xref: /netbsd-src/lib/libc/gen/glob.3 (revision bbde328be4e75ea9ad02e9715ea13ca54b797ada)
1.\"	$NetBSD: glob.3,v 1.36 2010/03/22 19:30:54 joerg Exp $
2.\"
3.\" Copyright (c) 1989, 1991, 1993, 1994
4.\"	The Regents of the University of California.  All rights reserved.
5.\"
6.\" This code is derived from software contributed to Berkeley by
7.\" Guido van Rossum.
8.\" Redistribution and use in source and binary forms, with or without
9.\" modification, are permitted provided that the following conditions
10.\" are met:
11.\" 1. Redistributions of source code must retain the above copyright
12.\"    notice, this list of conditions and the following disclaimer.
13.\" 2. Redistributions in binary form must reproduce the above copyright
14.\"    notice, this list of conditions and the following disclaimer in the
15.\"    documentation and/or other materials provided with the distribution.
16.\" 3. Neither the name of the University nor the names of its contributors
17.\"    may be used to endorse or promote products derived from this software
18.\"    without specific prior written permission.
19.\"
20.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30.\" SUCH DAMAGE.
31.\"
32.\"     @(#)glob.3	8.3 (Berkeley) 4/16/94
33.\"
34.Dd April 8, 2009
35.Dt GLOB 3
36.Os
37.Sh NAME
38.Nm glob ,
39.Nm globfree ,
40.Nm glob_pattern_p
41.Nd generate pathnames matching a pattern
42.Sh LIBRARY
43.Lb libc
44.Sh SYNOPSIS
45.In glob.h
46.Ft int
47.Fn glob "const char * restrict pattern" "int flags" "const int (*errfunc)(const char *, int)" "glob_t * restrict pglob"
48.Ft void
49.Fn globfree "glob_t *pglob"
50.Ft int
51.Fn glob_pattern_p "const char *pattern" "int quote"
52.Sh DESCRIPTION
53The
54.Fn glob
55function
56is a pathname generator that implements the rules for file name pattern
57matching used by the shell.
58.Pp
59The include file
60.Pa glob.h
61defines the structure type
62.Fa glob_t ,
63which contains at least the following fields:
64.Bd -literal
65typedef struct {
66	size_t gl_pathc;	/* count of total paths so far */
67	size_t gl_matchc;	/* count of paths matching pattern */
68	size_t gl_offs;		/* reserved at beginning of gl_pathv */
69	int gl_flags;		/* returned flags */
70	char **gl_pathv;	/* list of paths matching pattern */
71} glob_t;
72.Ed
73.Pp
74The argument
75.Fa pattern
76is a pointer to a pathname pattern to be expanded.
77The
78.Fn glob
79argument
80matches all accessible pathnames against the pattern and creates
81a list of the pathnames that match.
82In order to have access to a pathname,
83.Fn glob
84requires search permission on every component of a path except the last
85and read permission on each directory of any filename component of
86.Fa pattern
87that contains any of the special characters
88.Ql * ,
89.Ql \&?
90or
91.Ql \&[ .
92.Pp
93The
94.Fn glob
95argument
96stores the number of matched pathnames into the
97.Fa gl_pathc
98field, and a pointer to a list of pointers to pathnames into the
99.Fa gl_pathv
100field.
101The first pointer after the last pathname is
102.Dv NULL .
103If the pattern does not match any pathnames, the returned number of
104matched paths is set to zero.
105.Pp
106It is the caller's responsibility to create the structure pointed to by
107.Fa pglob .
108The
109.Fn glob
110function allocates other space as needed, including the memory pointed
111to by
112.Fa gl_pathv .
113.Pp
114The argument
115.Fa flags
116is used to modify the behavior of
117.Fn glob .
118The value of
119.Fa flags
120is the bitwise inclusive
121.Tn OR
122of any of the following
123values defined in
124.Pa glob.h :
125.Bl -tag -width GLOB_ALTDIRFUNC
126.It Dv GLOB_APPEND
127Append pathnames generated to the ones from a previous call (or calls)
128to
129.Fn glob .
130The value of
131.Fa gl_pathc
132will be the total matches found by this call and the previous call(s).
133The pathnames are appended to, not merged with the pathnames returned by
134the previous call(s).
135Between calls, the caller must not change the setting of the
136.Dv GLOB_DOOFFS
137flag, nor change the value of
138.Fa gl_offs
139when
140.Dv GLOB_DOOFFS
141is set, nor (obviously) call
142.Fn globfree
143for
144.Fa pglob .
145.It Dv GLOB_DOOFFS
146Make use of the
147.Fa gl_offs
148field.
149If this flag is set,
150.Fa gl_offs
151is used to specify how many
152.Dv NULL
153pointers to prepend to the beginning
154of the
155.Fa gl_pathv
156field.
157In other words,
158.Fa gl_pathv
159will point to
160.Fa gl_offs
161.Dv NULL
162pointers,
163followed by
164.Fa gl_pathc
165pathname pointers, followed by a
166.Dv NULL
167pointer.
168.It Dv GLOB_ERR
169Causes
170.Fn glob
171to return when it encounters a directory that it cannot open or read.
