Regular expressions are the wrong tool for this job.

For Krugle, we actually parsed the various languages so it actually knew the differences between e.g. definitions, invocations, comments, strings, etc. It affords the capabilities to do all sorts of cool queries and analyses.