API: date_range, timedelta_range infer unit from start/end/freq

doc/source/whatsnew/v3.0.0.rst

@@ -745,6 +745,7 @@ Other API changes
   the dtype of the resulting Index (:issue:`60797`)
 - :class:`IncompatibleFrequency` now subclasses ``TypeError`` instead of ``ValueError``. As a result, joins with mismatched frequencies now cast to object like other non-comparable joins, and arithmetic with indexes with mismatched frequencies align (:issue:`55782`)
 - :class:`Series` "flex" methods like :meth:`Series.add` no longer allow passing a :class:`DataFrame` for ``other``; use the DataFrame reversed method instead (:issue:`46179`)
+- :func:`date_range` and :func:`timedelta_range` no longer default to ``unit="ns"``, instead will infer a unit from the ``start``, ``end``, and ``freq`` parameters. Explicitly specify a desired ``unit`` to override these (:issue:`59031`)
 - :meth:`CategoricalIndex.append` no longer attempts to cast different-dtype indexes to the caller's dtype (:issue:`41626`)
 - :meth:`ExtensionDtype.construct_array_type` is now a regular method instead of a ``classmethod`` (:issue:`58663`)
 - Comparison operations between :class:`Index` and :class:`Series` now consistently return :class:`Series` regardless of which object is on the left or right (:issue:`36759`)

pandas/core/indexes/datetimes.py

@@ -25,7 +25,11 @@
     timezones,
     to_offset,
 )
-from pandas._libs.tslibs.offsets import prefix_mapping
+from pandas._libs.tslibs.dtypes import abbrev_to_npy_unit
+from pandas._libs.tslibs.offsets import (
+    DateOffset,
+    prefix_mapping,
+)
 from pandas.errors import Pandas4Warning
 from pandas.util._decorators import (
     cache_readonly,
@@ -883,7 +887,7 @@ def date_range(
     name: Hashable | None = None,
     inclusive: IntervalClosedType = "both",
     *,
-    unit: TimeUnit = "ns",
+    unit: TimeUnit | None = None,
     **kwargs,
 ) -> DatetimeIndex:
     """
@@ -922,8 +926,9 @@ def date_range(
         Name of the resulting DatetimeIndex.
     inclusive : {"both", "neither", "left", "right"}, default "both"
         Include boundaries; Whether to set each bound as closed or open.
-    unit : {'s', 'ms', 'us', 'ns'}, default 'ns'
+    unit : {'s', 'ms', 'us', 'ns', None}, default None
         Specify the desired resolution of the result.
+        If not specified, this is inferred from the 'start', 'end', and 'freq'
 
         .. versionadded:: 2.0.0
     **kwargs
@@ -1062,6 +1067,50 @@ def date_range(
     """
     if freq is None and com.any_none(periods, start, end):
         freq = "D"
+    if freq is not None:
+        freq = to_offset(freq)
+
+    if start is NaT or end is NaT:
+        # This check needs to come before the `unit = start.unit` line below
+        raise ValueError("Neither `start` nor `end` can be NaT")
+
+    if unit is None:
+        # Infer the unit based on the inputs
+
+        if start is not None and end is not None:
+            start = Timestamp(start)
+            end = Timestamp(end)
+            if abbrev_to_npy_unit(start.unit) > abbrev_to_npy_unit(end.unit):
+                unit = start.unit
+            else:
+                unit = end.unit
+        elif start is not None:
+            start = Timestamp(start)
+            unit = start.unit
+        elif end is not None:
+            end = Timestamp(end)
+            unit = end.unit
+        else:
+            raise ValueError(
+                "Of the four parameters: start, end, periods, "
+                "and freq, exactly three must be specified"
+            )
+
+        # Last we need to watch out for cases where the 'freq' implies a higher
+        #  unit than either start or end
+        if freq is not None:
+            creso = abbrev_to_npy_unit(unit)
+            if isinstance(freq, Tick):
+                if freq._creso > creso:
+                    unit = freq.base.freqstr  # type: ignore[assignment]
+            elif hasattr(freq, "offset") and freq.offset is not None:
+                # e.g. BDay with an offset
+                td = Timedelta(freq.offset)
+                if abbrev_to_npy_unit(td.unit) > creso:
+                    unit = td.unit  # type: ignore[assignment]
+            elif type(freq) is DateOffset and getattr(freq, "nanoseconds", 0) != 0:
+                # e.g. test_freq_dateoffset_with_relateivedelta_nanos
+                unit = "ns"
 
