REF: get_supported_reso->get_supported_dtype (pandas-dev#56439)

Author: jbrockmendel

pandas/_libs/tslibs/__init__.py

@@ -30,20 +30,16 @@
     "get_unit_from_dtype",
     "periods_per_day",
     "periods_per_second",
-    "is_supported_unit",
-    "npy_unit_to_abbrev",
-    "get_supported_reso",
     "guess_datetime_format",
     "add_overflowsafe",
+    "get_supported_dtype",
+    "is_supported_dtype",
 ]
 
 from pandas._libs.tslibs import dtypes  # pylint: disable=import-self
 from pandas._libs.tslibs.conversion import localize_pydatetime
 from pandas._libs.tslibs.dtypes import (
     Resolution,
-    get_supported_reso,
-    is_supported_unit,
-    npy_unit_to_abbrev,
     periods_per_day,
     periods_per_second,
 )
@@ -58,6 +54,8 @@
     OutOfBoundsTimedelta,
     add_overflowsafe,
     astype_overflowsafe,
+    get_supported_dtype,
+    is_supported_dtype,
     is_unitless,
     py_get_unit_from_dtype as get_unit_from_dtype,
 )

pandas/_libs/tslibs/dtypes.pxd

@@ -3,13 +3,13 @@ from numpy cimport int64_t
 from pandas._libs.tslibs.np_datetime cimport NPY_DATETIMEUNIT
 
 
-cpdef str npy_unit_to_abbrev(NPY_DATETIMEUNIT unit)
+cdef str npy_unit_to_abbrev(NPY_DATETIMEUNIT unit)
 cpdef NPY_DATETIMEUNIT abbrev_to_npy_unit(str abbrev)
 cdef NPY_DATETIMEUNIT freq_group_code_to_npy_unit(int freq) noexcept nogil
 cpdef int64_t periods_per_day(NPY_DATETIMEUNIT reso=*) except? -1
 cpdef int64_t periods_per_second(NPY_DATETIMEUNIT reso) except? -1
-cpdef NPY_DATETIMEUNIT get_supported_reso(NPY_DATETIMEUNIT reso)
-cpdef bint is_supported_unit(NPY_DATETIMEUNIT reso)
+cdef NPY_DATETIMEUNIT get_supported_reso(NPY_DATETIMEUNIT reso)
+cdef bint is_supported_unit(NPY_DATETIMEUNIT reso)
 
 cpdef freq_to_period_freqstr(freq_n, freq_name)
 cdef dict c_OFFSET_TO_PERIOD_FREQSTR

pandas/_libs/tslibs/dtypes.pyi

@@ -4,9 +4,6 @@ OFFSET_TO_PERIOD_FREQSTR: dict[str, str]
 
 def periods_per_day(reso: int = ...) -> int: ...
 def periods_per_second(reso: int) -> int: ...
-def is_supported_unit(reso: int) -> bool: ...
-def npy_unit_to_abbrev(unit: int) -> str: ...
-def get_supported_reso(reso: int) -> int: ...
 def abbrev_to_npy_unit(abbrev: str) -> int: ...
 def freq_to_period_freqstr(freq_n: int, freq_name: str) -> str: ...
 

pandas/_libs/tslibs/dtypes.pyx

@@ -559,7 +559,7 @@ class NpyDatetimeUnit(Enum):
     NPY_FR_GENERIC = NPY_DATETIMEUNIT.NPY_FR_GENERIC
 
 
-cpdef NPY_DATETIMEUNIT get_supported_reso(NPY_DATETIMEUNIT reso):
+cdef NPY_DATETIMEUNIT get_supported_reso(NPY_DATETIMEUNIT reso):
     # If we have an unsupported reso, return the nearest supported reso.
     if reso == NPY_DATETIMEUNIT.NPY_FR_GENERIC:
         # TODO: or raise ValueError? trying this gives unraisable errors, but
@@ -572,7 +572,7 @@ cpdef NPY_DATETIMEUNIT get_supported_reso(NPY_DATETIMEUNIT reso):
     return reso
 
 
-cpdef bint is_supported_unit(NPY_DATETIMEUNIT reso):
+cdef bint is_supported_unit(NPY_DATETIMEUNIT reso):
     return (
         reso == NPY_DATETIMEUNIT.NPY_FR_ns
         or reso == NPY_DATETIMEUNIT.NPY_FR_us
@@ -581,7 +581,7 @@ cpdef bint is_supported_unit(NPY_DATETIMEUNIT reso):
     )
 
