Hyukjin Kwon created SPARK-61014:
------------------------------------
Summary: Run foreach and foreachPartition, and return the
DataFrame of toJSON, in the Python client as PySpark Connect does
Key: SPARK-61014
URL: https://issues.apache.org/jira/browse/SPARK-61014
Project: Spark
Issue Type: Bug
Components: Connect
Affects Versions: connect-rust-4.3.0
Reporter: Hyukjin Kwon
In PySpark Connect, {{df.foreach(f)}} calls {{f}} with each row, by a UDF of a
struct of the row, {{df.foreachPartition(f)}} calls {{f}} with an iterator of
the rows of each partition, by {{mapInArrow}}, and {{df.toJSON()}} returns the
DataFrame of the JSON string of each row in the column {{value}},
{{df.select(to_json(struct(col("*"))).alias("value"))}}.
The Rust-backed Python client's {{foreach}} and {{foreachPartition}} fail on
the worker with {{TypeError: cannot unpack non-iterable function object}}, as
the command of their UDF is the function rather than the pair of the function
and its return type; and {{toJSON()}} raises {{NotImplementedError}}. Run them
as PySpark does.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
