DEPR: Deprecate non-ISO date string formats in DatetimeIndex.loc

doc/source/user_guide/10min.rst

@@ -203,7 +203,7 @@ For a :class:`DataFrame`, passing a slice ``:`` selects matching rows:
 .. ipython:: python
 
    df[0:3]
-   df["20130102":"20130104"]
+   df["2013-01-02":"2013-01-04"]
 
 Selection by label
 ~~~~~~~~~~~~~~~~~~
@@ -226,7 +226,7 @@ For label slicing, both endpoints are *included*:
 
 .. ipython:: python
 
-   df.loc["20130102":"20130104", ["A", "B"]]
+   df.loc["2013-01-02":"2013-01-04", ["A", "B"]]
 
 Selecting a single row and column label returns a scalar:
 

doc/source/user_guide/indexing.rst

@@ -314,7 +314,7 @@ Selection by label
 
   .. ipython:: python
 
-     dfl.loc['20130102':'20130104']
+     dfl.loc['2013-01-02':'2013-01-04']
 
 pandas provides a suite of methods in order to have **purely label based indexing**. This is a strict inclusion based protocol.
 Every label asked for must be in the index, or a ``KeyError`` will be raised.

doc/source/user_guide/timeseries.rst

@@ -566,11 +566,11 @@ Dates and strings that parse to timestamps can be passed as indexing parameters:
 
 .. ipython:: python
 
-   ts["1/31/2011"]
+   ts["2011-01-31"]
 
    ts[datetime.datetime(2011, 12, 25):]
 
-   ts["10/31/2011":"12/31/2011"]
+   ts["2011-10-31":"2011-12-31"]
 
 To provide convenience for accessing longer time series, you can also pass in
 the year or year and month as strings:

doc/source/whatsnew/v3.0.0.rst

@@ -805,6 +805,7 @@ Other Deprecations
 - Deprecated allowing ``fill_value`` that cannot be held in the original dtype (excepting NA values for integer and bool dtypes) in :meth:`Series.shift` and :meth:`DataFrame.shift` (:issue:`53802`)
 - Deprecated allowing strings representing full dates in :meth:`DataFrame.at_time` and :meth:`Series.at_time` (:issue:`50839`)
 - Deprecated backward-compatibility behavior for :meth:`DataFrame.select_dtypes` matching "str" dtype when ``np.object_`` is specified (:issue:`61916`)
+- Deprecated non-ISO date string formats in :meth:`DatetimeIndex.__getitem__` with string labels. Use ISO format (YYYY-MM-DD) instead. (:issue:`58302`)
 - Deprecated option "future.no_silent_downcasting", as it is no longer used. In a future version accessing this option will raise (:issue:`59502`)
 - Deprecated passing non-Index types to :meth:`Index.join`; explicitly convert to Index first (:issue:`62897`)
 - Deprecated silent casting of non-datetime 'other' to datetime in :meth:`Series.combine_first` (:issue:`62931`)

pandas/core/indexes/datetimes.py

@@ -2,6 +2,7 @@
 
 import datetime as dt
 import operator
+import re
 from typing import (
     TYPE_CHECKING,
     Self,
@@ -110,6 +111,42 @@ def _new_DatetimeIndex(cls, d):
     return result
 
 
+def _is_iso_format_string(date_str: str) -> bool:
+    """
+    Check if a date string follows ISO8601 format.
+
+    Uses date.fromisoformat() to validate full ISO dates, with fallback to regex
+    for reduced precision dates (YYYY or YYYY-MM) which are not supported by
+    fromisoformat().
+
+    Examples of ISO format (True):
+    - 2024 (reduced precision)
+    - 2024-01 (reduced precision)
+    - 2024-01-10
+    - 2024-01-10T00:00:00
+
+    Examples of non-ISO format (False):
+    - 2024/01/10 (/ separator)
+    - 2024 01 10 (space separator)
+    - 01/10/2024 (MM/DD/YYYY)
+    - 10/01/2024 (DD/MM/YYYY)
+    - 01-10-2024 (MM-DD-YYYY)
+    """
+    try:
+        # Standard library validates full ISO dates (YYYY-MM-DD format)
+        dt.date.fromisoformat(date_str)
+        return True
+    except (ValueError, TypeError):
+        # Fallback regex for reduced precision dates not supported by fromisoformat()
+        # Checks if string starts with ISO pattern (YYYY, YYYY-MM, YYYY-MM-DD, etc.)
