GH1379 Drop OffsetSeries replacing it with Series[BaseOffset] #1390
Conversation
pyproject.toml
Outdated
| SQLAlchemy = ">=2.0.39" | ||
| types-python-dateutil = ">=2.8.19" | ||
| beautifulsoup4 = "<=4.13.5" | ||
| beautifulsoup4 = ">=4.13.5" |
There was a problem hiding this comment.
| beautifulsoup4 = ">=4.13.5" | |
| beautifulsoup4 = ">=4.14.2" |
tests/test_timefuncs.py
Outdated
| o_s = cast( | ||
| "OffsetSeries", pd.Series([pd.DateOffset(days=1), pd.DateOffset(days=2)]) | ||
| "pd.Series[BaseOffset]", | ||
| pd.Series([pd.DateOffset(days=1), pd.DateOffset(days=2)]), | ||
| ) |
There was a problem hiding this comment.
Can you see if we can create a Series.__new__() overload that would accept a sequence of BaseOffset so that we don't need the cast here?
There was a problem hiding this comment.
So turns out this creates a Series[DateOffset] and not Series[BaseOffset] that is why this is an issue, not sure why though it is not working since it is supposed to inherit from BaseOffset.
There was a problem hiding this comment.
So I checked and seems like Series[DateOffset] does not match with Series[BaseOffset] even though DateOffset is always a BaseOffset:
assert_type(o_s, "pd.Series[BaseOffset]") ■ "assert_type" mismatch: expected "Series[BaseOffset]" but received "Series[DateOffset]"
There was a problem hiding this comment.
I think you need to add an overload for Series.__new__() that accepts a sequence of BaseOffset and then it should work. Similar to how there are overloads that produce Series[Timestamp], Series[Timedelta], etc.
There was a problem hiding this comment.
Removed the to_numpy overloads and added the new overloads and it is looking good!
assert_type()to assert the type of any return valueAlso dropped the locking of the version of
beautifulsoup, it seems like they shipped another version that fixed the issue.