Implement cudf-polars datetime extraction methods #16500
Conversation
lithomas1
added
feature request
lithomas1
requested review from
bdice and
galipremsagar
and removed request for
a team
github-actions
bot
added
Python
|
/ok to test |
|
/ok to test |
|
/ok to test |
|
/ok to test |
| "nanosecond", | ||
| ] | ||
|
|
||
| unsupported_extract_fields = ["millennium"] |
There was a problem hiding this comment.
I guess we could extract this from years rather than not support it…
There was a problem hiding this comment.
This was a rough solution for maintaining 100% coverage of the code, specifically getting us to hit the error line that complains about a rust node that is exposed but that we don't support. Because of this, adding millennium would in some sense just kick the can down the road until we come to a day where all the rust nodes are supported in python as well and we can get rid of the error.
Instead of doing that I went with an approach that we liked in a similar situation where we monkeypatch things to trigger an error at the desired location, regardless of what we support. See the latest commits for details and let me know what you think.
Co-authored-by: Bradley Dice <[email protected]> Co-authored-by: Lawrence Mitchell <[email protected]>
Co-authored-by: Lawrence Mitchell <[email protected]>
|
/ok to test |
|
Can fix the test that asserts that dtypes don't match for |
|
/ok to test |
| # libcudf can lose data here. | ||
| # https://github.com/rapidsai/cudf/issues/16196 | ||
| assert_gpu_result_equal(q) | ||
| q = ldf.select(field(pl.col("dates").dt)) |
There was a problem hiding this comment.
Now the dtypes should match, so we need to remove the check_dtypes=False part below
|
/ok to test |
Description
Checklist