+        # Pattern: ^\d{4}(?:-|T|$)
+        # - Requires exactly 4 digits at start (year)
+        # - Followed by: hyphen (YYYY-), T (YYYY-T...), or end (YYYY)
+        # Examples that match: "2024", "2024-01", "2024-01-10", "2024-01-10T00:00:00"
+        # Examples that don't: "01/10/2024", "2024 01 10", "1/1/2024"
+        return re.match(r"^\d{4}(?:-|T|$)", date_str) is not None
+
+
 @inherit_names(
     DatetimeArray._field_ops
     + [
@@ -613,6 +650,14 @@ def get_loc(self, key):
                 parsed, reso = self._parse_with_reso(key)
             except ValueError as err:
                 raise KeyError(key) from err
+            # GH#58302 - Deprecate non-ISO string formats in .loc indexing
+            if not _is_iso_format_string(key):
+                msg = (
+                    "Parsing non-ISO datetime strings in .loc is deprecated "
+                    "and will be removed in a future version. Use ISO format "
+                    f"(YYYY-MM-DD) instead. Got '{key}'."
+                )
+                warnings.warn(msg, Pandas4Warning, stacklevel=find_stack_level())
             self._disallow_mismatched_indexing(parsed)
 
             if self._can_partial_date_slice(reso):
@@ -688,6 +733,23 @@ def slice_indexer(self, start=None, end=None, step=None):
         def check_str_or_none(point) -> bool:
             return point is not None and not isinstance(point, str)
 
+        # GH#58302 - Deprecate non-ISO string formats in .loc indexing
+        if isinstance(start, str) and not _is_iso_format_string(start):
+            msg = (
+                "Parsing non-ISO datetime strings in .loc is deprecated "
+                "and will be removed in a future version. Use ISO format "
+                f"(YYYY-MM-DD) instead. Got '{start}'."
+            )
+            warnings.warn(msg, Pandas4Warning, stacklevel=find_stack_level())
+
+        if isinstance(end, str) and not _is_iso_format_string(end):
+            msg = (
+                "Parsing non-ISO datetime strings in .loc is deprecated "
+                "and will be removed in a future version. Use ISO format "
+                f"(YYYY-MM-DD) instead. Got '{end}'."
+            )
+            warnings.warn(msg, Pandas4Warning, stacklevel=find_stack_level())
+
         # GH#33146 if start and end are combinations of str and None and Index is not
         # monotonic, we can not use Index.slice_indexer because it does not honor the
         # actual elements, is only searching for start and end

pandas/tests/groupby/methods/test_value_counts.py

@@ -72,6 +72,9 @@ def seed_df(seed_nans, n, m):
 @pytest.mark.parametrize("bins", [None, [0, 5]], ids=repr)
 @pytest.mark.parametrize("isort", [True, False])
 @pytest.mark.parametrize("normalize, name", [(True, "proportion"), (False, "count")])
+@pytest.mark.filterwarnings(
+    "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+)
 def test_series_groupby_value_counts(
     seed_nans,
     num_rows,

pandas/tests/groupby/test_groupby.py

@@ -2851,6 +2851,9 @@ def test_groupby_with_Time_Grouper(unit):
     tm.assert_frame_equal(result, expected_output)
 
 
+@pytest.mark.filterwarnings(
+    "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+)
 def test_groupby_series_with_datetimeindex_month_name():
     # GH 48509
     s = Series([0, 1, 0], index=date_range("2022-01-01", periods=3), name="jan")

pandas/tests/indexes/datetimes/test_indexing.py

@@ -491,9 +491,9 @@ def test_get_loc_timedelta_invalid_key(self, key):
 
     def test_get_loc_reasonable_key_error(self):
         # GH#1062
-        index = DatetimeIndex(["1/3/2000"])
+        index = DatetimeIndex(["2000-01-03"])
         with pytest.raises(KeyError, match="2000"):
-            index.get_loc("1/1/2000")
+            index.get_loc("2000-01-01")
 
     def test_get_loc_year_str(self):
         rng = date_range("1/1/2000", "1/1/2010")

pandas/tests/indexes/datetimes/test_partial_slicing.py

@@ -5,6 +5,8 @@
 import numpy as np
 import pytest
 
+from pandas.errors import Pandas4Warning
+
 from pandas import (
     DataFrame,
     DatetimeIndex,
@@ -19,6 +21,10 @@
 
 
 class TestSlicing:
+    pytestmark = pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
+
     def test_string_index_series_name_converted(self):
         # GH#1644
         df = DataFrame(
@@ -464,3 +470,94 @@ def test_slice_reduce_to_series(self):
         )
         result = df.loc["2000", "A"]
         tm.assert_series_equal(result, expected)
+
+
+class TestDatetimeIndexNonISODeprecation:
+    """Tests for deprecation of non-ISO string formats in .loc indexing. GH#58302"""
+
+    @pytest.fixture
+    def ser_daily(self):
+        """Create a Series with daily DatetimeIndex for testing."""
