De-duplicate add_offset_array methods #19835
Conversation
pandas/core/indexes/datetimelike.py
Outdated
| kwargs = {} | ||
| if not is_period_dtype(self): | ||
| kwargs['freq'] = 'infer' | ||
| return self.__class__(res_values, **kwargs) |
There was a problem hiding this comment.
Changed it to shallow_copy and it broke several tests, so just pushed with it changed back to self.__class__. For some reason DatetimeIndex._simple_new returns an ndarray if its input values have object-dtype, which is the case here.
There was a problem hiding this comment.
so try using _shallow_copy_with_infer, we by definition never ever use self.__class__ in Index construction.
| ---------- | ||
| other : Index, np.ndarray | ||
| object-dtype containing pd.DateOffset objects | ||
| op : {operator.add, operator.sub} |
There was a problem hiding this comment.
can you assert on the operator
Codecov Report
@@ Coverage Diff @@
## master #19835 +/- ##
==========================================
+ Coverage 91.57% 91.58% +<.01%
==========================================
Files 150 150
Lines 48903 48870 -33
==========================================
- Hits 44785 44758 -27
+ Misses 4118 4112 -6
Continue to review full report at Codecov.
|
|
thanks! |
Combine DTI/TDI/PI add_offset_array and sub_offset_array methods.
Re-order
__add__and__sub__if/else branches to put scalars up front.This was going to include fixes for DTI/TDI with ndarray[timedelta64]/ndarray[datetime64], but the diff got big enough to split that off to the next one.