braces

What is braces?

The braces npm package is a library for expanding braces in a string, similar to how shells like bash perform brace expansion. It is used to generate permutations of strings based on patterns with braces, which is useful for file globbing, configuration, and other tasks where such patterns are needed.

What are braces's main functionalities?

Brace Expansion

Expands a pattern with comma-separated values inside braces into an array of strings.

"{a,b,c}d" -> ['ad', 'bd', 'cd']

Sequence Expansion

Expands a pattern with numerical sequences inside braces into an array of strings.

"{1..3}d" -> ['1d', '2d', '3d']

Nested Expansion

Expands a pattern with nested braces into an array of strings.

"{a,{b,c}d}e" -> ['ae', 'bde', 'cde']

Other packages similar to braces

minimatch

Minimatch is a minimal matching utility that works by converting glob expressions into JavaScript RegExp objects. It is similar to braces in that it supports pattern matching, but it is more focused on matching file paths and does not support brace expansion.

micromatch

Micromatch is a fast and accurate glob matching library, optimized for speed and accuracy. It extends the capabilities of minimatch and braces, providing more powerful matching features and better performance for complex patterns.

globby

Globby is a user-friendly glob matching library that uses micromatch under the hood. It provides a higher-level API for file matching and supports multiple patterns, making it a convenient alternative to braces for file system operations.

README

braces

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

Install

Install with npm:

$ npm install --save braces

Install with yarn:

$ yarn add braces

Usage

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]);

Why use braces?

• 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.

Optimized vs. expanded braces

Optimized

Patterns are optimized for regex and matching by default:

console.log(braces('a/{x,y,z}/b'));
//=> ['a/(x|y|z)/b']

Expanded

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

console.log(braces.expand('a/{x,y,z}/b'));
//=> ['a/x/b', 'a/y/b', 'a/z/b']

Or use options.expand:

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

Features

• 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

Lists

Uses fill-range for expanding alphabetical or numeric lists:

console.log(braces('a/{foo,bar,baz}/*.js'));
//=> ['a/(foo|bar|baz)/*.js']

console.log(braces.expand('a/{foo,bar,baz}/*.js'));
//=> ['a/foo/*.js', 'a/bar/*.js', 'a/baz/*.js']

Sequences

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

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']

// supports padded ranges
console.log(braces('a{01..03}b'));   //=> [ 'a(0[1-3])b' ]
console.log(braces('a{001..300}b')); //=> [ 'a(0{2}[1-9]|0[1-9][0-9]|[12][0-9]{2}|300)b' ]

Steps

Steps, or increments, may be used with ranges:

console.log(braces.expand('{2..10..2}'));
//=> ['2', '4', '6', '8', '10']

console.log(braces('{2..10..2}'));
//=> ['(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.

Nesting

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

"Expanded" braces

console.log(braces.expand('a{b,c,/{x,y}}/e'));
//=> ['ab/e', 'ac/e', 'a/x/e', 'a/y/e']

console.log(braces.expand('a/{x,{1..5},y}/c'));
//=> ['a/x/c', 'a/1/c', 'a/2/c', 'a/3/c', 'a/4/c', 'a/5/c', 'a/y/c']

"Optimized" braces

console.log(braces('a{b,c,/{x,y}}/e'));
//=> ['a(b|c|/(x|y))/e']

console.log(braces('a/{x,{1..5},y}/c'));
//=> ['a/(x|([1-5])|y)/c']

Escaping

Escaping braces

A brace pattern will not be expanded or evaluted if either the opening or closing brace is escaped:

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

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

Escaping commas

Commas inside braces may also be escaped:

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

console.log(braces.expand('a{d\\,c,b}e'));
//=> ['ad,ce', 'abe']

Single items

Following bash conventions, a brace pattern is also not expanded when it contains a single character:

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

Options

options.expand

Type: Boolean

Default: undefined

Description: 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' ]

options.optimize

Type: Boolean

Default: true

Description: Enabled by default.

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

options.nodupes

Type: Boolean

Default: true

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

options.rangeLimit

Type: Number

Default: 250

Description: 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.

Examples

// pattern exceeds the "rangeLimit", so it's optimized automatically
console.log(braces.expand('{1..1000}'));
//=> ['([1-9]|[1-9][0-9]{1,2}|1000)']

// pattern does not exceed "rangeLimit", so it's NOT optimized
console.log(braces.expand('{1..100}'));
//=> ['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']

options.transform

Type: Function

Default: undefined

Description: Customize range expansion.

var range = braces.expand('x{a..e}y', {
  transform: function(str) {
    return 'foo' + str;
  }
});

console.log(range);
//=> [ 'xfooay', 'xfooby', 'xfoocy', 'xfoody', 'xfooey' ]

options.quantifiers

Type: Boolean

Default: undefined

Description: 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.

Examples

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

options.unescape

Type: Boolean

Default: undefined

Description: Strip backslashes that were used for escaping from the result.

Benchmarks

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

Performance

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)

Braces is fast

• caching
• parsing
• looping
• regex: smaller patterns, optimized to match faster

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: the parser makes every effort to only evaluate each character in the pattern once.
• 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 freeze or 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:

• 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)
• we can't control how users define their patterns (and we shouldn't have to care)
• 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:

• Braces prevents malicious strings from being used by checking the length of the input strings, as well as generated regex
• Braces optimizes the generated result by default, so that only one output string is generated for every input string.
• 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)

About

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

Contributing

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

Contributors

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

Building docs

(This project's readme.md is generated by verb, please don't edit the readme directly. Any changes to the readme must be made in the .verb.md readme template.)

To generate the readme, run the following command:

$ npm install -g verbose/verb#dev verb-generate-readme && verb

Running tests

Running and reviewing unit tests is a great way to get familiarized with a library and its API. You can install dependencies and run tests with the following command:

$ npm install && npm test

Author

Jon Schlinkert

• github/jonschlinkert
• twitter/jonschlinkert

License

Copyright © 2017, Jon Schlinkert. Released under the MIT License.

This file was generated by verb-generate-readme, v0.6.0, on April 26, 2017.