 
-cpdef str npy_unit_to_abbrev(NPY_DATETIMEUNIT unit):
+cdef str npy_unit_to_abbrev(NPY_DATETIMEUNIT unit):
     if unit == NPY_DATETIMEUNIT.NPY_FR_ns or unit == NPY_DATETIMEUNIT.NPY_FR_GENERIC:
         # generic -> default to nanoseconds
         return "ns"

pandas/_libs/tslibs/np_datetime.pyi

@@ -23,3 +23,5 @@ def add_overflowsafe(
     left: npt.NDArray[np.int64],
     right: npt.NDArray[np.int64],
 ) -> npt.NDArray[np.int64]: ...
+def get_supported_dtype(dtype: np.dtype) -> np.dtype: ...
+def is_supported_dtype(dtype: np.dtype) -> bool: ...

pandas/_libs/tslibs/np_datetime.pyx

@@ -39,6 +39,8 @@ from numpy cimport (
 )
 
 from pandas._libs.tslibs.dtypes cimport (
+    get_supported_reso,
+    is_supported_unit,
     npy_unit_to_abbrev,
     npy_unit_to_attrname,
 )
@@ -91,6 +93,28 @@ def py_get_unit_from_dtype(dtype):
     return get_unit_from_dtype(dtype)
 
 
+def get_supported_dtype(dtype: cnp.dtype) -> cnp.dtype:
+    reso = get_unit_from_dtype(dtype)
+    new_reso = get_supported_reso(reso)
+    new_unit = npy_unit_to_abbrev(new_reso)
+
+    # Accessing dtype.kind here incorrectly(?) gives "" instead of "m"/"M",
+    #  so we check type_num instead
+    if dtype.type_num == cnp.NPY_DATETIME:
+        new_dtype = np.dtype(f"M8[{new_unit}]")
+    else:
+        new_dtype = np.dtype(f"m8[{new_unit}]")
+    return new_dtype
+
+
+def is_supported_dtype(dtype: cnp.dtype) -> bool:
+    if dtype.type_num not in [cnp.NPY_DATETIME, cnp.NPY_TIMEDELTA]:
+        raise ValueError("is_unitless dtype must be datetime64 or timedelta64")
+    cdef:
+        NPY_DATETIMEUNIT unit = get_unit_from_dtype(dtype)
+    return is_supported_unit(unit)
+
+
 def is_unitless(dtype: cnp.dtype) -> bool:
     """
     Check if a datetime64 or timedelta64 dtype has no attached unit.

pandas/core/arrays/_mixins.py

@@ -13,10 +13,7 @@
 
 from pandas._libs import lib
 from pandas._libs.arrays import NDArrayBacked
-from pandas._libs.tslibs import (
-    get_unit_from_dtype,
-    is_supported_unit,
-)
+from pandas._libs.tslibs import is_supported_dtype
 from pandas._typing import (
     ArrayLike,
     AxisInt,
@@ -141,16 +138,12 @@ def view(self, dtype: Dtype | None = None) -> ArrayLike:
             cls = dtype.construct_array_type()  # type: ignore[assignment]
             dt64_values = arr.view(f"M8[{dtype.unit}]")
             return cls(dt64_values, dtype=dtype)
-        elif lib.is_np_dtype(dtype, "M") and is_supported_unit(
-            get_unit_from_dtype(dtype)
-        ):
+        elif lib.is_np_dtype(dtype, "M") and is_supported_dtype(dtype):
             from pandas.core.arrays import DatetimeArray
 
             dt64_values = arr.view(dtype)
             return DatetimeArray(dt64_values, dtype=dtype)
-        elif lib.is_np_dtype(dtype, "m") and is_supported_unit(
-            get_unit_from_dtype(dtype)
-        ):
+        elif lib.is_np_dtype(dtype, "m") and is_supported_dtype(dtype):
             from pandas.core.arrays import TimedeltaArray
 
             td64_values = arr.view(dtype)

pandas/core/arrays/datetimes.py

@@ -27,14 +27,13 @@
     astype_overflowsafe,
     fields,
     get_resolution,
-    get_supported_reso,
+    get_supported_dtype,
     get_unit_from_dtype,
     ints_to_pydatetime,
     is_date_array_normalized,
-    is_supported_unit,
+    is_supported_dtype,
     is_unitless,
     normalize_i8_timestamps,
-    npy_unit_to_abbrev,
     timezones,
     to_offset,
     tz_convert_from_utc,
@@ -712,7 +711,7 @@ def astype(self, dtype, copy: bool = True):
             self.tz is None
             and lib.is_np_dtype(dtype, "M")
             and not is_unitless(dtype)
-            and is_supported_unit(get_unit_from_dtype(dtype))
+            and is_supported_dtype(dtype)
         ):
             # unit conversion e.g. datetime64[s]
             res_values = astype_overflowsafe(self._ndarray, dtype, copy=True)
@@ -2307,7 +2306,7 @@ def _sequence_to_dt64(
     assert isinstance(result, np.ndarray), type(result)
     assert result.dtype.kind == "M"
     assert result.dtype != "M8"
-    assert is_supported_unit(get_unit_from_dtype(result.dtype))
+    assert is_supported_dtype(result.dtype)
     return result, tz
 
