Skip to content

Latest commit

 

History

History
519 lines (319 loc) · 14.6 KB

.verb.md

File metadata and controls

519 lines (319 loc) · 14.6 KB
Raw

Quickstart

var mm = require('micromatch');
mm(list, patterns[, options]);

The main export takes a list of strings and one or more glob patterns:

console.log(mm(['foo', 'bar', 'qux'], ['f*', 'b*'])); // ['foo', 'bar']

Use .isMatch to get true/false:

console.log(mm.isMatch('foo', 'f*'));  // true

Switching from minimatch and multimatch is easy:

Heads up!

There is one notable difference between micromatch and minimatch in regards to how backslashes are handled. See the notes about backslashes for more information.

Why use micromatch?

micromatch is a drop-in replacement for minimatch and multimatch

Speed and accuracy

Micromatch uses [snapdragon][] for parsing and compiling globs, which results in:

  • Granular control over the entire conversion process in a way that is easy to understand, reason about, and customize.

  • Faster matching, from a combination of optimized glob patterns and (optional) caching.

  • Much greater accuracy than minimatch. In fact, nanomatch passes all of the spec tests from bash, including some that bash still fails. However, since there is no real specification for globs, if you encounter a pattern that yields unexpected match results after researching previous issues, please let us know.

  • Micromatch is safer, and is not subject to DoS with brace patterns, like minimatch and multimatch.

  • faster matching

  • More accurate, with more than 36,000 unit tests - and thousands more patterns tested - to prove it. (minimatch and multimatch fail a great number of the tests).

  • More complete support for the Bash 4.3 specification than minimatch and multimatch

  • More reliable windows support than minimatch and multimatch

  • Supports all of the same matching features as minimatch and multimatch

  • Micromatch uses real parsers and compilers. Although edge cases are inevitable, micromatch is better equipped to handle them.

Features

  • Native support for multiple glob patterns (no need for wrappers like multimatch)
  • Basic wildcard support (**/*, a/b/*.js, or ['a/*.js', '*b.js'])
  • Negation support (['!a/*.js', '*!(b).js'])
  • [extglob][] support (+(x|y), !(a|b), etc)
  • POSIX character class support (**/[[:alpha:][:digit:]]/)
  • brace expansion support (a/b-{1..5}.md, one/{two,three}/four.md)
  • regex character classes (a/b/baz-[1-5].js)
  • regex logical or (a/b/(abc|xyz).js)

You can mix and match these features to create whatever patterns you need!

Example

mm(['a/b.js', 'a/b.md'], 'a/*.!(js)');
//=> ['a/b.md']

Switching from minimatch

Use mm.isMatch() instead of minimatch():

mm.isMatch('foo', 'b*');
//=> false

Use mm.match() instead of minimatch.match():

mm.match(['foo', 'bar'], 'b*');
//=> 'bar'

Switching from multimatch

Same signature:

mm(['foo', 'bar', 'baz'], ['f*', '*z']);
//=> ['foo', 'baz']

API

{%= apidocs("index.js") %}

Options

options.basename

Allow glob patterns without slashes to match a file path based on its basename. Same behavior as [minimatch][] option matchBase.

Type: Boolean

Default: false

Example

mm(['a/b.js', 'a/c.md'], '*.js');
//=> []

mm(['a/b.js', 'a/c.md'], '*.js', {matchBase: true});
//=> ['a/b.js']

options.bash

Enabled by default, this option enforces bash-like behavior with stars immediately following a bracket expression. Bash bracket expressions are similar to regex character classes, but unlike regex, a star following a bracket expression does not repeat the bracketed characters. Instead, the star is treated the same as an other star.

Type: Boolean

Default: true

Example

var files = ['abc', 'ajz'];
console.log(mm(files, '[a-c]*'));
//=> ['abc', 'ajz']

console.log(mm(files, '[a-c]*', {bash: false}));

options.cache

Disable regex and function memoization.

Type: Boolean

Default: undefined

options.dot

Match dotfiles. Same behavior as [minimatch][] option dot.

Type: Boolean

Default: false

options.failglob

Similar to the --failglob behavior in Bash, throws an error when no matches are found.

Type: Boolean

Default: undefined

options.ignore

String or array of glob patterns to match files to ignore.

