201573 opened a new pull request, #55755: URL: https://github.com/apache/spark/pull/55755
### What changes were proposed in this pull request? This PR refactors `SQL_SCALAR_PANDAS_ITER_UDF` so that the scalar iterator Pandas UDF logic is handled inside `read_udfs()`. The worker now uses `ArrowStreamSerializer` for the scalar iterator Pandas UDF I/O path, converts input Arrow batches to pandas arguments in `read_udfs()`, validates the iterator and element return types there, and converts the pandas outputs back to Arrow record batches. This keeps the existing row-count checks and return validation while making the serializer a pure Arrow I/O layer for this eval type. ### Why are the changes needed? This is part of SPARK-55388. The scalar iterator Pandas UDF path should be self-contained in `read_udfs()` instead of relying on `ArrowStreamPandasUDFSerializer` to perform UDF-specific logic. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Added a test for scalar iterator Pandas UDFs with struct input and struct output. Verified locally: - `SPARK_HOME=/tmp/spark-56757-wt PYTHON_EXECUTABLE=/tmp/spark-venv-56757/bin/python ./python/run-tests --python-executables /tmp/spark-venv-56757/bin/python --testnames 'pyspark.sql.tests.pandas.test_pandas_udf_scalar ScalarPandasUDFTests.test_scalar_iter_udf_struct_input_and_output' -p 1` - `SPARK_HOME=/tmp/spark-56757-wt PYTHON_EXECUTABLE=/tmp/spark-venv-56757/bin/python ./python/run-tests --python-executables /tmp/spark-venv-56757/bin/python --testnames 'pyspark.sql.tests.pandas.test_pandas_udf_scalar ScalarPandasUDFTests.test_vectorized_udf_invalid_length,pyspark.sql.tests.pandas.test_pandas_udf_scalar ScalarPandasUDFTests.test_vectorized_udf_chained_struct_type,pyspark.sql.tests.pandas.test_pandas_udf_scalar ScalarPandasUDFTests.test_scalar_iter_pandas_udf_with_single_output_batch' -p 1` - `PYTHON_EXECUTABLE=/tmp/spark-venv-56757/bin/python ./dev/lint-python --compile` - `/tmp/spark-venv-56757/bin/ruff format --check python/pyspark/worker.py python/pyspark/sql/tests/pandas/test_pandas_udf_scalar.py` - `/tmp/spark-venv-56757/bin/ruff check python/pyspark/worker.py python/pyspark/sql/tests/pandas/test_pandas_udf_scalar.py` - `git diff --check` ### Was this patch authored or co-authored using generative AI tooling? Generated-by: OpenAI Codex (GPT-5) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
