1# Glob
2
3Match files using the patterns the shell uses, like stars and stuff.
4
5[![Build Status](https://travis-ci.org/isaacs/node-glob.svg?branch=master)](https://travis-ci.org/isaacs/node-glob/) [![Build Status](https://ci.appveyor.com/api/projects/status/kd7f3yftf7unxlsx?svg=true)](https://ci.appveyor.com/project/isaacs/node-glob) [![Coverage Status](https://coveralls.io/repos/isaacs/node-glob/badge.svg?branch=master&service=github)](https://coveralls.io/github/isaacs/node-glob?branch=master)
6
7This is a glob implementation in JavaScript.  It uses the `minimatch`
8library to do its matching.
9
10![a fun cartoon logo made of glob characters](logo/glob.png)
11
12## Usage
13
14Install with npm
15
16```
17npm i glob
18```
19
20```javascript
21var glob = require("glob")
22
23// options is optional
24glob("**/*.js", options, function (er, files) {
25  // files is an array of filenames.
26  // If the `nonull` option is set, and nothing
27  // was found, then files is ["**/*.js"]
28  // er is an error object or null.
29})
30```
31
32## Glob Primer
33
34"Globs" are the patterns you type when you do stuff like `ls *.js` on
35the command line, or put `build/*` in a `.gitignore` file.
36
37Before parsing the path part patterns, braced sections are expanded
38into a set.  Braced sections start with `{` and end with `}`, with any
39number of comma-delimited sections within.  Braced sections may contain
40slash characters, so `a{/b/c,bcd}` would expand into `a/b/c` and `abcd`.
41
42The following characters have special magic meaning when used in a
43path portion:
44
45* `*` Matches 0 or more characters in a single path portion
46* `?` Matches 1 character
47* `[...]` Matches a range of characters, similar to a RegExp range.
48  If the first character of the range is `!` or `^` then it matches
49  any character not in the range.
50* `!(pattern|pattern|pattern)` Matches anything that does not match
51  any of the patterns provided.
52* `?(pattern|pattern|pattern)` Matches zero or one occurrence of the
53  patterns provided.
54* `+(pattern|pattern|pattern)` Matches one or more occurrences of the
55  patterns provided.
56* `*(a|b|c)` Matches zero or more occurrences of the patterns provided
57* `@(pattern|pat*|pat?erN)` Matches exactly one of the patterns
58  provided
59* `**` If a "globstar" is alone in a path portion, then it matches
60  zero or more directories and subdirectories searching for matches.
61  It does not crawl symlinked directories.
62
63### Dots
64
65If a file or directory path portion has a `.` as the first character,
66then it will not match any glob pattern unless that pattern's
67corresponding path part also has a `.` as its first character.
68
69For example, the pattern `a/.*/c` would match the file at `a/.b/c`.
70However the pattern `a/*/c` would not, because `*` does not start with
71a dot character.
72
73You can make glob treat dots as normal characters by setting
74`dot:true` in the options.
75
76### Basename Matching
77
78If you set `matchBase:true` in the options, and the pattern has no
79slashes in it, then it will seek for any file anywhere in the tree
80with a matching basename.  For example, `*.js` would match
81`test/simple/basic.js`.
82
83### Empty Sets
84
85If no matching files are found, then an empty array is returned.  This
86differs from the shell, where the pattern itself is returned.  For
87example:
88
89    $ echo a*s*d*f
90    a*s*d*f
91
92To get the bash-style behavior, set the `nonull:true` in the options.
93
94### See Also:
95
96* `man sh`
97* `man bash` (Search for "Pattern Matching")
98* `man 3 fnmatch`
99* `man 5 gitignore`
100* [minimatch documentation](https://github.com/isaacs/minimatch)
101
102## glob.hasMagic(pattern, [options])
103
104Returns `true` if there are any special characters in the pattern, and
105`false` otherwise.
106
107Note that the options affect the results.  If `noext:true` is set in
108the options object, then `+(a|b)` will not be considered a magic
109pattern.  If the pattern has a brace expansion, like `a/{b/c,x/y}`
110then that is considered magical, unless `nobrace:true` is set in the
111options.
