@tomhundley/grafene

An in-memory graph library for Node.js. Build directed or undirected graphs, walk them, find shortest paths, detect cycles, compute components, spanning trees, and PageRank.

Zero runtime dependencies. Node 18+.

Install

npm install @tomhundley/grafene

Quick start

import { createGraph, shortestPath, topologicalSort } from "@tomhundley/grafene";

const g = createGraph({ directed: true });
g.addNode("build");
g.addNode("test");
g.addNode("deploy");
g.addEdge("build", "test", { weight: 1 });
g.addEdge("test", "deploy", { weight: 2 });

topologicalSort(g); // ["build", "test", "deploy"]
shortestPath(g, "build", "deploy");
// { path: ["build", "test", "deploy"], cost: 3 }

Graph API

import { Graph } from "@tomhundley/grafene";

const g = new Graph({ directed: false });
g.addNode("a", { label: "Ada" });
g.addNode("b");
g.addEdge("a", "b", { weight: 4, data: { kind: "friend" } });

g.neighbors("a");
g.degree("a");
g.clone();
g.subgraph(["a", "b"]);
g.toJSON();
Graph.fromJSON(saved);

Algorithms

| Function | Purpose | | --- | --- | | bfs / dfs | Walk from a start node | | reachable | Nodes you can reach | | shortestPath / dijkstra | Weighted shortest path | | astar | Heuristic search | | hasCycle | Cycle detection | | topologicalSort | Order a DAG | | connectedComponents | Undirected groups | | stronglyConnectedComponents | Directed SCCs (Tarjan) | | isBipartite | Two-color check | | minimumSpanningTree | Kruskal MST | | pageRank / density / degrees | Metrics |

import { connectedComponents, minimumSpanningTree, pageRank } from "@tomhundley/grafene";

const net = createGraph({ directed: false });
net.addNode("a");
net.addNode("b");
net.addNode("c");
net.addEdge("a", "b", { weight: 1 });
net.addEdge("b", "c", { weight: 2 });
net.addEdge("a", "c", { weight: 9 });

minimumSpanningTree(net).cost; // 3

What it is for

Use Grafene when your data is a network: dependencies, maps, org charts, social links, routing. It is not a graph database server — it lives in one Node process.

License

MIT