Originally posted by: LaurentB
Hi Jignesh,
I usually treat this using a 3 steps scenario.
In an output card #1, which goes to Sink, I use an ad-hoc type tree which handles :
- a copy of an input record (usually in a blob, except if I need further processing in step 3)
- a new output record, in which I map the input record format to the output record format, including the LOOKUP. If the LOOKUP fails, I like to write a "dummy" keyword in place of the expected value (let's say something like ####, that is easily recognizable and cannot be part of the data)
- a status field, which gets "OK" if LOOKUP succeeded, and "KO" otherwise (so KO if field was maped to ####)
In output card #2, which is the actual output, it is quite easy to EXTRACT all the records from card #1 which have OK as the status. Since card #2 uses the same type tree and type as the output record of card #1, no functional map is needed there
In output card #3, using whatever format is needed (if possible a set of blobs), you then can EXTRACT the records which have KO as the status.
Hope this helps
Laurent
#IBM-Websphere-Transformation-Extender#DataExchange#IBMSterlingTransformationExtender