Fuse

A fuzzy search library for PHP

This is a PHP port of the awesome Fuse.js project and aims to provide full API compatibility wherever possible.

Check out their demo and examples to get a good feel for what this library is capable of.

Latest compatible Fuse.js version: 6.6.2

Table of Contents:

• Installation
• Usage
• Options
• Methods
• Differences with Fuse.js
• Development

Installation

This package is available via Composer. To add it to your project, just run:

composer require loilo/fuse

Note that at least PHP 7.4 is needed to use Fuse.

Usage

Here's a simple usage example:

<?php
require_once 'vendor/autoload.php';

$list = [
    [
        'title' => "Old Man's War",
        'author' => 'John Scalzi',
    ],
    [
        'title' => 'The Lock Artist',
        'author' => 'Steve Hamilton',
    ],
    [
        'title' => 'HTML5',
        'author' => 'Remy Sharp',
    ],
    [
        'title' => 'Right Ho Jeeves',
        'author' => 'P.D Woodhouse',
    ],
];

$options = [
    'keys' => ['title', 'author'],
];

$fuse = new \Fuse\Fuse($list, $options);

$fuse->search('hamil');

This leads to the following results (where each result's item refers to the matched entry itself and refIndex provides the item's position in the original $list):

[
    [
        'item' => [
            'title' => 'The Lock Artist',
            'author' => 'Steve Hamilton',
        ],
        'refIndex' => 1,
    ],
    [
        'item' => [
            'title' => 'HTML5',
            'author' => 'Remy Sharp',
        ],
        'refIndex' => 2,
    ],
];

Options

Fuse has a lot of options to refine your search:

Basic Options

isCaseSensitive

• Type: bool
• Default: false

Indicates whether comparisons should be case sensitive.

includeScore

• Type: bool
• Default: false

Whether the score should be included in the result set. A score of 0 indicates a perfect match, while a score of 1 indicates a complete mismatch.

includeMatches

• Type: bool
• Default: false

Whether the matches should be included in the result set. When true, each record in the result set will include the indices of the matched characters. These can consequently be used for highlighting purposes.

minMatchCharLength

• Type: int
• Default: 1

Only the matches whose length exceeds this value will be returned. (For instance, if you want to ignore single character matches in the result, set it to 2).

shouldSort

• Type: bool
• Default: true

Whether to sort the result list, by score.

findAllMatches

• Type: bool
• Default: false

When true, the matching function will continue to the end of a search pattern even if a perfect match has already been located in the string.

keys

• Type: array
• Default: []

List of keys that will be searched. This supports nested paths, weighted search, searching in arrays of strings and objects.

Fuzzy Matching Options

location

• Type: int
• Default: 0

Determines approximately where in the text is the pattern expected to be found.

threshold

• Type: float
• Default: 0.6

At what point does the match algorithm give up. A threshold of 0.0 requires a perfect match (of both letters and location), a threshold of 1.0 would match anything.

distance

• Type: int
• Default: 100

Determines how close the match must be to the fuzzy location (specified by location). An exact letter match which is distance characters away from the fuzzy location would score as a complete mismatch. A distance of 0 requires the match be at the exact location specified. A distance of 1000 would require a perfect match to be within 800 characters of the location to be found using a threshold of 0.8.

ignoreLocation

• Type: bool
• Default: false

When true, search will ignore location and distance, so it won't matter where in the string the pattern appears.

Tip: The default options only search the first 60 characters. This should suffice if it is reasonably expected that the match is within this range. To modify this behavior, set the appropriate combination of location, threshold, distance (or ignoreLocation).

To better understand how these options work together, read about Fuse.js' Scoring Theory.

Advanced Options

useExtendedSearch

• Type: bool
• Default: false

When true, it enables the use of unix-like search commands. See example.

getFn

• Type: callable
• Default: source

The function to use to retrieve an object's value at the provided path. The default will also search nested paths.

sortFn

• Type: callable
• Default: source

The function to use to sort all the results. The default will sort by ascending relevance score, ascending index.

ignoreFieldNorm

• Type: bool
• Default: false

When true, the calculation for the relevance score (used for sorting) will ignore the field-length norm.

Tip: The only time it makes sense to set ignoreFieldNorm to true is when it does not matter how many terms there are, but only that the query term exists.

fieldNormWeight

• Type: float
• Default: 1

Determines how much the field-length norm affects scoring. A value of 0 is equivalent to ignoring the field-length norm. A value of 0.5 will greatly reduce the effect of field-length norm, while a value of 2.0 will greatly increase it.