 
@@ -2321,14 +2320,10 @@ def _construct_from_dt64_naive(
     #  lib.is_np_dtype(data.dtype)
 
     new_dtype = data.dtype
-    data_unit = get_unit_from_dtype(new_dtype)
-    if not is_supported_unit(data_unit):
+    if not is_supported_dtype(new_dtype):
         # Cast to the nearest supported unit, generally "s"
-        new_reso = get_supported_reso(data_unit)
-        new_unit = npy_unit_to_abbrev(new_reso)
-        new_dtype = np.dtype(f"M8[{new_unit}]")
+        new_dtype = get_supported_dtype(new_dtype)
         data = astype_overflowsafe(data, dtype=new_dtype, copy=False)
-        data_unit = get_unit_from_dtype(new_dtype)
         copy = False
 
     if data.dtype.byteorder == ">":
@@ -2346,6 +2341,7 @@ def _construct_from_dt64_naive(
         if data.ndim > 1:
             data = data.ravel()
 
+        data_unit = get_unit_from_dtype(new_dtype)
         data = tzconversion.tz_localize_to_utc(
             data.view("i8"), tz, ambiguous=ambiguous, creso=data_unit
         )
@@ -2552,7 +2548,7 @@ def _validate_dt64_dtype(dtype):
 
         if (
             isinstance(dtype, np.dtype)
-            and (dtype.kind != "M" or not is_supported_unit(get_unit_from_dtype(dtype)))
+            and (dtype.kind != "M" or not is_supported_dtype(dtype))
         ) or not isinstance(dtype, (np.dtype, DatetimeTZDtype)):
             raise ValueError(
                 f"Unexpected value for 'dtype': '{dtype}'. "

pandas/core/arrays/masked.py

@@ -15,10 +15,7 @@
     lib,
     missing as libmissing,
 )
-from pandas._libs.tslibs import (
-    get_unit_from_dtype,
-    is_supported_unit,
-)
+from pandas._libs.tslibs import is_supported_dtype
 from pandas._typing import (
     ArrayLike,
     AstypeArg,
@@ -876,9 +873,7 @@ def _maybe_mask_result(
 
             return BooleanArray(result, mask, copy=False)
 
-        elif lib.is_np_dtype(result.dtype, "m") and is_supported_unit(
-            get_unit_from_dtype(result.dtype)
-        ):
+        elif lib.is_np_dtype(result.dtype, "m") and is_supported_dtype(result.dtype):
             # e.g. test_numeric_arr_mul_tdscalar_numexpr_path
             from pandas.core.arrays import TimedeltaArray
 

pandas/core/arrays/numpy_.py

@@ -8,10 +8,7 @@
 import numpy as np
 
 from pandas._libs import lib
-from pandas._libs.tslibs import (
-    get_unit_from_dtype,
-    is_supported_unit,
-)
+from pandas._libs.tslibs import is_supported_dtype
 from pandas.compat.numpy import function as nv
 
 from pandas.core.dtypes.astype import astype_array
@@ -553,9 +550,7 @@ def _cmp_method(self, other, op):
     def _wrap_ndarray_result(self, result: np.ndarray):
         # If we have timedelta64[ns] result, return a TimedeltaArray instead
         #  of a NumpyExtensionArray
-        if result.dtype.kind == "m" and is_supported_unit(
-            get_unit_from_dtype(result.dtype)
-        ):
+        if result.dtype.kind == "m" and is_supported_dtype(result.dtype):
             from pandas.core.arrays import TimedeltaArray
 
             return TimedeltaArray._simple_new(result, dtype=result.dtype)

pandas/core/arrays/timedeltas.py

@@ -19,11 +19,9 @@
     Tick,
     Timedelta,
     astype_overflowsafe,
-    get_supported_reso,
-    get_unit_from_dtype,
+    get_supported_dtype,
     iNaT,
-    is_supported_unit,
-    npy_unit_to_abbrev,
+    is_supported_dtype,
     periods_per_second,
 )
 from pandas._libs.tslibs.conversion import cast_from_unit_vectorized
@@ -352,7 +350,7 @@ def astype(self, dtype, copy: bool = True):
                     return self.copy()
                 return self
 
