BUG: fix TypeError for invalid integer dates %Y%m%d with errors='ignore' (# GH 26583)

array_strptime returned TypeError when trying to slice 'too long' integer for the given format %Y%m%d (for example 2121010101).
After parsing date in the first 8 symbols it tried to return the remaining symbols in ValueError message as a slice of integer which in turn caused TypeError.
Converted to string value is now used to make slice for that ValueError message.
In case of 20209911, it tried to parse 20209911 to datetime(2020, 9, 9) and had 11 unparsed.

Author: nathalier

doc/source/whatsnew/v0.25.0.rst

@@ -427,6 +427,7 @@ Datetimelike
 - Bug in :class:`Series` and :class:`DataFrame` repr where ``np.datetime64('NaT')`` and ``np.timedelta64('NaT')`` with ``dtype=object`` would be represented as ``NaN`` (:issue:`25445`)
 - Bug in :func:`to_datetime` which does not replace the invalid argument with ``NaT`` when error is set to coerce (:issue:`26122`)
 - Bug in adding :class:`DateOffset` with nonzero month to :class:`DatetimeIndex` would raise ``ValueError`` (:issue:`26258`)
+- Bug in :func:`to_datetime` which raises ``TypeError`` for ``format='%Y%m%d'`` when called for invalid integer dates with length >= 6 digits with ``errors='ignore'``
 
 Timedelta
 ^^^^^^^^^

pandas/_libs/tslibs/strptime.pyx

@@ -140,13 +140,13 @@ def array_strptime(object[:] values, object fmt,
                     iresult[i] = NPY_NAT
                     continue
                 raise ValueError("time data %r does not match "
-                                 "format %r (match)" % (values[i], fmt))
+                                 "format %r (match)" % (val, fmt))
             if len(val) != found.end():
                 if is_coerce:
                     iresult[i] = NPY_NAT
                     continue
                 raise ValueError("unconverted data remains: %s" %
-                                 values[i][found.end():])
+                                 val[found.end():])
 
         # search
         else:
@@ -156,7 +156,7 @@ def array_strptime(object[:] values, object fmt,
                     iresult[i] = NPY_NAT
                     continue
                 raise ValueError("time data %r does not match format "
-                                 "%r (search)" % (values[i], fmt))
+                                 "%r (search)" % (val, fmt))
 
         iso_year = -1
         year = 1900

pandas/tests/indexes/datetimes/test_tools.py

@@ -114,6 +114,25 @@ def test_to_datetime_format_integer(self, cache):
         result = to_datetime(s, format='%Y%m', cache=cache)
         assert_series_equal(result, expected)
 
+    @pytest.mark.parametrize('int_date, expected', [
+        # valid date, length == 8
+        [20121030, datetime(2012, 10, 30)],
+        # short valid date, length == 6
+        [199934, datetime(1999, 3, 4)],
+        # long integer date partially parsed to datetime(2012,1,1), length > 8
+        [2012010101, 2012010101],
+        # invalid date partially parsed to datetime(2012,9,9), length == 8
+        [20129930, 20129930],
+        # short integer date partially parsed to datetime(2012,9,9), length < 8
+        [2012993, 2012993],
+        # short invalid date, length == 4
+        [2121, 2121]])
+    def test_int_to_datetime_format_YYYYMMDD_typeerror(self, int_date,
+                                                       expected):
+        # GH 26583
+        result = to_datetime(int_date, format='%Y%m%d', errors='ignore')
+        assert result == expected
+
     @pytest.mark.parametrize('cache', [True, False])
     def test_to_datetime_format_microsecond(self, cache):