Fix highly-nested parens when formula parsing in Composition (#3569)

* fix highly-nested formula parsing in Composition

* add failure cases from gh-3559 as tests

Author: janosh

pymatgen/core/composition.py

@@ -558,15 +558,16 @@ def get_sym_dict(form: str, factor: float) -> dict[str, float]:
                 raise ValueError(f"{form} is an invalid formula!")
             return sym_dict
 
-        m = re.search(r"\(([^\(\)]+)\)\s*([\.e\d]*)", formula)
-        if m:
+        match = re.search(r"\(([^\(\)]+)\)\s*([\.e\d]*)", formula)
+        while match:
             factor = 1.0
-            if m.group(2) != "":
-                factor = float(m.group(2))
-            unit_sym_dict = get_sym_dict(m.group(1), factor)
+            if match.group(2) != "":
+                factor = float(match.group(2))
+            unit_sym_dict = get_sym_dict(match.group(1), factor)
             expanded_sym = "".join(f"{el}{amt}" for el, amt in unit_sym_dict.items())
-            expanded_formula = formula.replace(m.group(), expanded_sym)
-            return self._parse_formula(expanded_formula)
+            expanded_formula = formula.replace(match.group(), expanded_sym, 1)
+            formula = expanded_formula
+            match = re.search(r"\(([^\(\)]+)\)\s*([\.e\d]*)", formula)
         return get_sym_dict(formula, 1)
 
     @property
@@ -581,14 +582,14 @@ def anonymized_formula(self) -> str:
             reduced /= gcd(*(int(i) for i in self.values()))
 
         anon = ""
-        for e, amt in zip(string.ascii_uppercase, sorted(reduced.values())):
+        for elem, amt in zip(string.ascii_uppercase, sorted(reduced.values())):
             if amt == 1:
                 amt_str = ""
             elif abs(amt % 1) < 1e-8:
                 amt_str = str(int(amt))
             else:
                 amt_str = str(amt)
-            anon += f"{e}{amt_str}"
+            anon += f"{elem}{amt_str}"
         return anon
 
     @property

tests/core/test_composition.py

@@ -186,6 +186,16 @@ def test_formula(self):
 
         assert Composition("Na 3 Zr (PO 4) 3").reduced_formula == "Na3Zr(PO4)3"
 
+        # gh-3559
+        very_nested_formula = (
+            "(Bi2(Mg0.667Nb1.333)O7)((Bi2(Mg0.667Nb1.333)O7)0.9(SrCO3)0.1)((Bi2(Mg0.667Nb1.333)O7)0.7(SrCO3)0.3)"
+        )
+        assert Composition(very_nested_formula).formula == "Sr0.4 Mg1.7342 Nb3.4658 Bi5.2 C0.4 O19.4"
+        assert Composition(very_nested_formula) == Composition(
+            "(Bi2(Mg0.667Nb1.333)O7)1((Bi2(Mg0.667Nb1.333)O7)0.9(SrCO3)0.1)((Bi2(Mg0.667Nb1.333)O7)0.7(SrCO3)0.3)"
+        )
+        assert Composition("(C)((C)0.9(B)0.1)") == Composition("C1.9 B0.1")
+
         assert Composition("NaN").reduced_formula == "NaN"
         with pytest.raises(ValueError, match=r"float\('NaN'\) is not a valid Composition, did you mean str\('NaN'\)\?"):
             Composition(float("NaN"))