fuzzbunny

fuzzbunny

fuzzbunny is a small (1k), fast & memory efficient fuzzy string searching/matching/highlighting library. It works equally well in a browser environment or Node.js.

Why fuzzbunny?

• Human friendly - fuzzbunny scoring and algorithm is more tuned to "human" searching patterns. It surfaces what you're looking for with minimal keystrokes.
• Lightweight - ~3KB minified and has zero dependencies.
• Ultra fast - ~million lines/second on a 2.4Ghz virtual core.

Other similar libraries are fuzzymatch, fuzzy, fuzzy-search, fuzzyjs.

fuzzbunny aims to be nimble and fast. It has a simple api that can easily be integrated with any frontend library to build great search UI. We use it at mixpanel.com to power our UI dropdowns and tables.

Installation

npm install --save fuzzbunny or yarn add fuzzbunny

Demo

Fuzzbunny Gutenberg Catalog Demo →

Usage

const {fuzzyFilter, fuzzyMatch} = require(`fuzzbunny`);
// or import {fuzzyFilter, fuzzyMatch} from 'fuzzbunny';

const heroes = [
  {
    name: `Claire Bennet`,
    ability: `Rapid cellular regeneration`,
  },
  {
    name: `Micah Sanders`,
    ability: `Technopathy`,
  },
  {
    name: `Hiro Nakamura`,
    ability: `Space-time manipulation`,
  },
  {
    name: `Peter Petrelli`,
    ability: `Tactile power mimicry`,
  },
];

// Use fuzzyFilter to filter an array of items on specific fields and get filtered + score-sorted results with highlights.
const results = fuzzyFilter(heroes, `stm`, {fields: [`name`, `ability`]});
/*
results = [
  {
    item: {
      name: 'Peter Petrelli',
      ability: 'Tactile power mimicry',
    },
    score: 1786,
    highlights: {
      ability: ['', 'T', 'actile power ', 'm', 'imicry'],
    },
  },
  {
    item: {
      name: 'Hiro Nakamura',
      ability: 'Space-time manipulation',
    },
    score: 983,
    highlights: {
      ability: ['Space-', 't', 'ime ', 'm', 'anipulation'],
    },
  },
];
*/

// Use fuzzyMatch to match a single string to get score + highlights. Returns null if no match found.
const match = fuzzyMatch(heroes[0].name, `ben`);
/*
match = {
  score: 2893,
  highlights: ['Claire ', 'Ben', 'net'],
};
*/

Scoring and Sort order

fuzzbunny uses a scoring algorithm that prioritizes following signals. See _getMatchScore function.

Example 1:

• Start of string - {Mayfl}ower ranks above The {Mayfl}ower
• Closer to start - The {Mayfl}ower ranks above Story of the {Mayfl}ower
• Contiguous length - The {Mayfl}ower ranks above {May} {fl}ower
• Alphabetically - The {May} {fl}ower ranks above This {May} {fl}ower

Example 2:

const f = require(`fuzzbunny`);
f.fuzzyMatch(`Gobbling pupusas`, `usa`);
// {score: 2700, highlights: ['Gobbling pup', 'usa', 's']}
f.fuzzyMatch(`United Sheets of Antarctica`, `usa`);
// {score: 2276, highlights: ['', 'U', 'nited ', 'S', 'heets of ', 'A', 'ntarctica']}

Gobbling pup{usa}s wins because 3 letter contiguous sequence yields a higher score.

NOTE: fuzzbunny optmizes for meaningful results. It only does substring/prefix/acronym-matching, not greedy matching.

This is because humans brains are great at prefix recall. e.g words that start with "ca" are much easier to recall than words that contain the letters "c" and "a" somewhere. It's easy to remember that {usa} stands for {U}nited {S}tates of {A}merica, not F{u}ll Java{s}cript Fr{a}mework

Performance

fuzzbunny matches ~ million lines/second on modern hardware. Tested on 2018 MacBook Pro with 2.4Ghz CPU. See tests/performance.js

Types

fuzzbunny comes with autogenerated TypeScript types. See index.d.ts