Type: String|Array

Default: undefined

options.matchBase

Alias for options.basename.

options.nobrace

Disable expansion of brace patterns. Same behavior as [minimatch][] option nobrace.

Type: Boolean

Default: undefined

See braces for more information about extended brace expansion.

options.nocase

Use a case-insensitive regex for matching files. Same behavior as [minimatch][].

Type: Boolean

Default: undefined

options.nodupes

Remove duplicate elements from the result array.

Type: Boolean

Default: undefined

Example

Example of using the unescape and nodupes options together:

mm.match(['a/b/c', 'a/b/c'], 'a/b/c');
//=> ['a/b/c', 'a/b/c']

mm.match(['a/b/c', 'a/b/c'], 'a/b/c', {nodupes: true});
//=> ['abc']

options.noext

Disable extglob support, so that extglobs are regarded as literal characters.

Type: Boolean

Default: undefined

Examples

mm(['a/z', 'a/b', 'a/!(z)'], 'a/!(z)');
//=> ['a/b', 'a/!(z)']

mm(['a/z', 'a/b', 'a/!(z)'], 'a/!(z)', {noext: true});
//=> ['a/!(z)'] (matches only as literal characters)

options.nonegate

Disallow negation (!) patterns, and treat leading ! as a literal character to match.

Type: Boolean

Default: undefined

options.noglobstar

Disable matching with globstars (**).

Type: Boolean

Default: undefined

mm(['a/b', 'a/b/c', 'a/b/c/d'], 'a/**');
//=> ['a/b', 'a/b/c', 'a/b/c/d']

mm(['a/b', 'a/b/c', 'a/b/c/d'], 'a/**', {noglobstar: true});
//=> ['a/b']

options.nonull

Alias for options.nullglob.

options.nullglob

If true, when no matches are found the actual (arrayified) glob pattern is returned instead of an empty array. Same behavior as [minimatch][] option nonull.

Type: Boolean

Default: undefined

options.snapdragon

Pass your own instance of [snapdragon][], to customize parsers or compilers.

Type: Object

Default: undefined

options.sourcemap

Generate a source map by enabling the sourcemap option with the .parse, .compile, or .create methods.

(Note that sourcemaps are currently not enabled for brace patterns)

Examples

var mm = require('micromatch');
var pattern = '*(*(of*(a)x)z)';

var res = mm.create('abc/*.js', {sourcemap: true});
console.log(res.map);
// { version: 3,
//   sources: [ 'string' ],
//   names: [],
//   mappings: 'AAAA,GAAG,EAAC,iBAAC,EAAC,EAAE',
//   sourcesContent: [ 'abc/*.js' ] }

var ast = mm.parse('abc/**/*.js');
var res = mm.compile(ast, {sourcemap: true});
console.log(res.map);
// { version: 3,
//   sources: [ 'string' ],
//   names: [],
//   mappings: 'AAAA,GAAG,EAAC,2BAAE,EAAC,iBAAC,EAAC,EAAE',
//   sourcesContent: [ 'abc/**/*.js' ] }

var ast = mm.parse(pattern);
var res = mm.compile(ast, {sourcemap: true});
console.log(res.map);
// { version: 3,
//   sources: [ 'string' ],
//   names: [],
//   mappings: 'AAAA,CAAE,CAAE,EAAE,CAAE,CAAC,EAAC,CAAC,EAAC,CAAC,EAAC',
//   sourcesContent: [ '*(*(of*(a)x)z)' ] }

options.unescape

Remove backslashes from returned matches.

Type: Boolean

Default: undefined

Example

In this example we want to match a literal *:

mm.match(['abc', 'a\\*c'], 'a\\*c');
//=> ['a\\*c']

mm.match(['abc', 'a\\*c'], 'a\\*c', {unescape: true});
//=> ['a*c']

options.unixify

Convert path separators on returned files to posix/unix-style forward slashes.

Type: Boolean

Default: true on windows, false everywhere else

Example

mm.match(['a\\b\\c'], 'a/**');
//=> ['a/b/c']

mm.match(['a\\b\\c'], {unixify: false});
//=> ['a\\b\\c']

Extended globbing

Micromatch also supports extended globbing features.

extglobs

Extended globbing, as described by the bash man page:

pattern regex equivalent description
?(pattern-list) `(foo bar)?`
*(pattern-list) `(foo bar)*`
+(pattern-list) `(foo bar)+`
@(pattern-list) `(foo bar)` *
!(pattern-list) N/A (equivalent regex is much more complicated) Matches anything except one of the given patterns

* Note that @ isn't a RegEx character.