+        return Series(
+            range(15),
+            index=DatetimeIndex(date_range(start="2024-01-01", freq="D", periods=15)),
+        )
+
+    @pytest.mark.parametrize(
+        "date_string",
+        [
+            "1/10/2024",  # MM/DD/YYYY format
+            "01/10/2024",  # MM/DD/YYYY format with leading zero
+        ],
+    )
+    def test_loc_indexing_non_iso_single_key_deprecation(self, ser_daily, date_string):
+        # GH#58302
+        msg = "Parsing non-ISO datetime strings in .loc is deprecated"
+
+        with tm.assert_produces_warning(Pandas4Warning, match=msg):
+            result = ser_daily.loc[date_string]
+            assert result == 9
+
+    @pytest.mark.parametrize(
+        "date_string,expected",
+        [
+            ("2024-01-10", 9),  # YYYY-MM-DD (ISO format)
+        ],
+    )
+    def test_loc_indexing_iso_format_no_warning(self, ser_daily, date_string, expected):
+        # GH#58302 - ISO format (YYYY-MM-DD) should NOT warn
+        with tm.assert_produces_warning(None):
+            result = ser_daily.loc[date_string]
+            assert result == expected
+
+    @pytest.mark.parametrize(
+        "start_string",
+        [
+            "1/10/2024",  # MM/DD/YYYY format
+            "01/10/2024",  # MM/DD/YYYY format with leading zero
+        ],
+    )
+    def test_loc_slicing_non_iso_start_deprecation(self, ser_daily, start_string):
+        # GH#58302 - Non-ISO start in slice should warn
+        msg = "Parsing non-ISO datetime strings in .loc is deprecated"
+
+        with tm.assert_produces_warning(Pandas4Warning, match=msg):
+            result = ser_daily.loc[start_string:"2024-01-15"]
+            assert len(result) > 0
+
+    @pytest.mark.parametrize(
+        "end_string",
+        [
+            "5-01-2024",  # DD-MM-YYYY format
+            "05-01-2024",  # DD-MM-YYYY format with leading zero
+        ],
+    )
+    def test_loc_slicing_non_iso_end_deprecation(self, ser_daily, end_string):
+        # GH#58302 - Non-ISO end in slice should warn
+        msg = "Parsing non-ISO datetime strings in .loc is deprecated"
+
+        with tm.assert_produces_warning(Pandas4Warning, match=msg):
+            result = ser_daily.loc["2024-01-01":end_string]
+            assert len(result) > 0
+
+    def test_loc_slicing_both_non_iso_deprecation(self, ser_daily):
+        # GH#58302 - Both non-ISO should warn (twice)
+        msg = "Parsing non-ISO datetime strings in .loc is deprecated"
+
+        with tm.assert_produces_warning(
+            Pandas4Warning, match=msg, check_stacklevel=False
+        ):
+            result = ser_daily.loc["1/10/2024":"5-01-2024"]
+            assert len(result) > 0
+
+    def test_loc_slicing_iso_formats_no_warning(self, ser_daily):
+        # GH#58302 - ISO slice formats should NOT warn
+        with tm.assert_produces_warning(None):
+            result = ser_daily.loc["2024-01-05":"2024-01-10"]
+            assert len(result) == 6
+
+    def test_loc_non_string_keys_no_warning(self, ser_daily):
+        # GH#58302 - Non-string keys should not warn
+        with tm.assert_produces_warning(None):
+            result = ser_daily.loc[Timestamp("2024-01-10")]
+            assert result == 9

pandas/tests/indexes/period/test_indexing.py

@@ -174,17 +174,17 @@ def test_getitem_list_periods(self):
     @pytest.mark.arm_slow
     def test_getitem_seconds(self):
         # GH#6716
-        didx = date_range(start="2013/01/01 09:00:00", freq="s", periods=4000)
