srcgraph-metrics

Graph-theoretic code-analysis metrics over the srcgraph-core traits — SCC, LCOM4, betweenness centrality, instability, Halstead, cyclomatic, entropy, clone detection, association rules, process mining.

Algorithms are generic over petgraph::Graph<N, E> where N: ClassNode, E: EdgeKind, so you can run them against an OwnedGraph loaded from GraphML or against a live in-process graph (e.g. yah's kg-store::Store::class_graph()).

use srcgraph_metrics::{betweenness::compute_betweenness, scc::compute_scc};

let graph = srcgraph_core::OwnedGraph::from_graphml_file("graph.graphml")?;
let scc   = compute_scc(&graph);
let bet   = compute_betweenness(&graph);

Benchmarks against networkx on a 7000+ node graph land in the 50× range (cargo bench -p srcgraph-metrics --bench betweenness).

License: MIT OR Apache-2.0.