I think you just need a counter ...
You create a 2nd itab for the file creation (collector).
You loop over the main itab and based on your rule and the sy-tabix you fill the 2nd itab, write it to the file (work station or server), refresh collector itab and continue the loop.
Write process, file creation depends on used technology.
Example solution:
****************
c_constant type i value '5'.
split type i.
v_count type i.
v_file_count.
describe itab1 lines to v_count.
loop at itab1 into wa.
split = split + 1.
move wa into wa_collector.
append wa_collector to itab2.
If split EQ c_constant
or sy-tabix EQ v_count.
v_file_count = v_file_count + 1.
< open file / export however you already planed to do it>
< Filename will be dynamic on counter and other variable choice>
clear split.
refresh itab2.
endif.
enloop.
****************
But that was a very rough an simple one.
When I would write it into the program I would change the fixed parts. And you need to check from the example if the check agains the counted entries is sufficient to write the last two entries in the thrid file.
That can be done more dynamic, based on input fields, arithmetic rules, counting of entries etc. check the mod command ect. I have done things like that often with other purpose.
BR,
P.S.: The more I check the example above, the more I think you should take it really only as an sample idea. I would indeed make it nicer with less counters etc. and some assign ...
Bluesky
