ABAP assistance - start routine logic in update rule

Question

I have used an existing update rule and have based my logic around the same.  The purpose of the rule is to look up customer master data and get a subset of customer numbers from the transaction records so that the values for customer number from the transactional data will not be updated if it does not match with existing master data customer numbers.
The loads are full and we drop the data before we load.
I have listed the logic below (the number at the front is to be considered as the line number) and a list of open questions that I have thereafter:
Start routine logic:
1  DATA: l_index LIKE sy-tabix.
2  DATA: BEGIN OF ls_customer,
3        customer TYPE /BI0/OICUSTOMER,
4        objver TYPE RSOBJVERS,
5     END OF ls_customer,
6      lt_customer LIKE TABLE OF ls_customer.
7  REFRESH: lt_customer.
8  LOOP AT DATA_PACKAGE.
all customers from data package
9    ls_customer-custno = DATA_PACKAGE-custid.
10  ls_customer-objver = 'A'
11    APPEND ls_customer TO lt_customer.
12   ENDLOOP.
12  SORT lt_customer.
13  DELETE ADJACENT DUPLICATES FROM lt_customer.
14   IF NOT lt_customer[] IS INITIAL.
15    SELECT /BI0/OICUSTOMER RSOBJVERS
16      FROM /BI0/PCUSTOMER
17      INTO CORRESPONDING FIELDS OF TABLE lt_customer
18      FOR ALL ENTRIES IN lt_customer
19      WHERE ls_customer-custno = DATA_PACKAGE-custid
20      AND ls_customer-objver = 'A'
21    SORT lt_customer BY customer ASCENDING                    
22  ENDIF.
Questions
Line
1 - what is the purpose of this line? What is it that is being declared
2 - in some code I have seen this line with OCCURS 0 at the end what does this mean with and without the term?
4 - I am using the Data Element name is this correct or should I use the field name?
3 - 5 here I declare an internal structure/table is that correct?
6 - here I declare a work area based on the internal table is that correct?
7 - What would happen if I avoided using the REFRESH statement?
8 - 12 - Is this syntactically correct, I am trying to get a set of data which is the customer numbers which match the master data customers and the master data record is án active version and than appendíng to the work area?
13 - My understanding is this will reduce the number of records in the work area is this correct and needed?
14 - 22 I am trying to identify my required set of data but feel I am repeating myself, could someone advise?
Finally what logic would I actually need to write in the key figure object, could I use something like:
Result = lt_customer.
Thanks
Edited by: Niten Shah on Jun 30, 2008 8:06 PM

Answer 1

1. This line is not required
2. OCCURS 0 is the OLD way of defining an internal table with that structure.  As it is, it just defines a flat structure.
3. Data element is usually best
3-5 Yes
6. No.  Here you are declaring a table of the type of the flat structure.  Just as the ABAP says!
7. Nothing.  But by putting this in, you ensure that you know the state of the table (empty) before you start looping through the data package
8-12. You can tell if it is syntactically correct by pressing Ctrl-F2 when in the editor.  Looks ok.
13. Ensures your list of customers contains no duplicated.  The code up to this point is building a list of all the unique customers in the data package.
14-22. Goes to the database and brings back ONLY those customers which are found in the master data.  Looks ok.
This is a start routine (that's why you've got a data package).  You don't use result.  You should update the datapackage.  But this you haven't done.  Double click on the table name /BIC/PCUSTOMER to get the correct field names.
So you have to loop through the data package again, and check if the customer in the datapackage is lt_customer.  If it is, fine, otherwise you blank it and report an error, or set an error message or whatever.
I wouldn't do it like this.  I'd do something like this:
STATICS: st_customer TYPE HASHED TABLE OF TYPE /bi0/oicustomer
                              WITH UNIQUE KEY TABLE_LINE.
* st_customer retains its value between calls, so only populate if empty
* In one run of the infopackage, this will mean you do only one read of
* the master data, so very efficient.
IF st_customer IS INITIAL.
  SELECT customer FROM /BI0/PCUSTOMER
                          INTO TABLE st_customer
                          WHERE objvers EQ 'A'. " Only active values
ENDIF.
* Go through data package
LOOP AT DATA_PACKAGE.
* Check whether the customer exists.
  READ TABLE st_customer TRANSPORTING NO FIELDS
              WITH TABLE KEY table_line = DATA_PACKAGE-custid.
  CHECK sy-subrc IS NOT INITIAL.
* If you get here, the customer isn't valid.  So I'm just setting it blank
  CLEAR DATA_PACKAGE-custid.
  MODIFY DATA_PACKAGE. " Updates the datapackage record
ENDLOOP.
Even this is not fully optimised, but it's not bad.
I strongly suggest that you get yourself sent on the basic ABAP programming course if you're going to do a lot of this.  Otherwise, read the ABAP documentation in the help.sap.com, and, from the editor, get the cursor on each ABAP keyword and press F1 to read the ABAP help.
matt