112
113## glob(pattern, [options], cb)
114
115* `pattern` `{String}` Pattern to be matched
116* `options` `{Object}`
117* `cb` `{Function}`
118  * `err` `{Error | null}`
119  * `matches` `{Array<String>}` filenames found matching the pattern
120
121Perform an asynchronous glob search.
122
123## glob.sync(pattern, [options])
124
125* `pattern` `{String}` Pattern to be matched
126* `options` `{Object}`
127* return: `{Array<String>}` filenames found matching the pattern
128
129Perform a synchronous glob search.
130
131## Class: glob.Glob
132
133Create a Glob object by instantiating the `glob.Glob` class.
134
135```javascript
136var Glob = require("glob").Glob
137var mg = new Glob(pattern, options, cb)
138```
139
140It's an EventEmitter, and starts walking the filesystem to find matches
141immediately.
142
143### new glob.Glob(pattern, [options], [cb])
144
145* `pattern` `{String}` pattern to search for
146* `options` `{Object}`
147* `cb` `{Function}` Called when an error occurs, or matches are found
148  * `err` `{Error | null}`
149  * `matches` `{Array<String>}` filenames found matching the pattern
150
151Note that if the `sync` flag is set in the options, then matches will
152be immediately available on the `g.found` member.
153
154### Properties
155
156* `minimatch` The minimatch object that the glob uses.
157* `options` The options object passed in.
158* `aborted` Boolean which is set to true when calling `abort()`.  There
159  is no way at this time to continue a glob search after aborting, but
160  you can re-use the statCache to avoid having to duplicate syscalls.
161* `cache` Convenience object.  Each field has the following possible
162  values:
163  * `false` - Path does not exist
164  * `true` - Path exists
165  * `'FILE'` - Path exists, and is not a directory
166  * `'DIR'` - Path exists, and is a directory
167  * `[file, entries, ...]` - Path exists, is a directory, and the
168    array value is the results of `fs.readdir`
169* `statCache` Cache of `fs.stat` results, to prevent statting the same
170  path multiple times.
171* `symlinks` A record of which paths are symbolic links, which is
172  relevant in resolving `**` patterns.
173* `realpathCache` An optional object which is passed to `fs.realpath`
174  to minimize unnecessary syscalls.  It is stored on the instantiated
175  Glob object, and may be re-used.
176
177### Events
178
179* `end` When the matching is finished, this is emitted with all the
180  matches found.  If the `nonull` option is set, and no match was found,
181  then the `matches` list contains the original pattern.  The matches
182  are sorted, unless the `nosort` flag is set.
183* `match` Every time a match is found, this is emitted with the specific
184  thing that matched. It is not deduplicated or resolved to a realpath.
185* `error` Emitted when an unexpected error is encountered, or whenever
186  any fs error occurs if `options.strict` is set.
187* `abort` When `abort()` is called, this event is raised.
188
189### Methods
190
191* `pause` Temporarily stop the search
192* `resume` Resume the search
193* `abort` Stop the search forever
194
195### Options
196
197All the options that can be passed to Minimatch can also be passed to
198Glob to change pattern matching behavior.  Also, some have been added,
199or have glob-specific ramifications.
200
201All options are false by default, unless otherwise noted.
202
203All options are added to the Glob object, as well.
204
205If you are running many `glob` operations, you can pass a Glob object
206as the `options` argument to a subsequent operation to shortcut some
207`stat` and `readdir` calls.  At the very least, you may pass in shared
208`symlinks`, `statCache`, `realpathCache`, and `cache` options, so that
209parallel glob operations will be sped up by sharing information about
210the filesystem.
211
212* `cwd` The current working directory in which to search.  Defaults
213  to `process.cwd()`.  This option is always coerced to use
214  forward-slashes as a path separator, because it is not tested
215  as a glob pattern, so there is no need to escape anything.
216* `root` The place where patterns starting with `/` will be mounted
217  onto.  Defaults to `path.resolve(options.cwd, "/")` (`/` on Unix
218  systems, and `C:\` or some such on Windows.)  This option is
219  always coerced to use forward-slashes as a path separator,
220  because it is not tested as a glob pattern, so there is no need
221  to escape anything.
