[
https://issues.apache.org/jira/browse/FLINK-40214?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Qilong Wang updated FLINK-40214:
--------------------------------
Component/s: Table SQL / Planner
Fix Version/s: 2.3.0
Affects Version/s: 2.3.0
> JSON_OBJECTAGG produces incorrect keys when key and value use the same column
> -----------------------------------------------------------------------------
>
> Key: FLINK-40214
> URL: https://issues.apache.org/jira/browse/FLINK-40214
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Affects Versions: 2.3.0
> Reporter: Qilong Wang
> Priority: Major
> Fix For: 2.3.0
>
>
> `JSON_OBJECTAGG` produces incorrect JSON object keys when its key and value
> arguments reference the same input column.
> The generated keys contain additional escaped quotation marks instead of
> preserving the original column values.
> *Steps to Reproduce*
> Run the following query:
> {code:java}
> SELECT JSON_OBJECTAGG(f0 VALUE f0)
> FROM (VALUES ('A'), ('B')) AS T(f0); {code}
> The issue can also be reproduced with a group window aggregation:
> {code:java}
> SELECT JSON_OBJECTAGG(f0 VALUE f0)
> FROM T
> GROUP BY TUMBLE(rowtime_col, INTERVAL '5' SECOND); {code}
> *Expected Result*
> {"A":"A","B":"B"}
> *Actual Result*
> The object keys contain additional escaped quotation marks, similar to:
> {"\"A\"":"A","\"B\"":"B"}
> *Impact*
> The issue affects JSON_OBJECTAGG queries where the key and value use the same
> input column. It can occur in both regular and window aggregations, producing
> JSON objects with incorrect key names.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
