AWK seems to be a good choice (however I have not analyzed the example in details). It seems like you are going to greate a table with cells spanning across multiple rows/columns. Think twice if this is what you need. Probably it would be acceptable to have the same fields repeated in multiple rows instead of using rowspan. That would make your code much easier.
You might need to count the entries first and then make a second run to fill the output document. Easy way is to parse that file 2x with a different code - however this is not optimal (low performance, too much I/O). That might be the best option for you.
There are plenty of other ways to do this (including PERL with "split" which can be very slow).

If by any chance you experience problems with number of "columns" in AWM try changing the awk (mawk, nawk, gawk,... there is plenty of this stuff with different limitations).

Good luck