-            if is_supported_unit(get_unit_from_dtype(dtype)):
+            if is_supported_dtype(dtype):
                 # unit conversion e.g. timedelta64[s]
                 res_values = astype_overflowsafe(self._ndarray, dtype, copy=False)
                 return type(self)._simple_new(
@@ -1064,12 +1062,9 @@ def sequence_to_td64ns(
         copy = False
 
     elif lib.is_np_dtype(data.dtype, "m"):
-        data_unit = get_unit_from_dtype(data.dtype)
-        if not is_supported_unit(data_unit):
+        if not is_supported_dtype(data.dtype):
             # cast to closest supported unit, i.e. s or ns
-            new_reso = get_supported_reso(data_unit)
-            new_unit = npy_unit_to_abbrev(new_reso)
-            new_dtype = np.dtype(f"m8[{new_unit}]")
+            new_dtype = get_supported_dtype(data.dtype)
             data = astype_overflowsafe(data, dtype=new_dtype, copy=False)
             copy = False
 
@@ -1173,7 +1168,7 @@ def _validate_td64_dtype(dtype) -> DtypeObj:
 
     if not lib.is_np_dtype(dtype, "m"):
         raise ValueError(f"dtype '{dtype}' is invalid, should be np.timedelta64 dtype")
-    elif not is_supported_unit(get_unit_from_dtype(dtype)):
+    elif not is_supported_dtype(dtype):
         raise ValueError("Supported timedelta64 resolutions are 's', 'ms', 'us', 'ns'")
 
     return dtype

pandas/core/construction.py

@@ -24,8 +24,8 @@
 from pandas._libs import lib
 from pandas._libs.tslibs import (
     Period,
-    get_unit_from_dtype,
-    is_supported_unit,
+    get_supported_dtype,
+    is_supported_dtype,
 )
 from pandas._typing import (
     AnyArrayLike,
@@ -370,9 +370,9 @@ def array(
     #   1. datetime64[ns,us,ms,s]
     #   2. timedelta64[ns,us,ms,s]
     # so that a DatetimeArray is returned.
-    if lib.is_np_dtype(dtype, "M") and is_supported_unit(get_unit_from_dtype(dtype)):
+    if lib.is_np_dtype(dtype, "M") and is_supported_dtype(dtype):
         return DatetimeArray._from_sequence(data, dtype=dtype, copy=copy)
-    if lib.is_np_dtype(dtype, "m") and is_supported_unit(get_unit_from_dtype(dtype)):
+    if lib.is_np_dtype(dtype, "m") and is_supported_dtype(dtype):
         return TimedeltaArray._from_sequence(data, dtype=dtype, copy=copy)
 
     elif lib.is_np_dtype(dtype, "mM"):
@@ -490,12 +490,14 @@ def ensure_wrapped_if_datetimelike(arr):
         if arr.dtype.kind == "M":
             from pandas.core.arrays import DatetimeArray
 
-            return DatetimeArray._from_sequence(arr)
+            dtype = get_supported_dtype(arr.dtype)
+            return DatetimeArray._from_sequence(arr, dtype=dtype)
 
         elif arr.dtype.kind == "m":
             from pandas.core.arrays import TimedeltaArray
 
-            return TimedeltaArray._from_sequence(arr)
+            dtype = get_supported_dtype(arr.dtype)
+            return TimedeltaArray._from_sequence(arr, dtype=dtype)
 
     return arr
 

pandas/core/dtypes/cast.py

@@ -36,8 +36,7 @@
     OutOfBoundsTimedelta,
     Timedelta,
     Timestamp,
-    get_unit_from_dtype,
-    is_supported_unit,
+    is_supported_dtype,
 )
 from pandas._libs.tslibs.timedeltas import array_to_timedelta64
 from pandas.errors import (
@@ -1266,8 +1265,7 @@ def _ensure_nanosecond_dtype(dtype: DtypeObj) -> None:
         pass
 
     elif dtype.kind in "mM":
-        reso = get_unit_from_dtype(dtype)
-        if not is_supported_unit(reso):
+        if not is_supported_dtype(dtype):
             # pre-2.0 we would silently swap in nanos for lower-resolutions,
             #  raise for above-nano resolutions
             if dtype.name in ["datetime64", "timedelta64"]: