node package manager


Fast, comprehensive, bash-like brace expansion implemented in JavaScript. Complete support for the Bash 4.3 braces specification, without sacrificing speed.

braces NPM version NPM monthly downloads NPM total downloads Linux Build Status Windows Build Status

Fast, comprehensive, bash-like brace expansion implemented in JavaScript. Complete support for the Bash 4.3 braces specification, without sacrificing speed.


Install with npm:

$ npm install --save braces


The main export is a function that takes a brace pattern to expand and an options object if necessary.

var braces = require('braces');
braces(pattern[, options]);


  • Safer: Braces isn't vulnerable to DoS attacks, unlike brace-expansion, minimatch and multimatch (this not the same DoS bug that was found in minimatch and multimatch recently)
  • Accurate: complete support for the Bash 4.3 Brace Expansion specification (passes all of the Bash braces tests)
  • fast and performant: not only runs fast, but scales well as patterns increase in complexity (other libs don't - see the notes on performance).
  • Organized code base: with parser and compiler that are easy to maintain and update when edge cases crop up.
  • Well-tested: 900+ unit tests with thousands of actual patterns tested. Passes 100% of the minimatch and brace-expansion unit tests as well.
  • Optimized braces: By default returns an optimized string that can be used for creating regular expressions for matching.
  • Expanded braces: Optionally returns an array (like bash). See a comparison between optimized and expanded.


Patterns are optimized for regex and matching by default:

//=> ['a/(x|y|z)/b'] 


To expand patterns the same way as Bash or minimatch, use the .expand method:

//=> ['a/x/b', 'a/y/b', 'a/z/b'] 


  • lists: Supports "lists": a/{b,c}/d => ['a/b/d', 'a/c/d']
  • sequences: Supports alphabetical or numerical "sequences" (ranges): {1..3} => ['1', '2', '3']
  • steps: Supports "steps" or increments: {2..10..2} => ['2', '4', '6', '8', '10']
  • escaping
  • options


Uses fill-range for expanding alphabetical or numeric lists:

//=> ['a/(foo|bar|baz)/*.js'] 
//=> ['a/foo/*.js', 'a/bar/*.js', 'a/baz/*.js'] 


Uses fill-range for expanding alphabetical or numeric ranges (bash "sequences"):

// padding is not retained when string is optimized 
console.log(braces('a{01..03}b')); // ['a([1-3])b'] 
console.log(braces('a{1..3}b'));   // ['a([1-3])b'] 
console.log(braces.expand('{1..3}'));     // ['1', '2', '3'] 
console.log(braces.expand('a{01..03}b')); // ['a01b', 'a02b', 'a03b'] 
console.log(braces.expand('a{1..3}b'));   // ['a1b', 'a2b', 'a3b'] 
console.log(braces.expand('{a..c}'));     // ['a', 'b', 'c'] 
console.log(braces.expand('foo/{a..c}')); // ['foo/a', 'foo/b', 'foo/c'] 


Steps, or increments, may be used with ranges:

//=> ['2', '4', '6', '8', '10'] 
//=> ['(2|4|6|8|10)'] 

When the .optimize method is used, or options.optimize is set to true, sequences are passed to to-regex-range for expansion.


Brace patterns may be nested. The results of each expanded string are not sorted, and left to right order is preserved.

"Expanded" examples

//=> ['ab/e', 'ac/e', 'a/x/e', 'a/y/e'] 
//=> ['a/x/c', 'a/1/c', 'a/2/c', 'a/3/c', 'a/4/c', 'a/5/c', 'a/y/c'] 

"Optimized" examples

//=> ['a(b|c|/(x|y))/e'] 
//=> ['a/(x|([1-5])|y)/c'] 


Escaping braces

Prevent braces from being expanded or evaluted by escaping either the opening or closing brace:

//=> ['a{d,c,b}e'] 
//=> ['a{d,c,b}e'] 

Escaping commas

Commas inside braces may also be escaped:

//=> ['ad,ce', 'abe'] 

Single items

A brace pattern is also considered to be escaped when it contains a single item:

//=> ['a{b}c'] 
//=> ['a{b,c}d'] 



Type: Boolean

Default: undefined

Generate an "expanded" brace pattern (this option is unncessary with the .expand method, which does the same thing).

console.log(braces('a/{b,c}/d', {expand: true}));
//=> [ 'a/b/d', 'a/c/d' ] 


Type: Boolean

Default: true

Enabled by default.

