Vanillalog

A Datalog variant that is as vanilla as it gets.

When used as a library, it provides an abstract AST acting as an intermediate target for Datalog variants with interesting operators. For example, see Temporalog for using Computation Tree Logic (CTL) operators in Datalog formulae.

This frontend (or intermediate representation) compiles to Exalog for evaluation and optimisation.

Installation

Clone the repository, run stack build, run stack exec -- vanillalog, let me know if it fails.

If you do stack install, you shouldn't need the prefix stack exec --. From there on, I'll omit it.

Usage

Running vanillalog lists all the options, but in summary, one can run or inspect (pp) a datalog program. There is a repl as well.

The most common usage to evaluate a program and see the results to queries in the source is:

vanillalog run -f <filepath>

The repl command takes an optional path to Vanillalog program which it evaluates and make available for queries.

vanillalog repl -f <filepath>

You can use :q, :quit, :e, :exit, or Ctrl+D to exit it.

The REPL doesn't yet allow extending existing predicates or defining recursive queries. For this reason omitting -f makes REPL useless unles you use foreign predicates.

Syntax

Best way to get a feel is to look at the examples. Definitely let me know if any of them causes problems!

A program is a list of clauses, facts, and queries.

A clause is of the form head :- body. (don't omit the full stop!). Example clause:

ancestor(X,Y) :- ancestor(X,Z), parent(Y,Z).

A fact is of the form head. (again, don't omit the full stop!). Example fact:

parent("Dad","Son").

A query is of the form ?- body. (you know the deal). Example query:

?- ancestor(Ancestor,"Son").

Head

A clause head is an atomic formula.

An atomic formula is of the form pred([Term]...). It consits of a predicate name and zero or more terms. Predicate names starts with lowercase letters, thereafter they can contain upper or lowercase alphanumeric characters as well as underscore (_) and a tick (').

A term is a variable, a literal, or a wildcard.

Variables start with uppercase letters followed by alphanumeric characters _ and '.

There are three kinds of literals: bools, integers, texts.

Bools: true and false

Integers: exactly what you'd expect, e.g., 42, -24, 0.

Texts: quoted arbitrary characters with no means to escape, e.g., "Horn Clauses are nice!"

A wildcard starts with an underscore (_) and is followed by anything that a variable can be.

Body

A body is structured like expressions are in most languages. The binary operators are conjunction (,) and disjunction (;). The unary operator is negation (!). There are also parantheses for disambiguation. Here's the BNF syntax:

Body ::= Atomic Formula | Body , Body | Body ; Body | ! Body | ( Body )

Foreign predicates

There are a number of foreign predicates included for your convenience.

Arithmetic

Predicate name	Type	Mode	Description
add	Int x Int x Int	++?	Puts the sum of first two params. into the third.
subtract	Int x Int x Int	++?	Take a guess.
lt	Int x Int	++	Holds when first param. is less than the second.
lte	Int x Int	++	Take a guess.
gt	Int x Int	++	Take a guess.
gte	Int x Int	++	Take a guess.

Unification

Predicate name	Type	Mode	Description
unify_int	Int x Int	+?	Unifies the first parameter with the second.
unify_text	Text x Text	+?	Ditto.
unify_bool	Bool x Bool	+?	Ditto.

IO

Predicate name	Type	Mode	Description
read_csv1	Text x Text	+?	Reads a CSV file, from the path in the first parameter and places each row in the second parameter.
read_csv2	Text x Text x Text	+??	Take a guess.
read_csv3	Text x Text x Text x Text	+???	Take a guess.
read_csv4	Text x Text x Text x Text x Text	+????	Take a guess.
append_csv1	Text x Text	++	Appends to the CSV file the path of which stored in the first parameter and the column in the second parameter.
append_csv2	Text x Text x Text	+++	Take a guess.
append_csv3	Text x Text x Text x Text	++++	Take a guess.
append_csv4	Text x Text x Text x Text x Text	+++++	Take a guess.

Semantics

To be continued...