fastest-levenshtein 🚀

Fastest JavaScript implementation of the Levenshtein distance algorithm - (up to 10x+ faster than closest competitor). This algorithm can be used to measure the similarity of two strings.

Note: if you need to take the edit distance of one string against many strings (maybe billions), I have created a GPU-implementation that will be atleast 1000x faster than this. Contact me, if you’re interested.

$ npm i fastest-levenshtein

Usage

Node

const {distance, closest} = require('fastest-levenshtein')

// Print levenshtein-distance between 'fast' and 'faster' 
console.log(distance('fast', 'faster'))
//=> 2

// Print string from array with lowest edit-distance to 'fast'
console.log(closest('fast', ['slow', 'faster', 'fastest']))
//=> 'faster'

Deno

import {distance, closest} from 'https://deno.land/x/fastest_levenshtein/mod.ts'

// Print levenshtein-distance between 'fast' and 'faster' 
console.log(distance('fast', 'faster'))
//=> 2

// Print string from array with lowest edit-distance to 'fast'
console.log(closest('fast', ['slow', 'faster', 'fastest']))
//=> 'faster'

Benchmark

I generated 500 pairs of strings with length N. I measured the ops/sec each library achieves to process all the given pairs. Higher is better. fastest-levenshtein is a lot faster in all cases.

Relative Performance

This image shows the relative performance between fastest-levenshtein and js-levenshtein (the 2nd fastest). fastest-levenshtein is always a lot faster. x-axis shows “times faster”.

License

This project is licensed under the MIT License - see the LICENSE.md file for details