fixup

Author: MarcoGorelli

doc/source/whatsnew/v2.0.0.rst

@@ -64,7 +64,8 @@ Other enhancements
 - :func:`date_range` now supports a ``unit`` keyword ("s", "ms", "us", or "ns") to specify the desired resolution of the output index (:issue:`49106`)
 - :func:`timedelta_range` now supports a ``unit`` keyword ("s", "ms", "us", or "ns") to specify the desired resolution of the output index (:issue:`49824`)
 - :meth:`DataFrame.to_json` now supports a ``mode`` keyword with supported inputs 'w' and 'a'. Defaulting to 'w', 'a' can be used when lines=True and orient='records' to append record oriented json lines to an existing json file. (:issue:`35849`)
-- :func:`to_datetime` now skips ``datetime.datetime`` and :class:`Timestamp` objects when passing ``format`` argument instead of raising a ``ValueError``. (:issue:`49298`)
+- Added ``name`` parameter to :meth:`IntervalIndex.from_breaks`, :meth:`IntervalIndex.from_arrays` and :meth:`IntervalIndex.from_tuples` (:issue:`48911`)
+-
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_200.notable_bug_fixes:
@@ -631,6 +632,7 @@ Datetimelike
 - Bug in subtracting a ``datetime`` scalar from :class:`DatetimeIndex` failing to retain the original ``freq`` attribute (:issue:`48818`)
 - Bug in ``pandas.tseries.holiday.Holiday`` where a half-open date interval causes inconsistent return types from :meth:`USFederalHolidayCalendar.holidays` (:issue:`49075`)
 - Bug in rendering :class:`DatetimeIndex` and :class:`Series` and :class:`DataFrame` with timezone-aware dtypes with ``dateutil`` or ``zoneinfo`` timezones near daylight-savings transitions (:issue:`49684`)
+- Bug in :func:`to_datetime` was raising ``ValueError`` when parsing :class:`Timestamp` or ``datetime`` objects with non-ISO8601 ``format`` (:issue:`49298`)
 -
 
 Timedelta

pandas/_libs/tslibs/strptime.pyx

@@ -20,6 +20,7 @@ from numpy cimport (
 )
 
 from pandas._libs.missing cimport checknull_with_nat_and_na
+from pandas._libs.tslibs.conversion cimport convert_timezone
 from pandas._libs.tslibs.nattype cimport (
     NPY_NAT,
     c_nat_strings as nat_strings,
@@ -59,7 +60,13 @@ cdef dict _parse_code_table = {'y': 0,
                                'u': 22}
 
 
-def array_strptime(ndarray[object] values, str fmt, bint exact=True, errors='raise'):
+def array_strptime(
+    ndarray[object] values,
+    str fmt,
+    bint exact=True,
+    errors='raise',
+    bint utc=False,
+):
     """
     Calculates the datetime structs represented by the passed array of strings
 
@@ -84,6 +91,9 @@ def array_strptime(ndarray[object] values, str fmt, bint exact=True, errors='rai
         bint is_raise = errors=='raise'
         bint is_ignore = errors=='ignore'
         bint is_coerce = errors=='coerce'
+        bint found_naive = False
+        bint found_tz = False
+        tzinfo tz_out = None
 
     assert is_raise or is_ignore or is_coerce
 
@@ -127,7 +137,6 @@ 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]
@@ -139,15 +148,22 @@ def array_strptime(ndarray[object] values, str fmt, bint exact=True, errors='rai
             iresult[i] = NPY_NAT
             continue
         elif PyDateTime_Check(val):
+            if val.tzinfo is not None:
+                found_tz = True
+            else:
+                found_naive = True
+            tz_out = convert_timezone(
+                val.tzinfo,
+                tz_out,
+                found_naive,
+                found_tz,
+                utc,
+            )
             if isinstance(val, _Timestamp):
-                iresult[i] = val.tz_localize(None)._as_unit("ns").value
+                iresult[i] = val.tz_localize(None).as_unit("ns").value
             else:
-                iresult[i] = pydatetime_to_dt64(val, &dts)
+                iresult[i] = pydatetime_to_dt64(val.replace(tzinfo=None), &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/core/tools/datetimes.py

