[
https://issues.apache.org/jira/browse/SPARK-59512?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Uroš Bojanić updated SPARK-59512:
---------------------------------
Affects Version/s: 4.4.0
(was: 5.0.0)
> StructField.fromJson mutates collation metadata and fails on repeated parsing
> -----------------------------------------------------------------------------
>
> Key: SPARK-59512
> URL: https://issues.apache.org/jira/browse/SPARK-59512
> Project: Spark
> Issue Type: Bug
> Components: PySpark, Tests
> Affects Versions: 4.4.0
> Reporter: Akshay Thorat
> Assignee: Akshay Thorat
> Priority: Major
> Labels: pull-request-available
>
> h2. Problem
> Parsing a cached schema dictionary changes its collation metadata. A second
> StructType.fromJson call on the same dictionary raises AssertionError because
> the provider prefix was removed.
> h2. Reproduction
> Reproduced from upstream master 39776477a3d using repository Python source
> (PySpark 5.0.0.dev0), Python 3.10.11 and NumPy 2.2.6. No SparkSession is
> required.
> {code:python}
> from pyspark.sql.types import StringType, StructField, StructType
> payload = StructType([StructField("s", StringType("UTF8_LCASE"))]).jsonValue()
> StructType.fromJson(payload)
> StructType.fromJson(payload) # AssertionError
> {code}
> h2. Expected behavior
> Both calls should return the same schema and leave payload unchanged.
> h2. Proposed fix and verification
> Copy the collation lookup before stripping provider names in
> StructField.fromJson. Regression tests cover repeated parsing with collated
> strings, arrays, maps, and nested structs. The new regression tests fail
> before the fix and pass afterward. Python compilation, custom-error checks
> and Ruff checks pass.
> Pull request: https://github.com/apache/spark/pull/58787
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
