REF: Stop mixing DTA/TDA into DTI/TDI #24476
Conversation
Codecov Report
@@ Coverage Diff @@
## master #24476 +/- ##
==========================================
- Coverage 92.32% 92.32% -0.01%
==========================================
Files 166 166
Lines 52298 52359 +61
==========================================
+ Hits 48285 48339 +54
- Misses 4013 4020 +7
Continue to review full report at Codecov.
|
Codecov Report
@@ Coverage Diff @@
## master #24476 +/- ##
==========================================
- Coverage 92.32% 92.31% -0.01%
==========================================
Files 166 166
Lines 52328 52379 +51
==========================================
+ Hits 48310 48354 +44
- Misses 4018 4025 +7
Continue to review full report at Codecov.
|
pandas/core/indexes/datetimes.py
Outdated
|
|
||
| return cls(dtarr, name=name) | ||
| dtarr = DatetimeArray._generate_range( | ||
| start, end, periods, |
There was a problem hiding this comment.
should the warning message actually be in DTA._generate_range?
There was a problem hiding this comment.
No, the non-deprecated way it gets called is via date_range
pandas/core/indexes/datetimes.py
Outdated
| @@ -286,16 +281,16 @@ def __new__(cls, data=None, | |||
| verify_integrity = True | |||
|
|
|||
| if data is None: | |||
| dtarr = DatetimeArray._generate_range( | |||
There was a problem hiding this comment.
Can this change be (partially) reverted? We want the warning after the array creation, in case the array creation raises an exception.
I'm not sure whether you want cls(...) or cls._simple_new right now. Probably simple_new.
There was a problem hiding this comment.
We want the warning after the array creation, in case the array creation raises an exception.
Makes sense, will change.
I'm not sure whether you want cls(...) or cls._simple_new right now. Probably simple_new
_simple_new. It's performant and correct regardless of what becomes of the discussed changes in DTA/TDA constructors/validation.
pandas/core/indexes/datetimes.py
Outdated
| _has_same_tz = ea_passthrough("_has_same_tz") | ||
| __array__ = ea_passthrough("__array__") | ||
|
|
||
| def round(self, freq, ambiguous='raise', nonexistent='raise'): |
There was a problem hiding this comment.
It looks like you can (maybe?) remove road, floor, and ciel. I haven't run the tests, but a manual example works.
They should be dispatched via DatetimeArray._datetimelike_methods.
There was a problem hiding this comment.
excellent, I'll give it a shot
There was a problem hiding this comment.
tentatively yes. we'll see if the CI agrees in a bit
|
|
||
| self._freq = to_offset(value) | ||
|
|
||
| def __getitem__(self, key): |
There was a problem hiding this comment.
In #24024 we move __getitem__ to the base DatetimeIndexOpsMixin and DatetimeArrayMixin. That seems relatively easy to do if we're already changing __getitem__ here and in TimedeltaIndex, but it may not be possible yet.
|
can you merge master |
jreback
added
Datetime
|
lgtm. @TomAugspurger |
|
Thanks. I'll merge this into #24024. |
|
|
||
| _timezone = cache_readonly(DatetimeArray._timezone.fget) | ||
| is_normalized = cache_readonly(DatetimeArray.is_normalized.fget) | ||
| _resolution = cache_readonly(DatetimeArray._resolution.fget) | ||
|
|
||
| strftime = ea_passthrough("strftime") |
There was a problem hiding this comment.
This removed the docstring. Is there a TODO / issue to add it back?
| return type(self)(result, name=self.name) | ||
|
|
||
| @property | ||
| def freq(self): # TODO: get via eadata |
There was a problem hiding this comment.
The same here, this had a docstring before
* implement _index_data parts of pandas-dev#24024 * implement _eadata, dispatch arithmetic methods to it * dont mix DatetimeLikeArrayMixin into DatetimeIndexOpsMixin * dont inherit TimedeltaIndex from TimedeltaArray * dont inherit from DatetimeArray * use ea_passthrough * remove previously-overriden overridings * stop double-mixing * stop over-writing * handle+test object arrays * Remove unused import * flake8 fixup * edits per comments
* implement _index_data parts of pandas-dev#24024 * implement _eadata, dispatch arithmetic methods to it * dont mix DatetimeLikeArrayMixin into DatetimeIndexOpsMixin * dont inherit TimedeltaIndex from TimedeltaArray * dont inherit from DatetimeArray * use ea_passthrough * remove previously-overriden overridings * stop double-mixing * stop over-writing * handle+test object arrays * Remove unused import * flake8 fixup * edits per comments
part of the #24024 process
along the way fixes+tests
DTA.__eq__(ndarray[object])bug