OK, I finally got off my butt and checked out Language::Prolog::Yaswi. The docs need some work and it helps if you know Prolog, and if you use a 5.1 or 5.2 version of SWI-Prolog (compiled with thread support: ./configure --enable-mt) and a version of Perl compiled with ithreads. However, if you jump through all of those hoops, you can figure out who's a thief.
steals(PERP, STUFF) :-
thief(PERP),
valuable(STUFF),
owns(VICTIM,STUFF),
not(knows(PERP,VICTIM)).
thief(badguy).
valuable(gold).
valuable(rubies).
owns(merlyn,gold).
owns(ovid,rubies).
knows(badguy,merlyn).
In short, I finally admitted defeat. Implementing a predicate logic engine in pure Perl that is easy to use and extensible involves understanding streams, patterns, a pattern matcher and generating arbitrary streams from complex data structure and handling unification without the benefit of tail call recursion optimizations. In short, to fully enjoy the First Order Predicate Calculus in pure Perl involves writing an NFA regex engine that operates on data structures instead of strings.
I am not that smart.
I went ahead and used the SWI-Prolog version. And it works :)
#!/usr/local/bin/perl
use strict;
use warnings;
use Language::Prolog::Yaswi qw/:query :assert/;
use Language::Prolog::Types::overload;
use Language::Prolog::Sugar
functors => [qw/steals thief valuable owns knows/],
functors => { NOT => 'not' }, # bugfix: This was in chains
vars => [qw/PERP STUFF VICTIM/],
chains => {
AND => ',',
};
swi_assert(
steals(PERP, STUFF) =>
AND(
valuable(STUFF),
thief(PERP),
owns(VICTIM, STUFF),
NOT(knows(PERP, VICTIM)),
));
swi_facts(
thief('badguy'),
valuable('gold'),
valuable('rubies'),
owns(qw/merlyn gold/),
owns(qw/ovid rubies/),
knows(qw/badguy merlyn/),
);
swi_set_query(steals('badguy', STUFF));
while(swi_next) {
print swi_var(STUFF), $/;
}