[HIVE-25459] Wrong VectorUDFMapIndexBaseScalar child class is used, leading to ClassCastException

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Assignee: László Bodor
• Reporter: László Bodor
• Created: 17/Aug/21 12:28
• Updated: 31/Aug/21 07:12

Description

given this query:

CREATE TABLE map_of_doubles (
id int,
doubleMap map<double, double>
) stored as orc;
insert overwrite table map_of_doubles SELECT 1, MAP(CAST(1.0 as DOUBLE), null, CAST(2.0 as DOUBLE), CAST(3.0 as DOUBLE));

select id, doubleMap from map_of_doubles;
select id, doubleMap[1] from map_of_doubles group by id, doubleMap[1]; -- this fails

error is:

Caused by: java.lang.ClassCastException: org.apache.hadoop.hive.ql.exec.vector.DoubleColumnVector cannot be cast to org.apache.hadoop.hive.ql.exec.vector.LongColumnVector
	at org.apache.hadoop.hive.ql.exec.vector.expressions.VectorUDFMapIndexLongScalar.findScalarInMap(VectorUDFMapIndexLongScalar.java:67)
	at org.apache.hadoop.hive.ql.exec.vector.expressions.VectorUDFMapIndexBaseScalar.evaluate(VectorUDFMapIndexBaseScalar.java:132)
	at org.apache.hadoop.hive.ql.exec.vector.VectorSelectOperator.process(VectorSelectOperator.java:146)
	... 23 more

I found this error while was trying to write q test cases for all data types in HIVE-23688, so this issue needs to be addressed first
HIVE-23688 is Parquet-specific, this one is not, it can be reproduced for ORC and Parquet too

Issue Links

• relates to: HIVE-23688 Vectorization: IndexArrayOutOfBoundsException For map type column which includes null value