pattern finding problem

On the BioPerl mailing list we often get requests like the following:

Within a given biosequence with length X, find substrings of min. 
length A and max. length B that contain the pattern P at least C times 
but no more than D times.

A more concrete example: Find all substrings 12 characters long (A = B 
= 12) that have at least 7 (C = 7, D = 12 implictly) 'I' or 'L' 
characters (P = [IL]) in it.

The naive approach is a "sliding window" method, but it seems to me 
that a pattern matching approach would be more efficient.  And it 
sounds like a great little challenge for the brilliant minds of FWP.  
The "best" version will find it's way into a BioPerl module (with 
appropriate attribution, of course).  Golfing is not the goal here (but 
Golf-ed solutions are still welcome, if you must).

Enjoy,

-Aaron