My understanding from the documentation for PPI is that is analyses Perl Documents in isolation so I suspect there are simple cases where a regex may be provided by a module and used in the Perl Code being analysed that may not be recognised by PPI. However I don't have PPI available (it doesn't seem to be in ActiveState's ppm repositories) so I can't test that.

Note that PPI doesn't claim to parse Perl Code, 'only' Perl Documents. I agree PPI very likely suffices for the OP's purpose, but it's not clear that PPI invalidates the mantra. :)

Right. From the documentation for PPI:

Since that time I've been able to prove to my own satisfaction that it is truly impossible to accurately parse Perl as both code and document at once. For the academics, parsing Perl suffers from the "Halting Problem".