Matt Zhang created SPARK-57624:
----------------------------------
Summary: from_xml to variant should honor the parse mode
Key: SPARK-57624
URL: https://issues.apache.org/jira/browse/SPARK-57624
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 5.0.0
Reporter: Matt Zhang
from_xml(..., 'variant', map('mode', ...)) ignores the parse mode.
XmlToStructsEvaluator.evaluate calls StaxXmlParser.parseVariant directly for
variant output, bypassing the FailureSafeParser the struct path uses, so a
malformed record (e.g. an illegal XML control char) aborts the whole job even
under PERMISSIVE.
It also fails under whole-stage codegen: XmlToStructs.doGenCode hardcasts the
result to InternalRow (compile error for variant output) and does not
null-check the result.
PR: https://github.com/apache/spark/pull/56681
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
