[SPARK-51206][PYTHON][CONNECT] Move Arrow conversion helpers out of Spark Connect #49941
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…park connect module
|
cc @HyukjinKwon |
| elif isinstance(dataType, ArrayType): | ||
| return ArrowTableToRowsConversion._need_converter(dataType.elementType) | ||
| elif isinstance(dataType, MapType): | ||
| # Different from PySpark, here always needs conversion, |
There was a problem hiding this comment.
I remember there is another version of conversion for PySpark classic, do we plan to unify them in the future?
There was a problem hiding this comment.
Are you referring to pyspark.sql.types._create_converter? Seems like there are some differences, for example ArrowTableToRowsConversion handles UDT and variant but the one in types.py doesn't.
There was a problem hiding this comment.
Can we check where _create_converteris used and see if we can merge these two methods?
There was a problem hiding this comment.
That _create_converter turns dicts (StructType) into tuples of values. It has nothing to do with arrow conversion. Sorry for the confusion 🤦
There was a problem hiding this comment.
actually, I meant python/pyspark/sql/pandas/conversion.py. We can investigate it later.
HyukjinKwon
reviewed
Feb 18, 2025
HyukjinKwon
reviewed
Feb 18, 2025
HyukjinKwon
reviewed
Feb 18, 2025
zhengruifeng
approved these changes
Feb 24, 2025
HyukjinKwon
approved these changes
Feb 24, 2025
allisonwang-db
approved these changes
Feb 27, 2025
|
Merged to master and branch-4.0. |
HyukjinKwon
pushed a commit
that referenced
this pull request
Feb 27, 2025
…park Connect ### What changes were proposed in this pull request? Refactor `pyspark.sql.connect.conversion` to move `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` into `pyspark.sql.conversion`. The reason is that `pyspark.sql.connect.conversion` checks for Spark Connect dependencies such as `pyarrow`, `grpcio` and `pandas`, but `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` only need `pyarrow`. `pyspark.sql.connect.conversion` still re-exports the two classes for backward compatibility. ### Why are the changes needed? Python Data Sources should work without Spark Connect dependencies but currently it imports `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` from `pyspark.sql.connect.conversion` making it require unnecessary dependencies. This change moves these two classes to `pyspark.sql.conversion` so that Python Data Sources runs without Spark Connect dependencies. ### Does this PR introduce _any_ user-facing change? Relaxed requirements for using Python Data Sources. ### How was this patch tested? Existing tests should make sure that the changes don't break anything. Manually tested to ensure that Python Data Sources can run without grpcio and pandas. ### Was this patch authored or co-authored using generative AI tooling? No Closes #49941 from wengh/spark-51206-pyds-fix-dependency. Authored-by: Haoyu Weng <wenghy02@gmail.com> Signed-off-by: Hyukjin Kwon <gurwls223@apache.org> (cherry picked from commit 727167a) Signed-off-by: Hyukjin Kwon <gurwls223@apache.org>
Pajaraja
pushed a commit
to Pajaraja/spark
that referenced
this pull request
Mar 6, 2025
…park Connect ### What changes were proposed in this pull request? Refactor `pyspark.sql.connect.conversion` to move `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` into `pyspark.sql.conversion`. The reason is that `pyspark.sql.connect.conversion` checks for Spark Connect dependencies such as `pyarrow`, `grpcio` and `pandas`, but `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` only need `pyarrow`. `pyspark.sql.connect.conversion` still re-exports the two classes for backward compatibility. ### Why are the changes needed? Python Data Sources should work without Spark Connect dependencies but currently it imports `LocalDataToArrowConversion` and `ArrowTableToRowsConversion` from `pyspark.sql.connect.conversion` making it require unnecessary dependencies. This change moves these two classes to `pyspark.sql.conversion` so that Python Data Sources runs without Spark Connect dependencies. ### Does this PR introduce _any_ user-facing change? Relaxed requirements for using Python Data Sources. ### How was this patch tested? Existing tests should make sure that the changes don't break anything. Manually tested to ensure that Python Data Sources can run without grpcio and pandas. ### Was this patch authored or co-authored using generative AI tooling? No Closes apache#49941 from wengh/spark-51206-pyds-fix-dependency. Authored-by: Haoyu Weng <wenghy02@gmail.com> Signed-off-by: Hyukjin Kwon <gurwls223@apache.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
Refactor
pyspark.sql.connect.conversionto moveLocalDataToArrowConversionandArrowTableToRowsConversionintopyspark.sql.conversion.The reason is that
pyspark.sql.connect.conversionchecks for Spark Connect dependencies such aspyarrow,grpcioandpandas, butLocalDataToArrowConversionandArrowTableToRowsConversiononly needpyarrow.pyspark.sql.connect.conversionstill re-exports the two classes for backward compatibility.Why are the changes needed?
Python Data Sources should work without Spark Connect dependencies but currently it imports
LocalDataToArrowConversionandArrowTableToRowsConversionfrompyspark.sql.connect.conversionmaking it require unnecessary dependencies. This change moves these two classes topyspark.sql.conversionso that Python Data Sources runs without Spark Connect dependencies.Does this PR introduce any user-facing change?
Relaxed requirements for using Python Data Sources.
How was this patch tested?
Existing tests should make sure that the changes don't break anything.
Manually tested to ensure that Python Data Sources can run without grpcio and pandas.
Was this patch authored or co-authored using generative AI tooling?
No