outmatch

An extremely fast and lightweight glob-matching library for JavaScript with advanced features

   

Outmatch takes one or more glob patterns, compiles them into a RegExp and returns a function for matching strings with it.

Glob patterns are strings that contain wildcards such as *, ?, [abc] and others. When such a pattern is compared with another string, these wildcards can replace one or more symbols. For example, src/* would match both src/foo and src/bar.

While globs are usually used to search file paths separated by slashes, with outmatch it is possible to match arbitrary strings, whether separated or not.

Quickstart

npm install outmatch

import outmatch from 'outmatch'

const isMatch = outmatch('src/**/*.{js,ts}')

isMatch('src/components/header/index.js') //=> true
isMatch('src/README.md') //=> false

isMatch.pattern //=> 'src/**/*.{js,ts}'
isMatch.options //=> { separator: true }
isMatch.regexp //=> /^(?:src[/\\]+? ... \.ts[/\\]*?)$/

More details are available in the Installation, Usage, Syntax and API sections.

Features

🍃	Lightweight No dependencies. Just 2.4 KB when minified and gzipped — less than ⅓ of picomatch and ⅕ of micromatch
💪	Powerful Supports basic and extended globbing, proper multi-pattern compilation and custom path separators — a feature unique to outmatch
🎯	Accurate The only library that can handle negated extglobs correctly. Actually expands braces instead of merely converting them to groups
🏎	Fast Compiles and matches patterns faster than minimatch, micromatch and picomatch
⚒	Reliable Written in TypeScript. Covered by thousands of unit tests
🌞	Simple The API is a single function. Options can be specified in easy to understand language rather than Linux slang
🔌	Compatible Works in any ES5 environment including older versions of Node.js, Deno, Bun, React Native and browsers

For comparison with the alternatives, see the corresponding section.

Installation

The package is distributed via the npm package registry. It can be installed using one of the compatible package managers or included directly from a CDN.

npm

npm install outmatch

Yarn

yarn add outmatch

pnpm

pnpm install outmatch

CDN

When included from a CDN, outmatch is available as the global function outmatch.

• unpkg
• jsDelivr

Usage

Outmatch comes built in ESM, CommonJS and UMD formats and includes TypeScript typings. The examples use ESM imports, which can be replaced with the following line for CommonJS: const outmatch = require('outmatch').

The default export is a function of two arguments, first of which can be either a single glob string or an array of such patterns. The second argument is optional and can be either an options object or a separator (which will be the value of the separator option). Outmatch compiles them into a regular expression and returns a function (usually called isMatch in the examples) that tests strings against the pattern. The pattern, options and the compiled RegExp object are available as properties on the returned function:

import outmatch from 'outmatch'

const isMatch = outmatch('src/[bc]ar', { '{}': false })

isMatch('src/bar') //=> true
isMatch('src/car') //=> true
isMatch('src/tar') //=> false

isMatch.pattern //=> 'src/[bc]ar'
isMatch.options //=> { '{}': false }
isMatch.regexp //=> /^src[/\\]+?(?!\.)(?![/\\])[bc]ar[/\\]*?$/

The returned function can be invoked immediately if there is no need to match a pattern more than once:

outmatch('src/**/*.js')('src/components/body/index.js') //=> true

Compiling a pattern is much slower than comparing a string to it, so it is recommended to always reuse the returned function when possible.

File Paths and Separators

Globs are most often used to search file paths, which are, essentially, strings split into segments by slashes. While other libraries are usually restricted to this use-case, outmatch is able to work with arbitrary strings by accepting a custom separator via the second argument:

const matchDomain = outmatch('*.example.com', { separator: '.' })
matchDomain('subdomain.example.com') //=> true

const matchLike = outmatch('wh?t like**like mean', 'like') // shorthand for { separator: 'like' }
matchLike('what like do like you like mean') //=> true

The only limitation is that backslashes \ cannot be used as separators in patterns because outmatch uses them for character escaping. However, when separator is undefined or true, / in patterns will match both / and \, so a single pattern with forward slashes can match both Unix and Windows paths:

const isMatchA = outmatch('foo\\bar') // throws an error

const isMatchB = outmatch('foo/bar') // same as passing `true` as the separator

isMatchB('foo/bar') //=> true
isMatchB('foo\\bar') //=> true

const isMatchC = outmatch('foo/bar', '/')

isMatchC('foo/bar') //=> true
isMatchC('foo\\bar') //=> false

A thing to note is that most matching features work with a segment rather than a whole pattern. For example, foo/b* will match foo/bar but not foo/b/ar. The two exceptions to this are brace expansion and pattern negation, both of which work with whole patterns:

outmatch('src/{foo/bar,baz}')('src/foo/bar') //=> true (brace expansion)
outmatch('src/@(foo/bar|baz)')('src/foo/bar') //=> false (extglob)

Any string that contains a segment starting with a dot (a dotfile) is excluded unless the dot is specified explicitly in the pattern. This behavior can be disabled by setting excludeDot to false, in which case leading dots are treated like any other symbol:

outmatch('project/*')('project/.git') //=> false
outmatch('project/.*')('project/.git') //=> true (dot is specified explicitly)
outmatch('project/*', { excludeDot: false })('project/.git') //=> true

Segmentation can be turned off completely by passing false as the separator, which makes outmatch treat whole patterns as a single segment. Slashes become regular symbols, ** works identically to * and leading dots get excluded only when they are the very first character of a pattern:

const isMatch = outmatch('foo?ba*', false)
isMatch('foo/bar/.qux') //=> true

A single separator in a pattern will match any number of separators in a sample string:

outmatch('foo/bar/baz')('foo/bar///baz') //=> true

When a pattern has an explicit separator at its end, samples also require one or more trailing separators:

const isMatch = outmatch('foo/bar/')

isMatch('foo/bar') //=> false
isMatch('foo/bar/') //=> true
isMatch('foo/bar///') //=> true

