Fact Engine

How to install Elixir

You can install Elixir by following the official docs or via asdf.

How to use Fact Engine

1. Download the fact_engine zip file and unzip it
2. cd into it
3. If you are using asdf you'll need to run asdf install
4. Run iex -S mix to get to the interactive shell
5. Run FactEngine.State.start() (this starts the process that holds the persistent state)
6. Run FactEngine.run(read_file_path, write_file_path) where write_file_path is the file path to the file you want to process and write_file_path is the file path to the file you want the results written to. For example: read_file_path = "/Users/jonericcook/github/fact_engine/examples/4/in.txt" and write_file_path = "/Users/jonericcook/github/fact_engine/examples/4/out.txt"

Example Run

Check the contents of the file that's to be processed

% cat /Users/jonericcook/github/fact_engine/examples/4/in.txt
INPUT make_a_triple (3, 4, 5)
INPUT make_a_triple (5, 12, 13)
QUERY make_a_triple (X, 4, Y)
QUERY make_a_triple (X, X, Y)

Start the Elixir interactive shell and use FactEngine

% iex -S mix
iex(1)> FactEngine.State.start()
iex(2)> read_file_path = "/Users/jonericcook/github/fact_engine/examples/4/in.txt"
iex(3)> write_file_path = "/Users/jonericcook/github/fact_engine/examples/4/out.txt"
iex(4)> FactEngine.run(read_file_path, write_file_path)
iex(5)> FactEngine.State.get
%{"make_a_triple" => #MapSet<[["12", "13", "5"], ["3", "4", "5"]]>}

Check the contents of the file that contains the results

% cat /Users/jonericcook/github/fact_engine/examples/4/out.txt
---
X: 3, Y: 5
---
false

Comments about solution

In an effort to not have duplicates in state I chose to sort the arguments. I chose this because to me it wouldn't make sense to have two entries - one for INPUT are_friends (alex, sam) and one for INPUT are_friends (sam, alex).

It was unclear how to create the constraint between the number of arguments and a statement therefore I opted to not verify the number of arguments for a given statement. For example INPUT love (cat, dog) and INPUT love (cheese) would both be accepted. As I read this I think I could handle that scenario by having a MapSet for each statement and each length of arguments provided with said statement. For example:

%{
    "loves" => %{
        2 => #MapSet<[["garfield", "lasagna"]]>,
        3 => #MapSet<[["garfield", "lasagna", "grilled_cheese"]]>
    }
}

Where the key at the second depth is the length of the lists in the MapSet.