     dtarr = DatetimeArray._generate_range(
         start=start,

pandas/core/indexes/timedeltas.py

@@ -2,7 +2,10 @@
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+from typing import (
+    TYPE_CHECKING,
+    cast,
+)
 
 from pandas._libs import (
     index as libindex,
@@ -13,6 +16,7 @@
     Timedelta,
     to_offset,
 )
+from pandas._libs.tslibs.dtypes import abbrev_to_npy_unit
 from pandas.util._decorators import set_module
 
 from pandas.core.dtypes.common import (
@@ -33,6 +37,10 @@
 
 if TYPE_CHECKING:
     from pandas._libs import NaTType
+    from pandas._libs.tslibs import (
+        Day,
+        Tick,
+    )
     from pandas._typing import (
         DtypeObj,
         TimeUnit,
@@ -252,7 +260,7 @@ def timedelta_range(
     name=None,
     closed=None,
     *,
-    unit: TimeUnit = "ns",
+    unit: TimeUnit | None = None,
 ) -> TimedeltaIndex:
     """
     Return a fixed frequency TimedeltaIndex with day as the default.
@@ -272,8 +280,9 @@ def timedelta_range(
     closed : str, default None
         Make the interval closed with respect to the given frequency to
         the 'left', 'right', or both sides (None).
-    unit : {'s', 'ms', 'us', 'ns'}, default 'ns'
+    unit : {'s', 'ms', 'us', 'ns', None}, default None
         Specify the desired resolution of the result.
+        If not specified, this is inferred from the 'start', 'end', and 'freq'
 
         .. versionadded:: 2.0.0
 
@@ -337,8 +346,40 @@ def timedelta_range(
     """
     if freq is None and com.any_none(periods, start, end):
         freq = "D"
-
     freq = to_offset(freq)
+
+    if com.count_not_none(start, end, periods, freq) != 3:
+        # This check needs to come before the `unit = start.unit` line below
+        raise ValueError(
+            "Of the four parameters: start, end, periods, "
+            "and freq, exactly three must be specified"
+        )
+
+    if unit is None:
+        # Infer the unit based on the inputs
+
+        if start is not None and end is not None:
+            start = Timedelta(start)
+            end = Timedelta(end)
+            if abbrev_to_npy_unit(start.unit) > abbrev_to_npy_unit(end.unit):
+                unit = cast("TimeUnit", start.unit)
+            else:
+                unit = cast("TimeUnit", end.unit)
+        elif start is not None:
+            start = Timedelta(start)
+            unit = cast("TimeUnit", start.unit)
+        else:
+            end = Timedelta(end)
+            unit = cast("TimeUnit", end.unit)
+
+        # Last we need to watch out for cases where the 'freq' implies a higher
+        #  unit than either start or end
+        if freq is not None:
+            freq = cast("Tick | Day", freq)
+            creso = abbrev_to_npy_unit(unit)
+            if freq._creso > creso:
+                unit = cast("TimeUnit", freq.base.freqstr)
+
     tdarr = TimedeltaArray._generate_range(
         start, end, periods, freq, closed=closed, unit=unit
     )

pandas/tests/apply/test_frame_apply.py

@@ -653,13 +653,15 @@ def test_apply_dict(df, dicts):
 
 def test_apply_non_numpy_dtype():
     # GH 12244
-    df = DataFrame({"dt": date_range("2015-01-01", periods=3, tz="Europe/Brussels")})
+    df = DataFrame(
+        {"dt": date_range("2015-01-01", periods=3, tz="Europe/Brussels", unit="ns")}
+    )
     result = df.apply(lambda x: x)
     tm.assert_frame_equal(result, df)
 
     result = df.apply(lambda x: x + pd.Timedelta("1day"))
     expected = DataFrame(
-        {"dt": date_range("2015-01-02", periods=3, tz="Europe/Brussels")}
+        {"dt": date_range("2015-01-02", periods=3, tz="Europe/Brussels", unit="ns")}
     )
     tm.assert_frame_equal(result, expected)
 
@@ -1425,7 +1427,7 @@ def test_nuiscance_columns():
             "A": [1, 2, 3],
             "B": [1.0, 2.0, 3.0],
             "C": ["foo", "bar", "baz"],
-            "D": date_range("20130101", periods=3),
+            "D": date_range("20130101", periods=3, unit="ns"),
         }
     )