V,

Perhaps ironically but for the greater good, NiFi (IMHO, with record
processors) performs better when you don't touch the content but just
describe what you want from it. So in your case you could send the
content to two different branches, one could be a PutDatabaseRecord
with a reader schema that ignores the "mylist" element, and the other
could use Jolt / SplitRecord / ForkRecord to "hoist" the fields from
the "mylist" element into a root array/RecordSet, then use
PutDatabaseRecord to process it. Then you don't need a "split" or "get
into attribute" or any other operation that isn't in line with your
business logic.

Regards,
Matt

On Fri, Aug 31, 2018 at 7:34 PM l vic <[email protected]> wrote:
>
> Hi,
> I have a json record that contains array "mylst":
> {
>     "id": 0,
>     "name": "Root",
>      "mylist": [{
>       "id": 10,
>       "info": "2am-3am"
>     },
>     {
>         "id": 11,
>         "info": "3AM-4AM"
>     },
>     {
>         "id": 12,
> "info": "4am-5am"
>     }]
>
> }
> I have to save root data into one db table and array into another... Can 
> someone recommend an approach to "splitting" of record for 2 different 
> database writers?
> Thank you,
> V

Reply via email to