It seems very specific and feasible.
Lucifer Tyrant <[email protected]> 于2022年8月8日周一 14:38写道：
> By the way, the row data format information also should be defined in the
> source connector. The final source config as the following:
>
>
>
>
>
>
>
>
>
>
>
>
> *Source { Kafka { shema { fields { field1 = "STRING"
> field2 = "INT" field3 = "DECIMAL(30, 3)" }
> format = "json" } }}*
>
>
>
>
>
>
>
>
>
> *Source { Kafka { shema { fields { field1 = "STRING"
> field2 = "INT" field3 = "DECIMAL(30, 3)" }
> format = "text"*
>
>
>
> * delimiter = "," } }}*
>
> In most use cases, I think we can support *json* or *text*, if we assign
> text, we also assign a *delimiter* option in the config file.
>
>
> 范佳 <[email protected]> 于2022年8月8日周一 14:15写道：
>
> > Hi, I think it’s great for add new config option named schema, but how to
> > implement complex nested types？
> > PS：your picture can’t display
> >
> > > 2022年8月8日 11:05，Lucifer Tyrant <[email protected]> 写道：
> > >
> > > Hi seatunnel community,
> > >
> > > According to issue #2371 and #2299, supporting user-defined schema in
> > source connectors is a import feature for seatuunel.
> > >
> > > In this email I want to discuss with you how to implement it.
> > >
> > > In my option, I think we can add a new config option named as schema in
> > all source connector source configs and for each connector they can parse
> > their own schema in getProducedType method. The implement function of
> > parsing schema we can add it in module seatunnel-common. And I agree with
> > your advice @ashulin to conform schema type to the habit of SQL the same
> as
> > defined in code as the following image show:
> > >
> > >
> > >
> > > What do you think about this?Waiting for your reply!
> > >
> > > TyrantLucifer
> > >
> > >
> > >
> > >
> >
> >
>
--
Best Regards
------------
EricJoy2048
[email protected]
