You appear to have record oriented data and want to get a subset of info out of those records...or so it appears to me.. There is another discussion going on about the range operator (/abc/.../xyz/), see: http://www.perlmonks.org/?node_id=783119

This is a different approach than your splitting on \n\n. And can also process huge files as this is done line by line, record by record without requiring the whole file to be read into a single scalar by redefining the record separator.

I am still considering some of the issues in this thread myself, but I think you will find it interesting and probably useful.