FIX: handle potential TypeError when determining variable type (#3516)

* FIX: handle potential TypeError when determining variable type

* Handle edge case in duplicated _core code

* Add tests

---------

Co-authored-by: Michael Waskom <mwaskom@gmail.com>

seaborn/_base.py

@@ -1520,8 +1520,13 @@ def variable_type(vector, boolean_type="numeric"):
         warnings.simplefilter(
             action='ignore', category=(FutureWarning, DeprecationWarning)
         )
-        if np.isin(vector, [0, 1]).all():
-            return VariableType(boolean_type)
+        try:
+            if np.isin(vector, [0, 1]).all():
+                return VariableType(boolean_type)
+        except TypeError:
+            # .isin comparison is not guaranteed to be possible under NumPy
+            # casting rules, depending on the (unknown) dtype of 'vector'
+            pass
 
     # Defer to positive pandas tests
     if pd.api.types.is_numeric_dtype(vector):

seaborn/_core/rules.py

@@ -97,7 +97,12 @@ def variable_type(
                 boolean_dtypes = ["bool"]
             boolean_vector = vector.dtype in boolean_dtypes
         else:
-            boolean_vector = bool(np.isin(vector, [0, 1]).all())
+            try:
+                boolean_vector = bool(np.isin(vector, [0, 1]).all())
+            except TypeError:
+                # .isin comparison is not guaranteed to be possible under NumPy
+                # casting rules, depending on the (unknown) dtype of 'vector'
+                boolean_vector = False
         if boolean_vector:
             return VarType(boolean_type)
 

tests/_core/test_rules.py

@@ -52,6 +52,11 @@ def test_variable_type():
     assert variable_type(s, boolean_type="categorical") == "categorical"
     assert variable_type(s, boolean_type="boolean") == "boolean"
 
+    # This should arguably be datmetime, but we don't currently handle it correctly
+    # Test is mainly asserting that this doesn't fail on the boolean check.
+    s = pd.timedelta_range(1, periods=3, freq="D").to_series()
+    assert variable_type(s) == "categorical"
+
     s_cat = s.astype("category")
     assert variable_type(s_cat, boolean_type="categorical") == "categorical"
     assert variable_type(s_cat, boolean_type="numeric") == "categorical"
@@ -61,6 +66,9 @@ def test_variable_type():
     assert variable_type(s, boolean_type="boolean") == "boolean"
     assert variable_type(s, boolean_type="boolean", strict_boolean=True) == "numeric"
 
+    s = pd.Series([1, 0, 0])
+    assert variable_type(s, boolean_type="boolean") == "boolean"
+
     s = pd.Series([pd.Timestamp(1), pd.Timestamp(2)])
     assert variable_type(s) == "datetime"
     assert variable_type(s.astype(object)) == "datetime"

tests/test_base.py

@@ -1508,6 +1508,11 @@ def test_variable_type(self):
         assert variable_type(s.to_numpy()) == "categorical"
         assert variable_type(s.to_list()) == "categorical"
 
+        # This should arguably be datmetime, but we don't currently handle it correctly
+        # Test is mainly asserting that this doesn't fail on the boolean check.
+        s = pd.timedelta_range(1, periods=3, freq="D").to_series()
+        assert variable_type(s) == "categorical"
+
         s = pd.Series([True, False, False])
         assert variable_type(s) == "numeric"
         assert variable_type(s, boolean_type="categorical") == "categorical"