Guess‑Rater

Fuzzy string matching for JavaScript. Handles typos, word order, accents, and punctuation — with five algorithms, a normalization pipeline, and full TypeScript support.

Install

npm install guess-rater

ESM only. Requires "type": "module" in your package.json.

Usage

import { rate, isMatch, findBestMatch, filterMatches, extract, createMatcher } from 'guess-rater'

rate('Molière', 'moliere')              // 100
rate('levenshtein', 'levenstein')       // ~88
rate('John Smith', 'Smith John', { algorithm: 'tokenSort' }) // 100

isMatch('Saint-Nazaire', 'saint nazaire', 85) // true

findBestMatch('peugeot', ['Renault Clio', 'Peugeot 208', 'BMW'])
// { value: 'Peugeot 208', score: 82, index: 1 }

filterMatches('iPhone', ['iPhone 15', 'Samsung S24', 'iPhone 14'], { threshold: 70 })
// ['iPhone 15', 'iPhone 14']

extract('apple', ['Apple Watch', 'Apple TV', 'Samsung TV'], { limit: 2 })
// ['Apple Watch', 'Apple TV']

Reuse a configuration with createMatcher():

const matcher = createMatcher({
  algorithm: 'hybrid',
  normalize: { removeAccents: true, sortTokens: true }
})

matcher.rate('Jean-Paul', 'jean paul')       // 100
matcher.isMatch('Café de Paris', 'cafe de paris') // true
matcher.filterMatches('iphone', catalog)

Documentation

tfoucher5.github.io/guess-rater

• Quickstart
• Algorithms — levenshtein, jaroWinkler, tokenSort, tokenSet, hybrid
• Normalization — accents, casing, stop words, replacements
• List helpers — rankCandidates, findBestMatch, filterMatches, extract
• API reference

License

MIT