//=> [ 'a/(b|c)/d' ] 


Type: Boolean

Default: true

Duplicates are removed by default. To keep duplicates, pass {nodupes: false} on the options


Type: Number

Default: 250

When braces.expand() is used, or options.expand is true, brace patterns will automatically be optimized when the difference between the range minimum and range maximum exceeds the rangeLimit. This is to prevent huge ranges from freezing your application.

You can set this to any number, or change options.rangeLimit to Inifinity to disable this altogether.


// pattern exceeds the "rangeLimit", so it's optimized automatically 
//=> ['([1-9]|[1-9][0-9]{1,2}|1000)'] 
// pattern does not exceed "rangeLimit", so it's NOT optimized 
//=> ['1', '2', '3', '4', '5', '6', '7', '8', '9', '10', '11', '12', '13', '14', '15', '16', '17', '18', '19', '20', '21', '22', '23', '24', '25', '26', '27', '28', '29', '30', '31', '32', '33', '34', '35', '36', '37', '38', '39', '40', '41', '42', '43', '44', '45', '46', '47', '48', '49', '50', '51', '52', '53', '54', '55', '56', '57', '58', '59', '60', '61', '62', '63', '64', '65', '66', '67', '68', '69', '70', '71', '72', '73', '74', '75', '76', '77', '78', '79', '80', '81', '82', '83', '84', '85', '86', '87', '88', '89', '90', '91', '92', '93', '94', '95', '96', '97', '98', '99', '100'] 


Type: Function

Default: undefined

Customize range expansion.

var range = braces.expand('x{a..e}y', {
  transform: function(str) {
    return 'foo' + str;
//=> [ 'xfooay', 'xfooby', 'xfoocy', 'xfoody', 'xfooey' ] 


Type: Boolean

Default: undefined

In regular expressions, quanitifiers can be used to specify how many times a token can be repeated. For example, a{1,3} will match the letter a one to three times.

Unfortunately, regex quantifiers happen to share the same syntax as Bash lists

The quantifiers option tells braces to detect when regex quantifiers are defined in the given pattern, and not to try to expand them as lists.


var braces = require('braces');
//=> [ 'a/b(1|3)/(x|y|z)' ] 
console.log(braces('a/b{1,3}/{x,y,z}', {quantifiers: true}));
//=> [ 'a/b{1,3}/(x|y|z)' ] 
console.log(braces('a/b{1,3}/{x,y,z}', {quantifiers: true, expand: true}));
//=> [ 'a/b{1,3}/x', 'a/b{1,3}/y', 'a/b{1,3}/z' ] 


Running benchmarks

Install dev dependencies:

