Filter and Join on same table

Question

Hi All,
I am having a bit of hard time, implementing following.
All suggestions welcome.
(1) I have a file being mapped into an initial table with say, 10 fields (field1...field10).
(2) I want to execute following logic for mapping
For all records in the table, in a cursor
If field1 = 10 Then
update field10=x
Else If field2 = 20 Then
update field10=y
End if
If field10=x then
update field3 =222
End if
I was thinking of using the filter, but am grappling with the problem that after I define the filter, how do I merge it with the original table data and execute the last conditional update (based on field10) on all records in the table?
- Am thinking of doing as below
Join the output from mother table with output from filter into a temp table (using non equal row id as the join condition), but this creates duplicate column names, and am wondering how to collapse them back into one column set again?
is there an alternative possible, for this is very kludgy (if it works at all).
Question 2
(1) I have a sql expression defined, which I want to use in the filter bifurcation and also after the join. The sql expression's input column and output columns are same, just the target are to be different?
Is it possible to do this, or do I have to duplicate the sql expression?
Question 3
My current load/stage is in PL/SQL procedure, which I am trying to model with OWB. Is there a guideline/ recommended best practice for doing this kind of activity?
Appreciate your help.
Deepak

Answer 1

1. I think I already gave an answer on this question (and there is one more of the same in the forum), but here it is again:
You can use an expression with a case statement:
CASE field1
WHEN 10 THEN 'x'
WHEN 20 THEN 'y'
WHEN ... THEN...
ELSE field10 END CASE
The input to this expression are fields field1 and field10, the output goes to field 10. So field10 will be updated with the value coming out of the expression - if field1 is 10 then it will be updated with 'x', when field1 is 20 then it will be updated with 'y' etc... when none of the CASE conditions are true (the ELSE case) it will be updated with the field10 (passt-hrough).
2. The best solution would be to create a transformation (a function, for example) that contains your expression, then use it throughout the project without having to retype it.
3. You should:
- Import the source object structures and (where possible) the target ones
- Design the new objects in OWB
- Import your custom transformation library, if there is one
- Design the extraction processes as mappings in OWB (you will not be able to reuse much of your old code if you want to take advantage of OWBs metadata management, runtime management etc. and if you want to maintain the system through OWB)
- Run the two systems side by side for some time until you are confortable that the process logic you designed in OWB gives the same results as the old process.
- Move the OWB system to production and switch the old system off.
Regards:
Igor