ekzhang / crepe

Datalog compiler embedded in Rust as a procedural macro
Apache License 2.0
460 stars 16 forks source link
datalog logic-programming rust souffle static-analysis translator

Crepe

github crates.io docs.rs build status

Crepe is a library that allows you to write declarative logic programs in Rust, with a Datalog-like syntax. It provides a procedural macro that generates efficient, safe code and interoperates seamlessly with Rust programs.

Features

Example

The program below computes the transitive closure of a directed graph. Note the use of the crepe! macro.

use crepe::crepe;

crepe! {
    @input
    struct Edge(i32, i32);

    @output
    struct Reachable(i32, i32);

    Reachable(x, y) <- Edge(x, y);
    Reachable(x, z) <- Edge(x, y), Reachable(y, z);
}

fn main() {
    let mut runtime = Crepe::new();
    runtime.extend([Edge(1, 2), Edge(2, 3), Edge(3, 4), Edge(2, 5)]);

    let (reachable,) = runtime.run();
    for Reachable(x, y) in reachable {
        println!("node {} can reach node {}", x, y);
    }
}

Output:

node 1 can reach node 2
node 1 can reach node 3
node 1 can reach node 4
node 1 can reach node 5
node 2 can reach node 3
node 2 can reach node 4
node 2 can reach node 5
node 3 can reach node 4

You can do much more with Crepe. The next example shows how you can use stratified negation, Rust expression syntax, and semi-naive evaluation to find all paths in a weighted graph with length at most MAX_PATH_LEN.

use crepe::crepe;

const MAX_PATH_LEN: u32 = 20;

crepe! {
    @input
    struct Edge(i32, i32, u32);

    @output
    struct Walk(i32, i32, u32);

    @output
    struct NoWalk(i32, i32);

    struct Node(i32);

    Node(x) <- Edge(x, _, _);
    Node(x) <- Edge(_, x, _);

    Walk(x, x, 0) <- Node(x);
    Walk(x, z, len1 + len2) <-
        Edge(x, y, len1),
        Walk(y, z, len2),
        (len1 + len2 <= MAX_PATH_LEN);

    NoWalk(x, y) <- Node(x), Node(y), !Walk(x, y, _);
}

fn main() {
    let n = 256;
    let mut edges = Vec::new();
    for i in 0..n {
        for j in 0..n {
            if rand::random::<f32>() < 0.02 {
                edges.push(Edge(i, j, 5));
            }
        }
    }

    let mut runtime = Crepe::new();
    runtime.extend(edges);
    let (walk, nowalk) = runtime.run();
    println!("Walk: {}", walk.len());
    println!("NoWalk: {}", nowalk.len());
}

Output:

Walk: 89203
NoWalk: 8207

Notes

From initial testing, the generated code is very fast. Variants of transitive closure for large graphs (~106 relations) run at comparable speed to compiled Souffle, and use a fraction of the compilation time.

For benchmarks, see the benches/ directory. The benchmarks can be run using cargo bench.

This macro generates a Crepe struct in the current module, as well as structs for all of the declared relations. This means that to integrate Crepe inside a larger program, you should put it in its own module with related code. See the documentation for more information.

Acknowledgements

This project was heavily inspired by Souffle and Formulog, which both use similar models of Datalog compilation for static analysis.