-        pidx = period_range(start="2013/01/01 09:00:00", freq="s", periods=4000)
+        didx = date_range(start="2013-01-01 09:00:00", freq="s", periods=4000)
+        pidx = period_range(start="2013-01-01 09:00:00", freq="s", periods=4000)
 
         for idx in [didx, pidx]:
             # getitem against index should raise ValueError
             values = [
                 "2014",
-                "2013/02",
-                "2013/01/02",
-                "2013/02/01 9h",
-                "2013/02/01 09:00",
+                "2013-02",
+                "2013-01-02",
+                "2013-02-01 9h",
+                "2013-02-01 09:00",
             ]
             for val in values:
                 # GH7116
@@ -194,9 +194,9 @@ def test_getitem_seconds(self):
                     idx[val]
 
             ser = Series(np.random.default_rng(2).random(len(idx)), index=idx)
-            tm.assert_series_equal(ser["2013/01/01 10:00"], ser[3600:3660])
-            tm.assert_series_equal(ser["2013/01/01 9h"], ser[:3600])
-            for d in ["2013/01/01", "2013/01", "2013"]:
+            tm.assert_series_equal(ser["2013-01-01 10:00"], ser[3600:3660])
+            tm.assert_series_equal(ser["2013-01-01 9h"], ser[:3600])
+            for d in ["2013-01-01", "2013-01", "2013"]:
                 tm.assert_series_equal(ser[d], ser)
 
     @pytest.mark.parametrize(
@@ -206,6 +206,9 @@ def test_getitem_seconds(self):
             period_range,
         ],
     )
+    @pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
     def test_getitem_day(self, idx_range):
         # GH#6716
         # Confirm DatetimeIndex and PeriodIndex works identically

pandas/tests/indexes/period/test_partial_slicing.py

@@ -12,6 +12,10 @@
 
 
 class TestPeriodIndex:
+    pytestmark = pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
+
     def test_getitem_periodindex_duplicates_string_slice(self):
         # monotonic
         idx = PeriodIndex([2000, 2007, 2007, 2009, 2009], freq="Y-JUN")

pandas/tests/indexing/multiindex/test_slice.py

@@ -308,6 +308,9 @@ def test_multiindex_slicers_datetimelike(self):
         ]
         tm.assert_frame_equal(result, expected)
 
+    @pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
     def test_multiindex_slicers_edges(self):
         # GH 8132
         # various edge cases

pandas/tests/indexing/test_loc.py

@@ -280,6 +280,9 @@ def test_loc_empty_slice_assignment_with_datetime(self, data):
 
 class TestLocBaseIndependent:
     # Tests for loc that do not depend on subclassing Base
+    @pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
     def test_loc_npstr(self):
         # GH#45580
         df = DataFrame(index=date_range("2021", "2022"))
@@ -1281,6 +1284,9 @@ def test_loc_setitem_str_to_small_float_conversion_type(self, using_infer_string
         expected = DataFrame(col_data, columns=["A"], dtype=float)
         tm.assert_frame_equal(result, expected)
 
+    @pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
     def test_loc_getitem_time_object(self, frame_or_series):
         rng = date_range("1/1/2000", "1/5/2000", freq="5min")
         mask = (rng.hour == 9) & (rng.minute == 30)
@@ -2434,6 +2440,9 @@ def test_loc_getitem_partial_slice_non_monotonicity(
 
 
 class TestLabelSlicing:
+    @pytest.mark.filterwarnings(
+        "ignore:Parsing non-ISO datetime strings:pandas.errors.Pandas4Warning"
+    )
     def test_loc_getitem_slicing_datetimes_frame(self):
         # GH#7523