Powered by extglob. Visit that library for the full range of options or to report extglob related issues.

braces

Expanded braces

Braces are expanded when ``

  • range expansion: a{1..3}b/*.js expands to: ['a1b/*.js', 'a2b/*.js', 'a3b/*.js']
  • nesting: a{c,{d,e}}b/*.js expands to: ['acb/*.js', 'adb/*.js', 'aeb/*.js']

Optimized braces (not expanded)

By default, brace patterns work the same way regex logical OR operators. For example, (a|b) will achieve the same result as {a,b}.

Visit braces to see the full range of features and options related to brace expansion, or to create brace matching or expansion related issues.

regex character classes

Given the list: ['a.js', 'b.js', 'c.js', 'd.js', 'E.js']:

  • [ac].js: matches both a and c, returning ['a.js', 'c.js']
  • [b-d].js: matches from b to d, returning ['b.js', 'c.js', 'd.js']
  • [b-d].js: matches from b to d, returning ['b.js', 'c.js', 'd.js']
  • a/[A-Z].js: matches and uppercase letter, returning ['a/E.md']

Learn about regex character classes.

regex groups

Given ['a.js', 'b.js', 'c.js', 'd.js', 'E.js']:

  • (a|c).js: would match either a or c, returning ['a.js', 'c.js']
  • (b|d).js: would match either b or d, returning ['b.js', 'd.js']
  • (b|[A-Z]).js: would match either b or an uppercase letter, returning ['b.js', 'E.js']

As with regex, parens can be nested, so patterns like ((a|b)|c)/b will work. Although brace expansion might be friendlier to use, depending on preference.

POSIX bracket expressions

POSIX brackets are intended to be more user-friendly than regex character classes. This of course is in the eye of the beholder.

Example

mm.isMatch('a1', '[[:alpha:][:digit:]]');
//=> true

mm.isMatch('a1', '[[:alpha:][:alpha:]]');
//=> false

See expand-brackets for more information about bracket expressions.

Notes

Bash 4.3 parity

Whenever possible matching behavior is based on behavior Bash 4.3, which is mostly consistent with minimatch.

However, it's suprising how many edge cases and rabbit holes there are with glob matching, and since there is no real glob specification, and micromatch is more accurate than both Bash and minimatch, there are cases where best-guesses were made for behavior. In a few cases where Bash had no answers, we used wildmatch (used by git) as a fallback.

Backslashes

There is an important, notable difference between minimatch and micromatch in regards to how backslashes are handled in glob patterns.

  • Micromatch exclusively and explicitly reserves backslashes for escaping characters in a glob pattern, even on windows. This is consistent with bash behavior.
  • Minimatch converts all backslashes to forward slashes, which means you can't use backslashes to escape any characters in your glob patterns.

We made this decision for micromatch for a couple of reasons:

  • consistency with bash conventions.
  • glob patterns are not filepaths. They are a type of regular language that is converted to a JavaScript regular expression. Thus, when forward slashes are defined in a glob pattern, the resulting regular expression will match windows or POSIX path separators just fine.

A note about joining paths to globs

Note that when you pass something like path.join('foo', '*') to micromatch, you are creating a filepath and expecting it to still work as a glob pattern. This causes problems on windows, since the path.sep is \\.

In other words, since \\ is reserved as an escape character in globs, on windows path.join('foo', '*') would result in foo\\*, which tells micromatch to match * as a literal character. This is the same behavior as bash.

Contributing

All contributions are welcome! Please read the contributing guide to get started.

Bug reports

Please feel free to create an issue if you find something that you think isn't right. However, if you find a matching-related issue, please try do the following first:

  • research existing issues first
  • visit the GNU Bash documentation to see how Bash deals with the pattern
  • visit the [minimatch][] documentation to cross-check expected behavior

Benchmarks

Running benchmarks

Install dev dependencies:

npm i -d && npm benchmark

Latest results

As of {%= date() %} (longer bars are better):

{%= bench() %}