Global Config

You can access and manipulate default values of all options above via the config method:

// Get an associative array of all options listed above
Fuse::config();

// Merge associative array of options into default config
Fuse::config(['shouldSort' => false]);

// Get single default option
Fuse::config('shouldSort');

// Set single default option
Fuse::config('shouldSort', false);

Methods

The following methods are available on each Fuse\Fuse instance:

search

Searches the entire collection of documents, and returns a list of search results.

public function search(mixed $pattern, ?array $options): array

The $pattern can be one of:

• String
• Path
• Extended query
• Logical query

The $options:

• limit (type: int): Denotes the max number of returned search results.

setCollection

Set/replace the entire collection of documents. If no index is provided, one will be generated.

public function setCollection(array $docs, ?\Fuse\Core\FuseIndex $index): void

Example:

$fruits = ['apple', 'orange'];
$fuse = new Fuse($fruits);

$fuse->setCollection(['banana', 'pear']);

add

Adds a doc to the collection and update the index accordingly.

public function add(mixed $doc): void

Example:

$fruits = ['apple', 'orange'];
$fuse = new Fuse($fruits);

$fuse->add('banana');

sizeof($fruits); // => 3

remove

Removes all documents from the list which the predicate returns truthy for, and returns an array of the removed docs. The predicate is invoked with two arguments: ($doc, $index).

public function remove(?callable $predicate): array

Example:

$fruits = ['apple', 'orange', 'banana', 'pear'];
$fuse = new Fuse($fruits);

$results = $fuse->remove(fn($doc) => $doc === 'banana' || $doc === 'pear');
sizeof($fuse->getCollection()); // => 2
$results; // => ['banana', 'pear']

removeAt

Removes the doc at the specified index.

public function removeAt(int $index): void

Example:

$fruits = ['apple', 'orange', 'banana', 'pear'];
$fuse = new Fuse($fruits);

$fuse->removeAt(1);

$fuse->getCollection(); // => ['apple', 'banana', 'pear']

getIndex

Returns the generated Fuse index.

public function getIndex(): \Fuse\Core\FuseIndex

Example:

$fruits = ['apple', 'orange', 'banana', 'pear'];
$fuse = new Fuse($fruits);

$fuse->getIndex()->size(); // => 4

Indexing

The following methods are available on each Fuse\Fuse instance:

Fuse::createIndex

Pre-generate the index from the list, and pass it directly into the Fuse instance. If the list is (considerably) large, it speeds up instantiation.

public static function createIndex(array $keys, array $docs, array $options = []): \Fuse\Core\FuseIndex

Example:

$list = [ ... ]; // See the example from the 'Usage' section
$options = [ 'keys' => [ 'title', 'author.firstName' ] ];

// Create the Fuse index
$myIndex = Fuse::createIndex($options['keys'], $list);

// Initialize Fuse with the index
$fuse = new Fuse($list, $options, $myIndex);

Fuse::parseIndex

Parses a JSON-serialized Fuse index.

public static function parseIndex(array $data, array $options = []): \Fuse\Core\FuseIndex

Example:

// (1) When the data is collected

$list = [ ... ]; // See the example from the 'Usage' section
$options = [ 'keys' => [ 'title', 'author.firstName' ] ];

// Create the Fuse index
$myIndex = Fuse::createIndex($options['keys'], $list);

// Serialize and save it
file_put_contents('fuse-index.json', json_encode($myIndex));


// (2) When the search is needed

// Load and deserialize index to an array
$fuseIndex = json_decode(file_get_contents('fuse-index.json'), true);
$myIndex = Fuse::parseIndex($fuseIndex);

// Initialize Fuse with the index
$fuse = new Fuse($list, $options, $myIndex);

Differences with Fuse.js

Development

Project Scope

Please note that I'm striving for feature parity with Fuse.js and therefore will add neither features nor fixes to the search logic that are not reflected in Fuse.js itself.

If you have any issues with search results that are not obviously bugs in this PHP port, and you happen to know JavaScript, please check if your use case works correctly in the online demo of Fuse.js as that is the canonical Fuse implementation. If the issue appears there as well, please open an issue in their repo.

Setup

To start development on Fuse, you need git, PHP (≥ 7.4) and Composer.

Since code is formatted using Prettier, it's also recommended to have Node.js/npm installed as well as using an editor which supports Prettier formatting.

Clone the repository and cd into it:

git clone https://github.com/loilo/fuse.git
cd fuse

Install Composer dependencies:

composer install

Install npm dependencies (optional but recommended). This is only needed for code formatting as npm dependencies include Prettier plugins used by this project.

npm ci

Quality Assurance

There are different kinds of code checks in place for this project. All of these are run when a pull request is submitted but can also be run locally:

Contributing

Before submitting a pull request, please add relevant tests to the test folder.