ENH: Expose split_part to Python API via pylibcudf #21068
Conversation
github-actions
bot
added
Python
davidwendt
added
feature request
|
You need to fix the style errors identified by the pre-commit.ci runner here. |
github-actions
bot
added
libcudf
|
pre-commit.ci autofix |
|
Something has gone wrong with your commit. Surely you do not mean to erase all of the .sh files. |
Umang-projects
force-pushed
the
feature/add-split-part
branch
from
40c99a0 to
cbab72f
Compare
|
pre-commit.ci autofix |
|
Also, could you add tests for when the delimiter is empty -- indicates split on whitespace. |
Co-authored-by: David Wendt <[email protected]>
|
pre-commit.ci autofix |
| from cudf.testing._utils import assert_eq | ||
|
|
||
|
|
||
| def test_split_part(): |
There was a problem hiding this comment.
Seems these should be moved to the cudf/python/cudf/cudf/tests/series/accessors/test_str.py
There was a problem hiding this comment.
This file should be deleted if everything has been moved to test_str.py.
Co-authored-by: David Wendt <[email protected]>
|
pre-commit.ci autofix |
| from cudf.testing._utils import assert_eq | ||
|
|
||
|
|
||
| def test_split_part(): |
There was a problem hiding this comment.
This file should be deleted if everything has been moved to test_str.py.
| # SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. | ||
| # SPDX-License-Identifier: Apache-2.0 | ||
|
|
||
| # SPDX-License-Identifier: Apache-2.0 |
There was a problem hiding this comment.
I think this source should be moved to https://github.com/rapidsai/cudf/blob/main/python/pylibcudf/tests/test_string_split_split.py
Co-authored-by: David Wendt <[email protected]>
Co-authored-by: David Wendt <[email protected]>
|
pre-commit.ci autofix |
Co-authored-by: David Wendt <[email protected]>
|
/ok to test f3cb63f |
Co-authored-by: David Wendt <[email protected]>
Co-authored-by: David Wendt <[email protected]>
|
/ok to test 7a75d80 |
| if delimiter is None: | ||
| delimiter = "" | ||
| return self._return_or_inplace( | ||
| self._column.split_part(delimiter, index) |
There was a problem hiding this comment.
Based on the delimiter annotation, you'll need to call plc.Scalar.from_py(delimiter) before passing it to split_part
3 participants
Description
This PR implements the Python bindings and API for
cudf::strings::split_part, as discussed in Issue #21042.This allows users to extract a specific token from a split string without materializing the entire list, significantly improving performance for ETL/Log parsing workloads.
Changes
split.pxdandsplit.pyx..str.split_part(delimiter, index)method toStringMethodsincore/column/string.py.tests/test_string.py.Checklist
Note to Reviewers
I am developing in a constrained environment (Google Colab) and could not run the full local build/test suite due to environment limitations. I am opening this as a Draft PR to rely on the CI/CD pipeline for compilation verification and testing.