Skip to content

[WIP][SPARK-52288][PS] Avoid INVALID_ARRAY_INDEX in split when ANSI mode is on #51006

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 1 commit into
base: master
Choose a base branch
from
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 7 additions & 1 deletion python/pyspark/pandas/strings.py
Original file line number Diff line number Diff line change
Expand Up @@ -2031,7 +2031,13 @@ def pudf(s: pd.Series) -> pd.Series:
if expand:
psdf = psser.to_frame()
scol = psdf._internal.data_spark_columns[0]
spark_columns = [scol[i].alias(str(i)) for i in range(n + 1)]

if ps.get_option("compute.ansi_mode_support"):
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It seems we can always use the new branch, and remove the configuration read here (which needs one Config RPC)

spark_columns = [
F.try_element_at(scol, F.lit(i + 1)).alias(str(i)) for i in range(n + 1)
Copy link

@abhiips07 abhiips07 May 26, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we try to calculate f.lit(i+1) outside loop? since this might avoid function calls and object creation during loop execution

literals = [F.lit(i + 1) for i in range(n + 1)]

spark_columns = [F.try_element_at(scol, lit).alias(str(i)) for i, lit in enumerate(literals)]

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for suggestion!
There might not be a significant perf difference between creating F.lit inside the loop or beforehand, it's just wrapping a Python literal into a Spark expression, which aren’t executed immediately(just nodes in the DAG), and will be deduplicated by Catalyst. With that being said I’d like to keep the original for simplicity, but feel free to share if you have other opinions!

]
else:
spark_columns = [scol[i].alias(str(i)) for i in range(n + 1)]
column_labels = [(i,) for i in range(n + 1)]
internal = psdf._internal.with_new_columns(
spark_columns,
Expand Down
2 changes: 2 additions & 0 deletions python/pyspark/pandas/tests/series/test_string_ops_adv.py
Original file line number Diff line number Diff line change
Expand Up @@ -185,6 +185,8 @@ def test_string_split(self):
with self.assertRaises(NotImplementedError):
self.check_func(lambda x: x.str.split(expand=True))

self.check_func_on_series(lambda x: repr(x.str.split("-", n=1, expand=True)), pser)

@unittest.skipIf(is_ansi_mode_test, ansi_mode_not_supported_message)
def test_string_rsplit(self):
self.check_func_on_series(lambda x: repr(x.str.rsplit()), self.pser[:-1])
Expand Down