However, if there is no trailing separator in a pattern, strings will match even if they have separators at the end:

const isMatch = outmatch('foo/bar')

isMatch('foo/bar') //=> true
isMatch('foo/bar/') //=> true
isMatch('foo/bar///') //=> true

Multiple Patterns

Outmatch can take an array of glob patterns as the first argument instead of a single pattern. In that case a string will be considered a match if it matches any of the given patterns:

const isMatch = outmatch(['src/*', 'tests/*'])

isMatch('src/utils.js') //=> true
isMatch('tests/utils.js') //=> true

If a negated pattern is given among positive patterns, it will work as an ignore filter for strings that match the positive patterns:

const isMatch = outmatch(['src/*', '!src/foo', '!src/bar'])

isMatch('src/foo') //=> false
isMatch('src/bar') //=> false
isMatch('src/baz') //=> true

Matching Arrays of Strings

The returned function can work with arrays of strings when used as the predicate of the native array methods:

const isMatch = outmatch(['src/**/*.js', '!**/body.js'])
const paths = ['readme.md', 'src/index.js', 'src/components/body.js']

paths.map(isMatch) //=> [ false, true, false ]
paths.filter(isMatch) //=> [ 'src/index.js' ]
paths.some(isMatch) //=> true
paths.every(isMatch) //=> false
paths.find(isMatch) //=> 'src/index.js'
paths.findIndex(isMatch) //=> 1

Syntax

Pattern	Description
Basic Wildcards
?	Matches exactly one arbitrary character excluding separators
*	Matches zero or more arbitrary characters excluding separators
Globstar
**	Matches zero or more segments when used as a whole segment in a separated pattern (e.g. /**/ if / is the separator)
Character Classes (Brackets)
[abc1_]	Matches a single character from the specified list of characters
[a-z] [0-9]	Matches a single character from the specified range of characters
[!abc] [!f-k]	Matches a single character not in the specified list or range
[[:alnum:]] [[:alpha:]] [[:lower:]]	POSIX classes are not supported. Other libraries claim to have support, but in reality it's just a gimmick. Their original purpose was to match symbols from the current locale rather than just English. For example, [[:alpha:]] would match the russian letter ы while [a-zA-Z] would not. However, making this work in JS is non-trivial, and most libraries that list POSIX classes as supported merely have them as aliases to regular character ranges, so [[:alpha:]] works identically to [a-zA-Z]
Extglobs (Parens)
@(bar|baz)	Matches one of the given subpatterns repeated exactly one time
?(foo) ?(bar|baz)	Matches one of the given subpatterns repeated zero or one time
*(foo) *(bar|baz)	Matches one of the given subpatterns repeated zero or more times
+(foo) +(bar|baz)	Matches one of the given subpatterns repeated one or more times
!(foo) !(bar|baz)	Matches anything except the given subpatterns. Cannot be nested inside another negated glob
Braces
{bar,baz}	Expands the pattern to an array of patterns, so outmatch('src/{foo,bar}/baz') is equivalent to outmatch(['src/foo/baz', 'src/bar/baz']) While braces are similar to extglobs, they work differently. Braces are expanded before anything else, so, unlike extglobs, they can handle subpatterns that contain separators. Braces can be nested: src/{foo,bar/{baz,qux}} expands to src/foo, src/bar/baz and src/bar/qux
{1..5}	Matches any character in the specified range
{01..300} {1..9..2}	Like most other libraries, outmatch doesn't support zero-padded and stepped ranges. The amount of code it would take to implement them is simply not justified by how rarely they are used
Negation
!	Negates a pattern when put at the start of it. If repeated multiple times, each ! will invert the effect, so !!foo/bar is the same as foo/bar and !!!baz/qux is the same as !baz/qux. A negated pattern matches any string that doesn't match the part after the !. When put in an array among positive patterns, negated patterns effectively work as ignores
Escaping
\	Escapes the following character making it be treated literally

API

outmatch(patterns, options?): isMatch
outmatch(patterns, separator?): isMatch

Takes a single pattern string or an array of patterns and compiles them into a regular expression. Returns an isMatch function that takes a sample string as its only argument and returns true if the string matches the pattern(s).

isMatch(sample): boolean

Tests if a sample string matches the patterns that were used to compile the regular expression and create this function.

isMatch.regexp

The compiled regular expression.

isMatch.pattern

The original pattern or array of patterns that was used to compile the regular expression and create the isMatch function.

isMatch.options

The options object that was used to compile the regular expression and create the isMatch function.

Options

Option	Type	Default Value	Description
separator	string | boolean	true	Separator to be used to split patterns and samples into segments: true — / in patterns match both / and \ in samples false — don't split any string — custom separator
flags	string	undefined	Flags to pass to the RegExp. For example, setting this option to 'i' will make the matching case-insensitive
excludeDot	boolean	true	Toggles whether to exclude strings that contain segments starting with a dot
!	boolean	true	Toggles pattern negation
?	boolean	true	Toggles single-char wildcards
*	boolean	true	Toggles multi-char wildcards
**	boolean	true	Toggles globstars
[]	boolean	true	Toggles character classes
()	boolean	true	Toggles extglobs
{}	boolean	true	Toggles brace expansion

Comparison

Pattern: src/test/**/*.?s
Sample: src/test/foo/bar.js

Compilation
  outmatch     650,098 ops/sec
  picomatch    136,927 ops/sec

Matching
  outmatch     34,524,204 ops/sec
  picomatch    10,223,739 ops/sec