Raise ValueError when tz-aware differs in input

Just like `to_datetime` when `format` is not passed, we should raise a
`ValueError` if the input contains some tz-aware objects and some
naive objects. The change is to identify if the `format` expects
tz-aware objects (check for `"%z"` and `"%Z"`) and then compare this
expectation with the `tzinfo` of each object. This modification
includes relevant tests.

Signed-off-by: Antonio Ossa Guerra <aaossa@uc.cl>

Author: aaossa

pandas/_libs/tslibs/strptime.pyx

@@ -128,6 +128,7 @@ def array_strptime(ndarray[object] values, str fmt, bint exact=True, errors='rai
     result_timezone = np.empty(n, dtype='object')
 
     dts.us = dts.ps = dts.as = 0
+    expect_tz_aware = "%z" in fmt or "%Z" in fmt
 
     for i in range(n):
         val = values[i]
@@ -144,6 +145,10 @@ def array_strptime(ndarray[object] values, str fmt, bint exact=True, errors='rai
             else:
                 iresult[i] = pydatetime_to_dt64(val, &dts)
                 check_dts_bounds(&dts)
+            if val.tzinfo is None and expect_tz_aware:
+                raise ValueError("Cannot mix tz-aware with tz-naive values")
+            elif val.tzinfo is not None and not expect_tz_aware:
+                raise ValueError("Cannot mix tz-aware with tz-naive values")
             result_timezone[i] = val.tzinfo
             continue
         else:

pandas/tests/tools/test_to_datetime.py

@@ -680,10 +680,27 @@ def test_to_datetime_includes_tz_dtype_on_pydatetime_and_timestamp(self, value):
         # Localized value
         america_santiago = pytz.timezone("America/Santiago")
         result_no_format = to_datetime([america_santiago.localize(value)])
-        result_with_format = to_datetime([america_santiago.localize(value)], format="%m-%d-%Y")
+        result_with_format = to_datetime([america_santiago.localize(value)], format="%m-%d-%Y %z")
         tm.assert_equal(result_with_format.dtype.tz, america_santiago)
         tm.assert_equal(result_no_format, result_with_format)
 
+    @pytest.mark.parametrize("value", [datetime(2010, 1, 2, 12, 13, 16), Timestamp("2010-01-02 12:13:17")])
+    def test_to_datetime_mixing_naive_tzaware_raises(self, value):
+        # GH 49298
+        msg = "Cannot mix tz-aware with tz-naive values"
+        america_santiago = pytz.timezone("America/Santiago")
+        # Fail if format expects tz but input is not localized
+        with pytest.raises(ValueError, match=msg):
+            to_datetime([value], format="%m-%d-%Y %z")
+        # Fail if format does not expect tz but input is localized
+        with pytest.raises(ValueError, match=msg):
+            to_datetime([america_santiago.localize(value)], format="%m-%d-%Y")
+        # Mixed input should fail in both cases
+        with pytest.raises(ValueError, match=msg):
+            to_datetime([value, america_santiago.localize(value)], format="%m-%d-%Y %z")
+        with pytest.raises(ValueError, match=msg):
+            to_datetime([value, america_santiago.localize(value)], format="%m-%d-%Y")
+
     def test_to_datetime_pydatetime(self):
         actual = to_datetime(datetime(2008, 1, 15))
         assert actual == datetime(2008, 1, 15)