BUG: fix to_datetime to handle int16 and int8

Fixes #13451

Author: ravinimmi

doc/source/whatsnew/v0.18.2.txt

@@ -517,3 +517,5 @@ Bug Fixes
 
 
 - Bug in ``Categorical.remove_unused_categories()`` changes ``.codes`` dtype to platform int (:issue:`13261`)
+
+- Bug in ``DataFrame.to_datetime()`` raises ValueError in case of dtype ``int8`` and ``int16`` (:issue:`13451`)

pandas/tseries/tests/test_timeseries.py

@@ -2563,6 +2563,33 @@ def test_dataframe(self):
         with self.assertRaises(ValueError):
             to_datetime(df2)
 
+    def test_dataframe_dtypes(self):
+        # #13451
+        df = DataFrame({'year': [2015, 2016],
+                        'month': [2, 3],
+                        'day': [4, 5]})
+
+        # int16
+        result = to_datetime(df.astype('int16'))
+        expected = Series([Timestamp('20150204 00:00:00'),
+                           Timestamp('20160305 00:00:00')])
+        assert_series_equal(result, expected)
+
+        # mixed dtypes
+        df['month'] = df['month'].astype('int8')
+        df['day'] = df['day'].astype('int8')
+        result = to_datetime(df)
+        expected = Series([Timestamp('20150204 00:00:00'),
+                           Timestamp('20160305 00:00:00')])
+        assert_series_equal(result, expected)
+
+        # float
+        df = DataFrame({'year': [2000, 2001],
+                        'month': [1.5, 1],
+                        'day': [1, 1]})
+        with self.assertRaises(ValueError):
+            to_datetime(df)
+
 
 class TestDatetimeIndex(tm.TestCase):
     _multiprocess_can_split_ = True

pandas/tseries/tools.py

@@ -508,7 +508,11 @@ def f(value):
 
     def coerce(values):
         # we allow coercion to if errors allows
-        return to_numeric(values, errors=errors)
+        values = to_numeric(values, errors=errors)
+        # prevent overflow in case of int8 or int16
+        if com.is_integer_dtype(values):
+            values = values.astype('int64', copy=False)
+        return values
 
     values = (coerce(arg[unit_rev['year']]) * 10000 +
               coerce(arg[unit_rev['month']]) * 100 +