222* `windowsPathsNoEscape` Use `\\` as a path separator _only_, and
223  _never_ as an escape character.  If set, all `\\` characters
224  are replaced with `/` in the pattern.  Note that this makes it
225  **impossible** to match against paths containing literal glob
226  pattern characters, but allows matching with patterns constructed
227  using `path.join()` and `path.resolve()` on Windows platforms,
228  mimicking the (buggy!) behavior of Glob v7 and before on
229  Windows.  Please use with caution, and be mindful of [the caveat
230  below about Windows paths](#windows).  (For legacy reasons,
231  this is also set if `allowWindowsEscape` is set to the exact
232  value `false`.)
233* `dot` Include `.dot` files in normal matches and `globstar` matches.
234  Note that an explicit dot in a portion of the pattern will always
235  match dot files.
236* `nomount` By default, a pattern starting with a forward-slash will be
237  "mounted" onto the root setting, so that a valid filesystem path is
238  returned.  Set this flag to disable that behavior.
239* `mark` Add a `/` character to directory matches.  Note that this
240  requires additional stat calls.
241* `nosort` Don't sort the results.
242* `stat` Set to true to stat *all* results.  This reduces performance
243  somewhat, and is completely unnecessary, unless `readdir` is presumed
244  to be an untrustworthy indicator of file existence.
245* `silent` When an unusual error is encountered when attempting to
246  read a directory, a warning will be printed to stderr.  Set the
247  `silent` option to true to suppress these warnings.
248* `strict` When an unusual error is encountered when attempting to
249  read a directory, the process will just continue on in search of
250  other matches.  Set the `strict` option to raise an error in these
251  cases.
252* `cache` See `cache` property above.  Pass in a previously generated
253  cache object to save some fs calls.
254* `statCache` A cache of results of filesystem information, to prevent
255  unnecessary stat calls.  While it should not normally be necessary
256  to set this, you may pass the statCache from one glob() call to the
257  options object of another, if you know that the filesystem will not
258  change between calls.  (See "Race Conditions" below.)
259* `symlinks` A cache of known symbolic links.  You may pass in a
260  previously generated `symlinks` object to save `lstat` calls when
261  resolving `**` matches.
262* `sync` DEPRECATED: use `glob.sync(pattern, opts)` instead.
263* `nounique` In some cases, brace-expanded patterns can result in the
264  same file showing up multiple times in the result set.  By default,
265  this implementation prevents duplicates in the result set.  Set this
266  flag to disable that behavior.
267* `nonull` Set to never return an empty set, instead returning a set
268  containing the pattern itself.  This is the default in glob(3).
269* `debug` Set to enable debug logging in minimatch and glob.
270* `nobrace` Do not expand `{a,b}` and `{1..3}` brace sets.
271* `noglobstar` Do not match `**` against multiple filenames.  (Ie,
272  treat it as a normal `*` instead.)
273* `noext` Do not match `+(a|b)` "extglob" patterns.
274* `nocase` Perform a case-insensitive match.  Note: on
275  case-insensitive filesystems, non-magic patterns will match by
276  default, since `stat` and `readdir` will not raise errors.
277* `matchBase` Perform a basename-only match if the pattern does not
278  contain any slash characters.  That is, `*.js` would be treated as
279  equivalent to `**/*.js`, matching all js files in all directories.
280* `nodir` Do not match directories, only files.  (Note: to match
281  *only* directories, simply put a `/` at the end of the pattern.)
282* `ignore` Add a pattern or an array of glob patterns to exclude matches.
283  Note: `ignore` patterns are *always* in `dot:true` mode, regardless
284  of any other settings.
285* `follow` Follow symlinked directories when expanding `**` patterns.
286  Note that this can result in a lot of duplicate references in the
287  presence of cyclic links.
288* `realpath` Set to true to call `fs.realpath` on all of the results.
