Along the same lines suggested by cbu, I found a very nice little snippet posted to the perl-unicode mail list, and included it in this reply to an older SoPW node.

I think you can see a nice way to adapt that little command-line tool into a module that uses perl's grep function in a suitable way to give you the sort of flexibility you want. In effect, a one-line idiom like this:

@names = grep /cyrillic/i, split /^/, do 'unicore/Name.pl';
[download]

is sort of what you're after, I think. Just put a variable into the regex for the grep, and you can fetch subsets based on all sorts of patterns in unicode character names. (But if you're going to do this sort of grep repeatedly in one script, you probably want to run unicore/Name.pl just once and keep all it's output in an array.)