@umanghome/fuzzysort

Fuzzysort

A fuzzy-searching library for JavaScript

This is a fork of farzher/fuzzysort with some significant changes:

• Tree-shakable
• Only synchronous searching
• No global cache
• Only works on objects

The original repository describes this as

Fast SublimeText-like fuzzy search for JavaScript. Sublime's fuzzy search is... sublime. I wish everything used it. So here's an open source js version.

Demo at https://umanghome.github.io/fuzzysort/demo.html

Installation

Node

npm install @umanghome/fuzzysort

and use as

const fuzzysort = require('@umanghome/fuzzysort');

or

import * as fuzzysort from '@umanghome/fuzzysort';

Browser

Download and serve dist/fuzzysort.umd.js

<script src="fuzzysort.umd.js"></script>

The library will be available under window.fuzzysort

Exports

• search - The core function you will use for searching.
• createCache - For better performance, you should pass a cache to search. This function creates the necessary cache.
• algorithmWithTypo - You will need to pass an algorithm to search. This algorithm allows a mismatch of one character.
• algorithmWithoutTypo - You will need to pass an algorithm to search. This algorithm does not allow a mismatch.

Usage

createCache

const cache = createCache();

A cache can be reused between searches for to gain performance improvement. The recommended way to use a cache using a single cache for different searches across the same targets. If the targets change entirely, use a different cache. This can translated loosely to mean use the same cache for multiple terms across the same targets, but a different cache for each target.

It's a good idea to free up memory when you know no further searches will be made using the cache. To clear all the internal caches, use

const cache = fuzzysort.createCache();

// Do stuff

cache.clear();

The unmount lifecycle hook of your UI component is a good place to use this.

search

search(
  term: string, // The search term
  targets: Array<Object>, // The list of objects to search on
  keys: Array<string>, // The keys on each object to consider
  options: Object // Misc. options (see below)
): ({
  results: Array<Object>,
  total: number 
})

options = {
  algorithm: fuzzysort.algorithmWithTypo, // The algorithm to use

  // Optional
  cache: createdCache, // The cache that is created. See `createCache` usage for details.
  limit: 10, // The limit of results to return. Picks the top `limit` results. Default: 9007199254740991
  threshold: -100, // Considers results with a score greater than or equal to `threshold`. Default: -9007199254740991
}

keys is the list of keys to search on. Nested keys can be represented as "foo.bar". Example: ["name", "contact.phone"] if your target looks like

{
  name: 'John Doe',
  contact: {
    phone: '9988776655'
  }
}

search returns an object with two keys:

• results - An array of objects of the shape

{
  ref: Object; // Reference to the original object in `targets`. Equality check will work since this is an object ref.
  score: number; // The score of the match
}

• total - The total number of matches. This might be different than results.length if options.limit is used.

• meta - An object containing meta-information for all the matches. It is an object with keys as every key of keys. See usage for an example. Each object under meta[key] looks like

{
  indices: Array<number> | null; // The indices matched, if at all
  score: number | null; // The score if we found any matching characters
  target: string; // The string that we performed the search on
}

Example

const Banks = [
  {
    "code": "HDFC",
    "name": "HDFC Bank"
  },
  {
    "code": "ICIC",
    "name": "ICICI Bank"
  },
  {
    "code": "IOBA",
    "name": "Indian Overseas Bank"
  },
  {
    "code": "SBIN",
    "name": "State Bank of India"
  },
  {
    "code": "UBIN",
    "name": "Union Bank of India"
  },
];

const cache = fuzzysort.createCache();

const results = fuzzysort.search('india', Banks, ['name'], {
  algorithm: fuzzysort.algorithmWithTypo,
  cache: cache,
  limit: 2,
});

cache.clear();

console.log(results);

will log

{
  "results": [
    {
      "ref": {
        "code": "IOBA",
        "name": "Indian Overseas Bank"
      },
      "meta": {
        "name": {
          "indices": [0, 1, 2, 3, 4],
          "score": -15,
          "target": "Indian Overseas Bank"
        }
      },
      "score": -15
    },
    {
      "ref": {
        "code": "SBIN",
        "name": "State Bank of India"
      },
      "meta": {
        "name": {
          "indices": [14, 15, 16, 17, 18],
          "score": -28,
          "target": "State Bank of India"
        }
      },
      "score": -28
    }
  ],
  "total": 3
}

TODO

• Document how to highlight
• Allow and document custom scoreFn
• Add tests for algorithmWithTypo
• Add tests for algorithmWithoutTypo
• Modernize algorithmWithTypo
• Modernize algorithmWithoutTypo
• Fix and add types
• Maybe migrate to TypeScript

License

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

Changelog

0.0.1

• Initial release