generating a single row from multiple rows

[sravanthi]

Hi,

For a parent multiple childs can be there,i need to concatenate all child records in one column in a row.I have used 3 stage variables.

s1=s2,s2=i/p column,s3=if s1=s2 then s1:',':s3 else i/p column

If i have 3 child records,i'm getting output as
1 row:1 ,
2nd row:1,2
3rd row :1,2,3

But i need only one row with 1,2,3 as my output

can anyone suggest some solution for me.
Thanks!

[DSguru2B]

Load that into a temp table, group by the key and get the last one. If you were on server, you could have achieved this by aggregator. I am not sure if the Last option is available in a parallel aggreagator.
Or, if the remove duplicate stage works like a hashed file stage, then pass it through the remove duplicate stage. This way you will retain the last record.

[kumar_s]

Not LAST but Maximum option it left out in aggregator.
Or simply SORT it based on the key and mark Allow Duplicate = False.

[DSguru2B]

O great. Then the Last option is going to work. Ive done something like that in the past and the last option works great.

[sud]

O great. Then the Last option is going to work. Ive done something like that in the past and the last option works great.

Simply using a filter(constraint) on the number of commas will serve the purpose.

[sravanthi]

Hi all,

Thanks for your suggestion.I have used remove duplicate stage and used retain last column it worked fine.

thanks! :D

[DSguru2B]

Simply using a filter(constraint) on the number of commas will serve the purpose.

Only if the number is known before hand

[kumar_s]

Hi all,

Thanks for your suggestion.I have used remove duplicate stage and used retain last column it worked fine.

thanks! :D

Make sure your data is sorted and partitioned on the key before using RemoveDuplicate stage.