Skip to content

Latest commit

 

History

History
100 lines (64 loc) · 2.78 KB

README.md

File metadata and controls

100 lines (64 loc) · 2.78 KB

Graham scan

Build Status codecov

A JavaScript implementation of the Graham scan algorithm for finding the convex hull of a set of points.

Installation

npm i @lucio/graham-scan

Usage

const grahamScan = new GrahamScan();
grahamScan.setPoints([[1,0], [0,1], [2,1], [1,0.5]]);
const hull = grahamScan.getHull();  // [1,0], [2,1], [0,1]

API

new GrahamScan()

Creates a new instance of the algorithm, meant to be reusable.

const grahamScan = new GrahamScan();

grahamScan.addPoint(point)

Adds a point to the set. The point must be an array of two numbers: the x and y coordinates.

const grahamScan = new GrahamScan();
grahamScan.addPoint([10, 20]);

grahamScan.clear()

Clear the underlying array of points.

const grahamScan = new GrahamScan();
grahamScan.addPoint([10, 20]);
grahamScan.clear();

grahamScan.getHull()

Computes the convex hull. Returns a new array containing all the hull vertices, starting from the one with the lowest y value (in case there are multiple ones, it will be the one with the lowest x as well) and going in counter-clockwise direction.

const grahamScan = new GrahamScan();
grahamScan.setPoints([[1,0], [0,1], [2,1], [1,0.5]]);
const hull = grahamScan.getHull();  // [1,0], [2,1], [0,1]

grahamScan.getPoints()

Returns a reference to the underlying array containing all the points.

const grahamScan = new GrahamScan();
grahamScan.addPoint([10, 20]);
grahamScan.addPoint([30, 40]);
grahamScan.getPoints();  // [10, 20], [30, 40]

grahamScan.setPoints(points)

Replace the underlying array with the given one.

const grahamScan = new GrahamScan();
grahamScan.setPoints([[10, 20], [30, 40]]);

Contributing

git clone [email protected]:luciopaiva/graham-scan.git
cd graham-scan
nvm install
npm install

Tests

npm test

FAQ

Is this the fastest way to build a convex hull?

No. Graham scan runs in O(n log n), where n is the total number of points in the set. Chan's algorithm does a bit better, O(n log h), where h is the number of vertices composing the final hull.

It is still much better than Jarvis march, though, which runs in O(nh).

Why use the .mjs extension?

Node.js projects that use this library by directly copying the script file instead of installing it via npm would need to add type: module to package.json, which may be undesired.