ruby-prolog allows you to solve complex logic problems on the fly using a dynamic, Prolog-like DSL inline with your normal Ruby code. Basic use is encompassed by stating basic facts using your data, defining rules, and then asking questions. Why is this cool? Because ruby-prolog allows you to leave your normal object-oriented vortex on demand and step into the alternate reality of declarative languages.
ruby-prolog has been used in projects ranging from complex realtime access control authorization in Rails apps, to headless 3D layout engines, and many other use cases.
- Example: Dynamic Access Control for IAM
- Example: Towers of Hanoi solution
With ruby-prolog:
- There are no classes.
- There are no functions.
- There are no variables.
- There are no control flow statements.
You can use all these wonder things -- it’s still Ruby after all -- but they’re not needed, and mainly useful for getting data and results into/out of the interpreter. Declarative langugaes like Prolog are often favored heavily in artificial intelligence and theorem proving applications and is also taught in computer science curricula, so I hope this updated release proves useful for your logic evaluation needs!
ruby-prolog is written using object-oriented-ish pure Ruby, and should work under all modern Ruby interpreters. Please report compatibility problems. The core engine is largely based on tiny_prolog, though numerous additional enhancements have been made such as object-oriented refactorings and integration of ideas from the interwebs. Unfortunately I cannot read Japanese and cannot give proper attribution to the original tiny_prolog author. (If you can, let me know and I'll update this document!)
Say you want to write the following Prolog code:
implication(a, b).
implication(b, c).
implication(c, d).
implication(c, x).
implies(A, B) :- implication(A, B).
implies(A, B) :- implication(A, Something), implies(Something, B).
Here's the equivalent Ruby code using this library:
db = RubyProlog.new do
implication['a', 'b'].fact
implication['b', 'c'].fact
implication['c', 'd'].fact
implication['c', 'x'].fact
implies[:A, :B] << implication[:A, :B]
implies[:A, :B] << [
implication[:A, :Something],
implies[:Something, :B]
]
end
Now you can run some queries:
# What are all the direct implications of 'c'?
db.query{ implication['c', :X] }
#=> [{ X: 'd' }, { X: 'x' }]
# What are all the things that can directly imply?
db.query{ implication[:X, :_] }
#=> [{ X: 'a' }, { X: 'b' }, { X: 'c' }, { X: 'c' }]
# What are all the things 'a' implies?
db.query{ implies['a', :X] }
#=> [{ X: 'b' }, { X: 'c' }, { X: 'd' }, { X: 'x' }]
Unfortunately if you have two predicates in a query, you can't just use a comma. There two ways to solve this problem:
# Solution 1: Use an array
db.query{[ implication['b', :S], implies[:S, :B] ]}
# Solution 2: Use a beneign assignment
db.query{_= implication['b', :S], implies[:S, :B] }
If you need to add to your database, you can call instance_eval
:
db = RubyProlog.new do
implication['a', 'b'].fact
implication['b', 'c'].fact
end
# Later...
db.instance_eval do
implication['c', 'd'].fact
implication['c', 'x'].fact
end
This will mutate your database. If you want to "fork" your database instead, you can call db.clone
, which will return a new instance with all stored data. Cloning like this is optimized to copy as little as possible.
If you're loading rules from a database, you might be generating predicates like this:
rules = Ruleset.find_by(org: ...).rules
db = RubyProlog.new do
rules['permissions'].map do |role, perm|
permission[role, perm].fact
end
end
However, if something doesn't work, how do find out why?
This is where #to_prolog
comes in handy. Just run it on your instance:
puts db.to_prolog
and you'll get something that looks like this:
permission('admin', 'invite').
permission('admin', 'ban').
permission('membe', 'create_post').
Then you can do a quick copy/paste into an environment like Tau Prolog's sandbox or SWISH and run some queries.
gem install ruby-prolog
Two runnable examples are included in the 'bin' directory. The first..
ruby-prolog-acls
..shows the ruby-prolog dynamic DSL used to trivially implement access control checks. The second..
ruby-prolog-hanoi
..is a ruby-prolog solution to the well-known "Towers of Hanoi" problem in computer science. It's not clear, but something Prolog hackers will be interested in. If you have other useful or clever examples, please send a pull request!
See the test/ directory for additional examples.
- Pure Ruby.
- No wacko dependencies.
- Tested with Ruby 2.0.0!
- Object-oriented.
- Multiple Prolog environments can be created and manipulated simultaneously.
- Concurrent access to different core instances should be safe.
- Concurrent access to a single core instance might probably explode in odd ways.
$ git clone https://github.com/preston/ruby-prolog
$ cd ruby-prolog
$ bundle
$ rake test
Released under the Apache 2.0 license. Copyright (c) 2013-2022 Preston Lee. All rights reserved. https://prestonlee.com