[
https://issues.apache.org/jira/browse/FLINK-20373?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17242359#comment-17242359
]
HideOnBush commented on FLINK-20373:
------------------------------------
{code:java}
class ExplodeArray() extends TableFunction[AnyRef] {
//FLINK SQL ARRAY<ROW> to ROW
def eval(array: Array[AnyRef])(): Unit = {
if (array == null) {
return
}
array.foreach(collect)
}
override def getTypeInference(typeFactory: DataTypeFactory): TypeInference = {
new TypeInference.Builder().inputTypeStrategy(new
ArrayInputTypeStrategy()).outputTypeStrategy(new TypeStrategy {
override def inferType(callContext: CallContext): Optional[DataType] = {
Optional.of(callContext.getArgumentDataTypes.get(0).getChildren.get(0))
}
})
}.build()
}
{code}
I Code the UDF to Explode Flink SQL JSON Array
> Flink table jsonArray access all
> --------------------------------
>
> Key: FLINK-20373
> URL: https://issues.apache.org/jira/browse/FLINK-20373
> Project: Flink
> Issue Type: New Feature
> Components: Table SQL / Client
> Affects Versions: 1.11.2
> Reporter: HideOnBush
> Priority: Major
>
> The official jsonArray is provided, and the array is also provided to access
> Row elements based on the subscript. Should we also consider the length of
> each jsonArray, and if the subscript is passed, the code will become longer
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
