[
https://issues.apache.org/jira/browse/FLINK-18590?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17191939#comment-17191939
]
silence commented on FLINK-18590:
---------------------------------
Can we do this in the form of computed columns like this:
CREATE TABLE source (
message as rowToColumn(), -- type:String,value:[\{"col1": 1, "col2":
"hello"},\{"col1": 2, "col2": "flink"}]
messageAsString as rowToColumn(String), -- type:String,value:[\{"col1": 1,
"col2": "hello"},\{"col1": 2, "col2": "flink"}]
messageAsArray as rowToColumn(Array<Row<col1 Int,col2 String>>) --
type:Array<Row<col1 Int,col2 String>>,value:[\{"col1": 1, "col2":
"hello"},\{"col1": 2, "col2": "flink"}]
) WITH (
...
'format' = 'json'
)
rowToColumn([SpecificType]) is a function, it can convert an entire row of
records into a column with specific type,Then we can analyze it flexibly in SQL
through JSON UDF
> Support json array explode to multi messages
> --------------------------------------------
>
> Key: FLINK-18590
> URL: https://issues.apache.org/jira/browse/FLINK-18590
> Project: Flink
> Issue Type: Improvement
> Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile)
> Reporter: Benchao Li
> Assignee: Benchao Li
> Priority: Major
> Labels: pull-request-available
>
> Currently our {{DeserializationSchema}} has supported to emit multi messages
> for one input record. There is some cases that the json data is not a json
> object in top level, for example:
> {code:SQL}
> CREATE TABLE source (
> col1 int,
> col2 varchar
> ) WITH (
> ...
> 'format' = 'json'
> )
> {code}
> With following data.
> {code:json}
> [
> {"col1": 1, "col2": "hello"},
> {"col1": 2, "col2": "flink"}
> ]
> {code}
> CC [~jark]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
