FIX BUG: Timestamp __add__/__sub__ DateOffset with nanoseconds lost. #43968
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -83,6 +83,7 @@ from pandas._libs.tslibs.np_datetime cimport ( | |
| from pandas._libs.tslibs.np_datetime import OutOfBoundsDatetime | ||
|
|
||
| from pandas._libs.tslibs.offsets cimport ( | ||
| is_dateoffset_object, | ||
| is_offset_object, | ||
| to_offset, | ||
| ) | ||
|
|
@@ -283,8 +284,11 @@ cdef class _Timestamp(ABCTimestamp): | |
| cdef: | ||
| int64_t nanos = 0 | ||
|
|
||
| if is_dateoffset_object(other): | ||
jreback marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| nanos += other.nanoseconds | ||
| other = other._offset | ||
| if is_any_td_scalar(other): | ||
| nanos += delta_to_nanoseconds(other) | ||
|
There was a problem hiding this comment. are there any cases that go through the There was a problem hiding this comment. If a case go through the There was a problem hiding this comment.
Doesn't it sometimes convert to a relativedelta object (which isnt a td_scalar)? There was a problem hiding this comment. @jbrockmendel According to the codes below, we can see self._offset is a relativedelta object only if there is no nanosecond / nanoseconds keyword parameter. kwds_no_nanos = dict(
(k, v) for k, v in kwds.items()
if k not in ('nanosecond', 'nanoseconds')
)
# TODO: Are nanosecond and nanoseconds allowed somewhere?
_kwds_use_relativedelta = ('years', 'months', 'weeks', 'days',
'year', 'month', 'week', 'day', 'weekday',
'hour', 'minute', 'second', 'microsecond')
use_relativedelta = False
if len(kwds_no_nanos) > 0:
if any(k in _kwds_use_relativedelta for k in kwds_no_nanos):
offset = relativedelta(**kwds_no_nanos)
use_relativedelta = True |
||
| result = type(self)(self.value + nanos, tz=self.tzinfo) | ||
| if result is not NaT: | ||
| result._set_freq(self._freq) # avoid warning in constructor | ||
|
|
@@ -307,7 +311,6 @@ cdef class _Timestamp(ABCTimestamp): | |
| elif not isinstance(self, _Timestamp): | ||
| # cython semantics, args have been switched and this is __radd__ | ||
| return other.__add__(self) | ||
| return NotImplemented | ||
|
|
||
| def __sub__(self, other): | ||
|
|
||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
can you add a docstring, specifically make it clear when to use this vs is_offset_object
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sure, that makes sense.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the comment you have here is not useful, add a doc-string not dependent on a particular patch.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe this function is not that necessary. It could be enough by using
is_offset_object(obj) and hasattr(obj, "nanoseconds")to determine the need to deal with "nanoseconds".There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
When calling
_TimeStamp.__add__(self, other)with other is an Offset object, it can not meet any condition of the if-else chain. And it will call_TimeStamp.__add__(self, other._offset)instead. I cannot find where this logic is implemented, but I guess we could explicitly check the Offset type in the if-else chain.