-1228185)
+        c = Composition("Ga8 As16 H102 C32 S36 O3")
+        assert c.hill_formula == "C32 H102 As16 Ga8 O3 S36"
+        # A test case with H but no C
+        c = Composition("Ga8 As16 H102 S36 O3")
+        assert c.hill_formula == "As16 Ga8 H102 O3 S36"
 
     def test_init_(self):
         with pytest.raises(ValueError):