Logic for populating internal table

[Former Member]

Hi Abapers,

I have two internal tables it_bseg and it_bseg_ven with header lines. the key fields common to these two tables are bukrs, belnr and gjahr. My requirement is to move records of it_bseg_ven to it_bseg which are not there in it_bseg by comparing key fields.

I coded this way. but 'ne' is not supported in read. how can i achieve this. because no.of records are more.

LOOP AT it_bseg.

READ TABLE it_bseg_ven WITH KEY bukrs ne it_bseg-bukrs

belnr ne it_bseg-belnr

gjahr ne it_bseg-gjahr.

IF sy-subrc EQ 0.

MOVE it_bseg_ven-bukrs TO it_bseg-bukrs.

MOVE it_bseg_ven-belnr TO it_bseg-belnr.

MOVE it_bseg_ven-gjahr TO it_bseg-gjahr.

APPEND it_bseg.

ENDIF.

ENDLOOP.

SORT it_bseg BY bukrs belnr gjahr.

pointa are sure.

Message was edited by:

Premraj Cheguri

[Former Member]

Hi Premraj,

You loop at it_bseg_ven and read table it_bseg with the keys u mentioned.

If record not found in it_bseg (SY_SUBRC <> 0); append that record in it_bseg.

Hope it wil solve your problem

Regards,

Pankaj Sharma

[Former Member]

Yeah i can do this but the problem is there could be 5000 records in it_bseg and it_bseg_ven. almost many records are common between these two tables. because of not doing correct archiving for the database table bseg some records are missing in it_bseg. so is there any way to move records which are not there with minimal execution time without effecting performance.

please help me ....

Regards,

Radhika.

[Former Member]

The TRANSPORTING NO FIELDS addition checks for existance, and doesn't have the overhead of moving the record to the work area. Much quicker

[Former Member]

Why don't you do it the other way around?

LOOP AT it_bseg_ven.

READ TABLE it_bseg TRANSPORTING NO FIELDS

WITH KEY bukrs = it_bseg_ven-bukrs

belnr = it_bseg_ven-belnr

gjahr = it_bseg_ven-gjahr.

IF sy-subrc NE 0.

MOVE it_bseg_ven-bukrs TO it_bseg-bukrs.

MOVE it_bseg_ven-belnr TO it_bseg-belnr.

MOVE it_bseg_ven-gjahr TO it_bseg-gjahr.

APPEND it_bseg.

ENDIF.

ENDLOOP.

[Clemenss]

Hi Premraj ,

since release 4.5 you can LOOP ASSIGNING <field-symbol>. This removes the overhead of moving every table body line to the header line completely. Especially for tables with that much fields as BSEG it will ibrease processing time 5 to 10 times.

You may do like this:

field-symbols:
  <bseg> like line of it_bseg[].
sort:
  it_bseg_ven by bukrs belnr jhahr.
loop at it_bseg assigning <bseg>. 
* check if record is already present
  read table it_bseg_ven transporting no fields
    with key 
    bukrs = <bseg>-bukrs
    belnr = <bseg>-belnr
    gjahr = <bseg>-gjahr
    binary search.
  check sy-subrc = 0.
* yes this record is already in it_bseg - don't need
  delete it_bseg_ven index sy-tabix.
endloop.
append lines of it_bseg_ven to it_bseg. 

I think this will be pretty fast - any better idea around?

Or like this:

append lines of it_bseg_ven to it_bseg.
sort it_bseg.
delete adjacent duplicates from it_bseg comparing bukrs belnr gjahr.

Just depend on if all other fields in the 2 internal tables are just the same.

Regards,

Clemens