BUG: DatetimeIndex.resolution with nanosecond reso (#46903)

doc/source/whatsnew/v1.5.0.rst

@@ -600,6 +600,8 @@ Other Deprecations
 - Deprecated the ``closed`` argument in :meth:`interval_range` in favor of ``inclusive`` argument; In a future version passing ``closed`` will raise (:issue:`40245`)
 - Deprecated the methods :meth:`DataFrame.mad`, :meth:`Series.mad`, and the corresponding groupby methods (:issue:`11787`)
 - Deprecated positional arguments to :meth:`Index.join` except for ``other``, use keyword-only arguments instead of positional arguments (:issue:`46518`)
+- Deprecated indexing on a timezone-naive :class:`DatetimeIndex` using a string representing a timezone-aware datetime (:issue:`46903`, :issue:`36148`)
+-
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_150.performance:
@@ -641,6 +643,7 @@ Datetimelike
 - Bug in :meth:`Index.astype` when casting from object dtype to ``timedelta64[ns]`` dtype incorrectly casting ``np.datetime64("NaT")`` values to ``np.timedelta64("NaT")`` instead of raising (:issue:`45722`)
 - Bug in :meth:`SeriesGroupBy.value_counts` index when passing categorical column (:issue:`44324`)
 - Bug in :meth:`DatetimeIndex.tz_localize` localizing to UTC failing to make a copy of the underlying data (:issue:`46460`)
+- Bug in :meth:`DatetimeIndex.resolution` incorrectly returning "day" instead of "nanosecond" for nanosecond-resolution indexes (:issue:`46903`)
 -
 
 Timedelta

pandas/_libs/tslib.pyx

@@ -60,6 +60,10 @@ from pandas._libs.tslibs.nattype cimport (
 )
 from pandas._libs.tslibs.timestamps cimport _Timestamp
 
+from pandas._libs.tslibs import (
+    Resolution,
+    get_resolution,
+)
 from pandas._libs.tslibs.timestamps import Timestamp
 
 # Note: this is the only non-tslibs intra-pandas dependency here
@@ -122,11 +126,11 @@ def format_array_from_datetime(
     """
     cdef:
         int64_t val, ns, N = len(values)
-        ndarray[int64_t] consider_values
         bint show_ms = False, show_us = False, show_ns = False
         bint basic_format = False
         ndarray[object] result = cnp.PyArray_EMPTY(values.ndim, values.shape, cnp.NPY_OBJECT, 0)
-        object ts, res
+        _Timestamp ts
+        str res
         npy_datetimestruct dts
 
     if na_rep is None:
@@ -136,16 +140,10 @@ def format_array_from_datetime(
     # a format based on precision
     basic_format = format is None and tz is None
     if basic_format:
-        consider_values = values[values != NPY_NAT]
-        show_ns = (consider_values % 1000).any()
-
-        if not show_ns:
-            consider_values //= 1000
-            show_us = (consider_values % 1000).any()
-
-            if not show_ms:
-                consider_values //= 1000
-                show_ms = (consider_values % 1000).any()
+        reso_obj = get_resolution(values)
+        show_ns = reso_obj == Resolution.RESO_NS
+        show_us = reso_obj == Resolution.RESO_US
+        show_ms = reso_obj == Resolution.RESO_MS
 
     for i in range(N):
         val = values[i]
@@ -178,6 +176,7 @@ def format_array_from_datetime(
                 # invalid format string
                 # requires dates > 1900
                 try:
+                    # Note: dispatches to pydatetime
                     result[i] = ts.strftime(format)
                 except ValueError:
                     result[i] = str(ts)

pandas/_libs/tslibs/vectorized.pyx

@@ -204,7 +204,9 @@ def ints_to_pydatetime(
 
 
 cdef inline c_Resolution _reso_stamp(npy_datetimestruct *dts):
-    if dts.us != 0:
+    if dts.ps != 0:
+        return c_Resolution.RESO_NS
+    elif dts.us != 0:
         if dts.us % 1000 == 0:
             return c_Resolution.RESO_MS
         return c_Resolution.RESO_US

pandas/core/indexes/datetimes.py

@@ -593,7 +593,7 @@ def _parsed_string_to_bounds(self, reso: Resolution, parsed: datetime):
         end = self._maybe_cast_for_get_loc(end)
         return start, end
 
-    def _deprecate_mismatched_indexing(self, key) -> None:
+    def _deprecate_mismatched_indexing(self, key, one_way: bool = False) -> None:
         # GH#36148
         # we get here with isinstance(key, self._data._recognized_scalars)
         try:
@@ -606,6 +606,10 @@ def _deprecate_mismatched_indexing(self, key) -> None:
                     "raise KeyError in a future version.  "
                     "Use a timezone-naive object instead."
                 )
+            elif one_way:
+                # we special-case timezone-naive strings and timezone-aware
+                #  DatetimeIndex
+                return
             else:
                 msg = (
                     "Indexing a timezone-aware DatetimeIndex with a "
@@ -640,6 +644,7 @@ def get_loc(self, key, method=None, tolerance=None):
                 parsed, reso = self._parse_with_reso(key)
             except ValueError as err:
                 raise KeyError(key) from err
+            self._deprecate_mismatched_indexing(parsed, one_way=True)
 
             if self._can_partial_date_slice(reso):
                 try:

pandas/tests/indexing/test_datetime.py

@@ -10,6 +10,22 @@
 
 
 class TestDatetimeIndex:
+    def test_get_loc_naive_dti_aware_str_deprecated(self):
+        # GH#46903
+        ts = Timestamp("20130101").value
+        dti = pd.DatetimeIndex([ts + 50 + i for i in range(100)])
+        ser = Series(range(100), index=dti)
+
+        key = "2013-01-01 00:00:00.000000050+0000"
+        msg = "Indexing a timezone-naive DatetimeIndex with a timezone-aware datetime"
+        with tm.assert_produces_warning(FutureWarning, match=msg):
+            res = ser[key]
+        assert res == 0
+
+        with tm.assert_produces_warning(FutureWarning, match=msg):
+            loc = dti.get_loc(key)
+        assert loc == 0
+
     def test_indexing_with_datetime_tz(self):
 
         # GH#8260

pandas/tests/series/methods/test_asof.py

@@ -23,9 +23,12 @@ def test_asof_nanosecond_index_access(self):
 
         first_value = ser.asof(ser.index[0])
 
+        # GH#46903 previously incorrectly was "day"
+        assert dti.resolution == "nanosecond"
+
         # this used to not work bc parsing was done by dateutil that didn't
         #  handle nanoseconds
-        assert first_value == ser["2013-01-01 00:00:00.000000050+0000"]
+        assert first_value == ser["2013-01-01 00:00:00.000000050"]
 
         expected_ts = np.datetime64("2013-01-01 00:00:00.000000050", "ns")
         assert first_value == ser[Timestamp(expected_ts)]

pandas/tests/tslibs/test_resolution.py

@@ -0,0 +1,13 @@
+import numpy as np
+
+from pandas._libs.tslibs import (
+    Resolution,
+    get_resolution,
+)
+
+
+def test_get_resolution_nano():
+    # don't return the fallback RESO_DAY
+    arr = np.array([1], dtype=np.int64)
+    res = get_resolution(arr)
+    assert res == Resolution.RESO_NS