@@ -318,7 +318,14 @@ def _return_parsed_timezone_results(
     )
     if utc:
         # Convert to the same tz
-        tz_results = np.array([tz_result.tz_convert("utc") for tz_result in tz_results])
+        tz_results = np.array(
+            [
+                tz_result.tz_convert("utc")
+                if tz_result.tzinfo is not None
+                else tz_result.tz_localize("utc")
+                for tz_result in tz_results
+            ]
+        )
 
     return Index(tz_results, name=name)
 
@@ -468,7 +475,9 @@ def _array_strptime_with_fallback(
     Call array_strptime, with fallback behavior depending on 'errors'.
     """
     try:
-        result, timezones = array_strptime(arg, fmt, exact=exact, errors=errors)
+        result, timezones = array_strptime(
+            arg, fmt, exact=exact, errors=errors, utc=utc
+        )
     except OutOfBoundsDatetime:
         if errors == "raise":
             raise
@@ -495,7 +504,7 @@ def _array_strptime_with_fallback(
             # Indicates to the caller to fallback to objects_to_datetime64ns
             return None
     else:
-        if any(timezones):
+        if any([i is not None for i in timezones]):
             return _return_parsed_timezone_results(result, timezones, utc, name)
 
     return _box_as_indexlike(result, utc=utc, name=name)

pandas/tests/tools/test_to_datetime.py

@@ -469,6 +469,89 @@ def test_to_datetime_mixed_datetime_and_string(self):
         expected = to_datetime([d1, d2]).tz_convert(pytz.FixedOffset(-60))
         tm.assert_index_equal(res, expected)
 
+    @pytest.mark.parametrize(
+        "fmt",
+        ["%Y-%d-%m %H:%M:%S%z", "%Y-%m-%d %H:%M:%S%z"],
+        ids=["non-ISO8601 format", "ISO8601 format"],
+    )
+    @pytest.mark.parametrize(
+        "utc, input, expected",
+        [
+            pytest.param(
+                True,
+                ["2000-01-01 01:00:00-08:00", "2000-01-01 02:00:00-08:00"],
+                DatetimeIndex(
+                    ["2000-01-01 09:00:00+00:00", "2000-01-01 10:00:00+00:00"],
+                    dtype="datetime64[ns, UTC]",
+                ),
+                id="all tz-aware, with utc",
+            ),
+            pytest.param(
+                False,
+                ["2000-01-01 01:00:00-08:00", "2000-01-01 02:00:00-08:00"],
+                DatetimeIndex(
+                    ["2000-01-01 01:00:00-08:00", "2000-01-01 02:00:00-08:00"],
+                    tz=pytz.FixedOffset(-480),
+                ),
+                id="all tz-aware, without utc",
+            ),
+            pytest.param(
+                True,
+                ["2000-01-01 01:00:00-08:00", "2000-01-01 02:00:00+00:00"],
+                DatetimeIndex(
+                    ["2000-01-01 09:00:00+00:00", "2000-01-01 02:00:00+00:00"],
+                    dtype="datetime64[ns, UTC]",
+                ),
+                id="all tz-aware, mixed offsets, with utc",
+            ),
+        ],
+    )
+    @pytest.mark.parametrize(
+        "constructor",
+        [Timestamp, lambda x: Timestamp(x).to_pydatetime()],
+    )
+    def test_to_datetime_mixed_datetime_and_string_with_format(
+        self, fmt, utc, input, expected, constructor
+    ):
+        # https://github.com/pandas-dev/pandas/issues/49298
+        # note: ISO8601 formats go down a fastpath, so we need to check both
+        # a ISO8601 format and a non-ISO8601 one
+        ts1 = constructor(input[0])
+        ts2 = input[1]
+        result = to_datetime([ts1, ts2], format=fmt, utc=utc)
+        tm.assert_index_equal(result, expected)
+
+    @pytest.mark.parametrize(
+        "fmt",
+        ["%Y-%d-%m %H:%M:%S%z", "%Y-%m-%d %H:%M:%S%z"],
+        ids=["non-ISO8601 format", "ISO8601 format"],
+    )
+    @pytest.mark.parametrize(
+        "input",
+        [
+            pytest.param(
+                ["2000-01-01 01:00:00-08:00", "2000-01-01 02:00:00-07:00"],
+                id="all tz-aware, mixed timezones, without utc",
+            ),
+        ],
+    )
+    @pytest.mark.parametrize(
+        "constructor",
+        [Timestamp, lambda x: Timestamp(x).to_pydatetime()],
+    )
+    def test_to_datetime_mixed_datetime_and_string_with_format_raises(
+        self, fmt, input, constructor
+    ):
+        # https://github.com/pandas-dev/pandas/issues/49298
+        # note: ISO8601 formats go down a fastpath, so we need to check both
+        # a ISO8601 format and a non-ISO8601 one
+        ts1 = constructor(input[0])
+        ts2 = constructor(input[1])
+        with pytest.raises(
+            ValueError, match="cannot be converted to datetime64 unless utc=True"
+        ):
+            to_datetime([ts1, ts2], format=fmt, utc=False)
+
     @pytest.mark.parametrize("infer_datetime_format", [True, False])
     def test_to_datetime_np_str(self, infer_datetime_format):
         # GH#32264
@@ -609,98 +692,6 @@ def test_to_datetime_dtarr(self, tz):
         result = to_datetime(arr)
         assert result is arr
 
