@graphrs

igraph for JavaScript — comprehensive graph algorithms at native speed.

The graph algorithm library JavaScript never had. 400+ algorithms (community detection, centrality, layout, flow, isomorphism) at native speed via WebAssembly. Works in Browser and Node.js. Zero native dependencies. MIT licensed.

---

Why @graphrs?

Python has networkx (pure Python, slow) and igraph (C core, fast). JavaScript has graphology (pure JS, limited) and... nothing fast.

@graphrs fills the "igraph for JavaScript" gap — a comprehensive, high-performance graph analysis library powered by Rust/WASM.

Capability	graphology	cytoscape.js	@graphrs
Community detection	2	0	10+
Centrality measures	7	2	15+
Layout engines	3	ext	16
Network flow	0	0	Full
Isomorphism	0	0	VF2
10k nodes PageRank	~5-10s	N/A	~100ms

Packages

Package	Description
@graphrs/core	Graph data structure + WASM runtime
@graphrs/community	Community detection (Louvain, Leiden, Infomap, ...)
@graphrs/centrality	Centrality measures (PageRank, betweenness, ...)
@graphrs/path	Shortest paths & traversal (Dijkstra, BFS, DFS, ...)
@graphrs/layout	Graph layout algorithms (FR, KK, Sugiyama, ...)
@graphrs/generators	Graph generators (ER, Barabasi-Albert, ...)
@graphrs/io	Import/export (GraphML, GML, DOT, edgelist, Pajek)
@graphrs/operators	Graph transforms (union, simplify, reverse, ...)
@graphrs/flow	Network flow & connectivity
@graphrs/isomorphism	Structural matching (VF2, canonical, automorphism)
@graphrs/g6	AntV G6 5.x integration (layout, analysis, adapters)
@graphrs/react-flow	React Flow integration (hook, adapters)

Quick Start

npm install @graphrs/core @graphrs/community @graphrs/centrality

import { Graph } from '@graphrs/core';
import { louvain } from '@graphrs/community';
import { pagerank } from '@graphrs/centrality';

// Build a graph
const g = Graph.fromEdges([
  [0, 1], [1, 2], [2, 0],  // cluster 1
  [3, 4], [4, 5], [5, 3],  // cluster 2
  [2, 3],                   // bridge
]);

// Community detection
const communities = await louvain(g);
// { membership: [0,0,0,1,1,1], modularity: 0.357, clusters: 2 }

// Centrality analysis
const pr = await pagerank(g, { damping: 0.85 });
// { scores: [0.12, 0.15, 0.23, 0.18, 0.16, 0.16] }

Framework Integration

# AntV G6 — plug-and-play layout + analysis
npm install @graphrs/g6 @antv/g6

# React Flow — useGraphrsLayout hook
npm install @graphrs/react-flow @xyflow/react

// G6: one-line layout registration
import { createGraphrsLayout } from '@graphrs/g6';
new G6Graph({ layout: createGraphrsLayout({ algorithm: 'fruchterman-reingold' }) });

// React Flow: auto-layout hook
import { useGraphrsLayout } from '@graphrs/react-flow';
const { nodes, edges } = useGraphrsLayout(initialNodes, initialEdges);

Use Cases

• Social network analysis — community detection, influence propagation
• Knowledge graphs — path queries, centrality ranking
• Visualization — drop-in layout engine for React Flow / G6 / Cytoscape / D3
• Supply chain — flow optimization, bottleneck identification
• Fraud detection — subgraph isomorphism, motif census

Architecture

┌─────────────────────────────────────────────────┐
│              Your Application                    │
│  import { Graph } from "@graphrs/core"           │
└───────────────────┬─────────────────────────────┘
                    │  ESM imports (tree-shakable)
┌───────────────────▼─────────────────────────────┐
│         TypeScript Wrapper Layer (MIT)            │
│  Type-safe Graph class, async API, marshalling   │
└───────────────────┬─────────────────────────────┘
                    │  wasm-bindgen glue
┌───────────────────▼─────────────────────────────┐
│          WASM Binary Layer (GPL-2.0)             │
│  Compiled from rust-igraph, lazy-loaded          │
└─────────────────────────────────────────────────┘

Documentation

Full documentation with interactive playground: totoro-jam.github.io/graphrs

Development

pnpm install          # install dependencies
pnpm turbo build      # build all packages
pnpm turbo test       # run all tests
pnpm turbo typecheck  # type check
pnpm format           # format code

License

The TypeScript wrapper code is MIT licensed.

This package bundles a WebAssembly binary compiled from rust-igraph (GPL-2.0-or-later). The TypeScript wrapper code is original work and is not a derivative of the GPL code.

Contributing

See CONTRIBUTING.md for guidelines.