289  In the case of a symlink that cannot be resolved, the full absolute
290  path to the matched entry is returned (though it will usually be a
291  broken symlink)
292* `absolute` Set to true to always receive absolute paths for matched
293  files.  Unlike `realpath`, this also affects the values returned in
294  the `match` event.
295* `fs` File-system object with Node's `fs` API. By default, the built-in
296  `fs` module will be used. Set to a volume provided by a library like
297  `memfs` to avoid using the "real" file-system.
298
299## Comparisons to other fnmatch/glob implementations
300
301While strict compliance with the existing standards is a worthwhile
302goal, some discrepancies exist between node-glob and other
303implementations, and are intentional.
304
305The double-star character `**` is supported by default, unless the
306`noglobstar` flag is set.  This is supported in the manner of bsdglob
307and bash 4.3, where `**` only has special significance if it is the only
308thing in a path part.  That is, `a/**/b` will match `a/x/y/b`, but
309`a/**b` will not.
310
311Note that symlinked directories are not crawled as part of a `**`,
312though their contents may match against subsequent portions of the
313pattern.  This prevents infinite loops and duplicates and the like.
314
315If an escaped pattern has no matches, and the `nonull` flag is set,
316then glob returns the pattern as-provided, rather than
317interpreting the character escapes.  For example,
318`glob.match([], "\\*a\\?")` will return `"\\*a\\?"` rather than
319`"*a?"`.  This is akin to setting the `nullglob` option in bash, except
320that it does not resolve escaped pattern characters.
321
322If brace expansion is not disabled, then it is performed before any
323other interpretation of the glob pattern.  Thus, a pattern like
324`+(a|{b),c)}`, which would not be valid in bash or zsh, is expanded
325**first** into the set of `+(a|b)` and `+(a|c)`, and those patterns are
326checked for validity.  Since those two are valid, matching proceeds.
327
328### Comments and Negation
329
330Previously, this module let you mark a pattern as a "comment" if it
331started with a `#` character, or a "negated" pattern if it started
332with a `!` character.
333
334These options were deprecated in version 5, and removed in version 6.
335
336To specify things that should not match, use the `ignore` option.
337
338## Windows
339
340**Please only use forward-slashes in glob expressions.**
341
342Though windows uses either `/` or `\` as its path separator, only `/`
343characters are used by this glob implementation.  You must use
344forward-slashes **only** in glob expressions.  Back-slashes will always
345be interpreted as escape characters, not path separators.
346
347Results from absolute patterns such as `/foo/*` are mounted onto the
348root setting using `path.join`.  On windows, this will by default result
349in `/foo/*` matching `C:\foo\bar.txt`.
350
351To automatically coerce all `\` characters to `/` in pattern
352strings, **thus making it impossible to escape literal glob
353characters**, you may set the `windowsPathsNoEscape` option to
354`true`.
355
356## Race Conditions
357
358Glob searching, by its very nature, is susceptible to race conditions,
359since it relies on directory walking and such.
360
361As a result, it is possible that a file that exists when glob looks for
362it may have been deleted or modified by the time it returns the result.
363
364As part of its internal implementation, this program caches all stat
365and readdir calls that it makes, in order to cut down on system
366overhead.  However, this also makes it even more susceptible to races,
367especially if the cache or statCache objects are reused between glob
368calls.
369
370Users are thus advised not to use a glob result as a guarantee of
371filesystem state in the face of rapid changes.  For the vast majority
372of operations, this is never a problem.
373
374## Glob Logo
375Glob's logo was created by [Tanya Brassie](http://tanyabrassie.com/). Logo files can be found [here](https://github.com/isaacs/node-glob/tree/master/logo).
376
377The logo is licensed under a [Creative Commons Attribution-ShareAlike 4.0 International License](https://creativecommons.org/licenses/by-sa/4.0/).
378
379## Contributing
380
381Any change to behavior (including bugfixes) must come with a test.
382
383Patches that fail tests or reduce performance will be rejected.
384
385```
386# to run tests
387npm test
388
389# to re-generate test fixtures
390npm run test-regen
391
392# to benchmark against bash/zsh
393npm run bench
394
395# to profile javascript
396npm run prof
397```
398
399![](oh-my-glob.gif)
400