-    def test_to_datetime_arraylike_contains_pydatetime_and_timestamp(self):
-        # GH 49298
-        # Test explicit custom format
-        case1 = [
-            Timestamp("2001-10-01 12:00:01.123456789"),
-            datetime(2001, 10, 2, 12, 30, 1, 123456),
-            "10/03/01",
-        ]
-        result = to_datetime(case1, format="%m/%d/%y")
-        expected_data = [
-            Timestamp("2001-10-01 12:00:01.123456789"),
-            Timestamp("2001-10-02 12:30:01.123456"),
-            Timestamp("2001-10-03 00:00:00"),
-        ]
-        tm.assert_equal(result, DatetimeIndex(expected_data))
-
-        # Test ISO8601 format
-        case2 = [
-            Timestamp("2001-10-01 13:18:05"),
-            datetime(2001, 10, 2, 13, 18, 5),
-            "2001-10-03T13:18:05",
-            "20011004",
-        ]
-        result = to_datetime(case2)
-        expected_data = [
-            Timestamp("2001-10-01 13:18:05"),
-            Timestamp("2001-10-02 13:18:05"),
-            Timestamp("2001-10-03 13:18:05"),
-            Timestamp("2001-10-04 00:00:00"),
-        ]
-        tm.assert_equal(result, DatetimeIndex(expected_data))
-
-    def test_to_datetime_arraylike_contains_pydatetime_and_timestamp_with_tz(self):
-        # GH 49298
-        # Different offsets when utc=True
-        data = [
-            "20100102 121314 +01:00",
-            "20100102 121315 -05:00",
-            pytz.timezone("Europe/Berlin").localize(datetime(2010, 1, 2, 12, 13, 16)),
-            pytz.timezone("US/Eastern").localize(Timestamp("2010-01-02 12:13:17")),
-        ]
-        expected_data = [
-            Timestamp("2010-01-02 11:13:14", tz="utc"),
-            Timestamp("2010-01-02 17:13:15", tz="utc"),
-            Timestamp("2010-01-02 11:13:16", tz="utc"),
-            Timestamp("2010-01-02 17:13:17", tz="utc"),
-        ]
-        result = to_datetime(data, format="%Y%m%d %H%M%S %z", utc=True)
-        tm.assert_equal(result, DatetimeIndex(expected_data))
-
-        # Different offsets when utc=False
-        expected_data = [
-            Timestamp("2010-01-02 12:13:14 +01:00"),
-            Timestamp("2010-01-02 12:13:15 -05:00"),
-            Timestamp("2010-01-02 12:13:16 +01:00"),
-            Timestamp("2010-01-02 12:13:17 -05:00"),
-        ]
-        result = to_datetime(data, format="%Y%m%d %H%M%S %z", utc=False)
-        tm.assert_equal(result, Index(expected_data))
-
-    @pytest.mark.parametrize("value", [datetime(2010, 1, 2, 12, 13, 16), Timestamp("2010-01-02 12:13:17")])
-    def test_to_datetime_includes_tz_dtype_on_pydatetime_and_timestamp(self, value):
-        # GH 49298
-        # No timezone
-        result_no_format = to_datetime([value])
-        result_with_format = to_datetime([value], format="%m-%d-%Y")
-        tm.assert_equal(result_no_format, result_with_format)
-
-        # 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 %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)