npm i -d && npm benchmark

Latest results

Benchmarking: (8 of 8)
 · combination-nested
 · combination
 · escaped
 · list-basic
 · list-multiple
 · no-braces
 · sequence-basic
 · sequence-multiple
# benchmark/fixtures/combination-nested.js (52 bytes)
  brace-expansion x 4,756 ops/sec ±1.09% (86 runs sampled)
  braces x 11,202,303 ops/sec ±1.06% (88 runs sampled)
  minimatch x 4,816 ops/sec ±0.99% (87 runs sampled)
  fastest is braces
# benchmark/fixtures/combination.js (51 bytes)
  brace-expansion x 625 ops/sec ±0.87% (87 runs sampled)
  braces x 11,031,884 ops/sec ±0.72% (90 runs sampled)
  minimatch x 637 ops/sec ±0.84% (88 runs sampled)
  fastest is braces
# benchmark/fixtures/escaped.js (44 bytes)
  brace-expansion x 163,325 ops/sec ±1.05% (87 runs sampled)
  braces x 10,655,071 ops/sec ±1.22% (88 runs sampled)
  minimatch x 147,495 ops/sec ±0.96% (88 runs sampled)
  fastest is braces
# benchmark/fixtures/list-basic.js (40 bytes)
  brace-expansion x 99,726 ops/sec ±1.07% (83 runs sampled)
  braces x 10,596,584 ops/sec ±0.98% (88 runs sampled)
  minimatch x 100,069 ops/sec ±1.17% (86 runs sampled)
  fastest is braces
# benchmark/fixtures/list-multiple.js (52 bytes)
  brace-expansion x 34,348 ops/sec ±1.08% (88 runs sampled)
  braces x 9,264,131 ops/sec ±1.12% (88 runs sampled)
  minimatch x 34,893 ops/sec ±0.87% (87 runs sampled)
  fastest is braces
# benchmark/fixtures/no-braces.js (48 bytes)
  brace-expansion x 275,368 ops/sec ±1.18% (89 runs sampled)
  braces x 9,134,677 ops/sec ±0.95% (88 runs sampled)
  minimatch x 3,755,954 ops/sec ±1.13% (89 runs sampled)
  fastest is braces
# benchmark/fixtures/sequence-basic.js (41 bytes)
  brace-expansion x 5,492 ops/sec ±1.35% (87 runs sampled)
  braces x 8,485,034 ops/sec ±1.28% (89 runs sampled)
  minimatch x 5,341 ops/sec ±1.17% (87 runs sampled)
  fastest is braces
# benchmark/fixtures/sequence-multiple.js (51 bytes)
  brace-expansion x 116 ops/sec ±0.77% (77 runs sampled)
  braces x 9,445,118 ops/sec ±1.32% (84 runs sampled)
  minimatch x 109 ops/sec ±1.16% (76 runs sampled)
  fastest is braces


What's the big deal?

If you use globbing a lot, whether in your build tool chain or in your application, the speed of the glob matcher not only impacts your experience, but a slow glob matcher can slow everything else down, limitimg what you thought was possible in your application.

Braces is fast

minimatch gets exponentially slower as patterns increase in complexity, braces does not

Brace patterns are meant to be expanded - at least, if you're using Bash, that is. It's not at all unusual for users to want to use brace patterns for dates or similar ranges. It's also very easy to define a brace pattern that appears to be very simple but in fact is fairly complex.

For example, here is how generated regex size and processing time compare as patterns increase in complexity:

(the following results were generated using braces() and minimatch.braceExpand())

Pattern minimatch braces
{1..9007199254740991}[1] N/A (freezes) 300 B (12ms 878μs)
{1..10000000} 98.9 MB (20s 193ms 13μs) 34 B (11ms 204μs)
{1..1000000} 8.89 MB (1s 838ms 718μs) 33 B (1ms 761μs)
{1..100000} 789 kB (181ms 518μs) 32 B (1ms 76μs)
{1..10000} 68.9 kB (17ms 436μs) 31 B (1ms 382μs)
{1..1000} 5.89 kB (1ms 773μs) 30 B (1ms 509μs)
{1..100} 491 B (321μs) 24 B (309μs)
{1..10} 40 B (56μs) 12 B (333μs)
{1..3} 11 B (80μs) 9 B (370μs)

These numbers are actually pretty small as far as numeric ranges are concerned. Regardless, we shouldn't have to consider such things when creating a glob pattern. The tool should get out of your way and let you be as creative as you want.

Braces is performant

Even when brace patterns are fully expanded, braces is still much faster.

(the following results were generated using braces.expand() and minimatch.braceExpand())

Pattern minimatch braces
a/{1..10000000} 98.9 MB (19s 754ms 376μs) 98.9 MB (5s 734ms 419μs)
a/{1..1000000} 8.89 MB (1s 866ms 968μs) 8.89 MB (561ms 594μs)
a/{1..100000} 789 kB (178ms 311μs) 789 kB (29ms 823μs)
a/{1..10000} 68.9 kB (17ms 692μs) 68.9 kB (2ms 351μs)
a/{1..1000} 5.89 kB (1ms 823μs) 5.89 kB (706μs)
a/{1..100} 491 B (609μs) 491 B (267μs)
a/{1..10} 40 B (61μs) 40 B (636μs)
a/{1..3} 11 B (206μs) 11 B (207μs)

Why is braces so fast?

