[jira] [Commented] (FLINK-28956) FIx non-multi insert statement fall into multi insert logic in Hive dialect

["luoyuxia (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-28956?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17579501#comment-17579501 ] 

luoyuxia commented on FLINK-28956:
----------------------------------

To fix it, we need a more strict check.

> FIx non-multi insert statement fall into multi insert logic  in Hive dialect
> ----------------------------------------------------------------------------
>
>                 Key: FLINK-28956
>                 URL: https://issues.apache.org/jira/browse/FLINK-28956
>             Project: Flink
>          Issue Type: Bug
>          Components: Connectors / Hive
>            Reporter: luoyuxia
>            Priority: Major
>
> In [FLINK-27387|https://issues.apache.org/jira/browse/FLINK-27387], we support multi insert statment in Hive dialect,  but the check for  multi insert statment  is not strict which will result non-multi insert statement fall into such logic.
> For example, 
>  
> {code:java}
> with cte as (select t.a as a,t.a as b,t.a as c from t where t.b is null) select * from cte {code}
>  
> . The AST for it is [TOK_CTE, TOK_FROM, TOK_INSERT], then it 
> will be mistaken as multi-insert statement
>  
> {code:java}
> private boolean isMultiDestQuery(HiveParserASTNode astNode) {
>     // Hive's multi dest insert will always be [FROM, INSERT+]
>     // so, if it's children count is more than 2, it should be a multi-dest query
>     return astNode.getChildCount() > 2;
> } {code}
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)