Braces was built from the ground up to be as performant as possible when matching, using a combination of the following:

  • minimizes loops and iterating: we try to pass over the pattern once to parse it before passing it to the compiler.
  • generates an optimized string: sequences/ranges are optimized by to-regex-range, which is not only highly accurate, but produces patterns that are a fraction of the size of patterns generated by other brace expansion libraries, such as Bash and minimatch (via brace-expansion)
  • generates results faster: can handle even the most complex patterns that cause other implementations like minimatch and Bash to fail.

Braces is safe

Last, but perhaps most important of all, unlike brace-expansion and minimatch, braces will not freeze your application when a user passes a complex brace pattern.

The trouble with tribbles brace patterns

There are unlimited scenarios where brace patterns cause the resulting array to grow geometrically. If you define two brace patterns for example, the result is multiplicative if not exponential. For example, given the pattern {1..1000}, we would end up with 1 thousand strings. But if two patterns are given, such as {1...1000}{1...1000}, we would end up with 1 million strings.

It's easy to think, "I wouldn't do that", but that's not the point. Even though those example patterns are contrived:

  1. it's not uncommon for users to define brace patterns that result in similarly huge strings (consider date ranges 1-2016, for example, and how easy it might be for someone to also add an alphabetical range if searching records of some kind)
  2. we can't control how users define their patterns (and we shouldn't have to care)
  3. even if it's uncommon for user to define unwieldy patterns, there is still potential for intentionally exploiting this feature to cause a DoS in your application (and given that this is a not-very-technically-advanced method for causing a DoS - e.g. anyone can do it, you should be concerned!)

Potential for DoS attacks

Most brace expansion libraries, including Bash, minimatch, multimatch and brace-expansion are vulnerable to DoS attacks (similar to the ReDoS bug minimatch had recently).

Braces mitigates this risk in three ways:

  1. Braces prevents malicious strings from being used by checking the length of the input strings, as well as generated regex
  2. Braces optimizes the generated result by default, so that only one output string is generated for every input string.
  3. When .expand or options.expand are used, braces has checks in place to prevent huge ranges from being (accidentally) created. By default, when more than 250 strings will result from the given pattern, braces will optimize the result instead of expanding it. You can override this limit of 250 by passing a custom number on options.rangeLimit.

(you might also want to consider using micromatch for glob matching, as a safer alternative to minimatch and multimatch)


Related projects

  • expand-brackets: Expand POSIX bracket expressions (character classes) in glob patterns. | homepage
  • extglob: Extended glob support for JavaScript. Adds (almost) the expressive power of regular expressions to glob… more | homepage
  • fill-range: Fill in a range of numbers or letters, optionally passing an increment or step to… more | homepage
  • micromatch: Glob matching for javascript/node.js. A drop-in replacement and faster alternative to minimatch and multimatch. | homepage
  • nanomatch: Fast, minimal glob matcher for node.js. Similar to micromatch, minimatch and multimatch, but complete Bash… more | homepage


Pull requests and stars are always welcome. For bugs and feature requests, please create an issue.


Commits Contributor
143 jonschlinkert
4 doowb
1 es128
1 eush77
1 hemanth

Building docs

(This document was generated by verb-generate-readme (a verb generator), please don't edit the readme directly. Any changes to the readme must be made in

To generate the readme and API documentation with verb:

$ npm install -g verb verb-generate-readme && verb

Running tests

Install dev dependencies:

$ npm install -d && npm test


Jon Schlinkert


Copyright © 2016, Jon Schlinkert. Released under the MIT license.

This file was generated by verb-generate-readme, v0.2.0, on December 10, 2016.

  1. this is the largest safe integer allowed in JavaScript.