172Ordinarily,
173.Fn glob
174continues to find matches.
175.It Dv GLOB_MARK
176Each pathname that is a directory that matches
177.Fa pattern
178has a slash
179appended.
180.It Dv GLOB_NOCHECK
181If
182.Fa pattern
183does not match any pathname, then
184.Fn glob
185returns a list
186consisting of only
187.Fa pattern ,
188with the number of total pathnames set to 1, and the number of matched
189pathnames set to 0.
190.It Dv GLOB_NOSORT
191By default, the pathnames are sorted in ascending
192.Tn ASCII
193order;
194this flag prevents that sorting (speeding up
195.Fn glob ) .
196.El
197.Pp
198The following values may also be included in
199.Fa flags ,
200however, they are non-standard extensions to
201.St -p1003.2 .
202.Bl -tag -width GLOB_ALTDIRFUNC
203.It Dv GLOB_ALTDIRFUNC
204The following additional fields in the pglob structure have been
205initialized with alternate functions for glob to use to open, read,
206and close directories and to get stat information on names found
207in those directories.
208.Bd -literal
209	void *(*gl_opendir)(const char * name);
210	struct dirent *(*gl_readdir)(void *);
211	void (*gl_closedir)(void *);
212	int (*gl_lstat)(const char *name, struct stat *st);
213	int (*gl_stat)(const char *name, struct stat *st);
214.Ed
215.Pp
216This extension is provided to allow programs such as
217.Xr restore 8
218to provide globbing from directories stored on tape.
219.It Dv GLOB_BRACE
220Pre-process the pattern string to expand
221.Ql {pat,pat,...}
222strings like
223.Xr csh 1 .
224The pattern
225.Ql {}
226is left unexpanded for historical reasons
227.Po
228.Xr csh 1
229does the same thing to ease typing of
230.Xr find 1
231patterns
232.Pc .
233.It Dv GLOB_MAGCHAR
234Set by the
235.Fn glob
236function if the pattern included globbing characters.
237See the description of the usage of the
238.Fa gl_matchc
239structure member for more details.
240.It Dv GLOB_NOMAGIC
241Is the same as
242.Dv GLOB_NOCHECK
243but it only appends the
244.Fa pattern
245if it does not contain any of the special characters ``*'', ``?'' or ``[''.
246.Dv GLOB_NOMAGIC
247is provided to simplify implementing the historic
248.Xr csh 1
249globbing behavior and should probably not be used anywhere else.
250.It Dv GLOB_NOESCAPE
251Disable the use of the backslash
252.Pq Ql \e
253character for quoting.
254.It Dv GLOB_TILDE
255Expand patterns that start with
256.Ql ~
257to user name home directories.
258.It Dv GLOB_LIMIT
259Limit the amount of memory used by matches to
260.Li ARG_MAX .
261This option should be set for programs that can be coerced to a denial of
262service attack via patterns that expand to a very large number of matches,
263such as a long string of
264.Li */../*/..
265.It Dv GLOB_PERIOD
266Allow metacharacters to match a leading period in a filename.
267.It Dv GLOB_NO_DOTDIRS
268Hide
269.Sq Li \&.
270and
271.Sq Li \&..
272from metacharacter matches, regardless of whether
273.Dv GLOB_PERIOD
274is set and whether the pattern component begins with a literal period.
275.El
276.Pp
277If, during the search, a directory is encountered that cannot be opened
278or read and
279.Fa errfunc
280is
281.Pf non- Dv NULL ,
282.Fn glob
283calls
284.Fa (*errfunc)(path, errno) .
285This may be unintuitive: a pattern like
286.Ql */Makefile
287will try to
288.Xr stat 2
289.Ql foo/Makefile
290even if
291.Ql foo
292is not a directory, resulting in a
293call to
294.Fa errfunc .
295The error routine can suppress this action by testing for
296.Dv ENOENT
297and
298.Dv ENOTDIR ;
299however, the
300.Dv GLOB_ERR
301flag will still cause an immediate
302return when this happens.
303.Pp
304If
305.Fa errfunc
306returns non-zero,
307.Fn glob
308stops the scan and returns
309.Dv GLOB_ABORTED
310after setting
311.Fa gl_pathc
312and
313.Fa gl_pathv
314to reflect any paths already matched.
315This also happens if an error is encountered and
316.Dv GLOB_ERR
317is set in
318.Fa flags ,
319regardless of the return value of
320.Fa errfunc ,
321if called.
322If
323.Dv GLOB_ERR
324is not set and either
325.Fa errfunc
326is
327.Dv NULL
328or
329.Fa errfunc
330returns zero, the error is ignored.
331.Pp
332The
333.Fn globfree
334function frees any space associated with
335.Fa pglob
336from a previous call(s) to
337.Fn glob .
338.Pp
339The
340.Fn glob_pattern_p
341returns
342.Dv 1
343if the
344.Fa pattern
345has any special characters that
346.Fn glob
347will interpret and
348.Dv 0
349otherwise.
350If the
351.Fa quote
352argument is non-zero, then backslash quoted characters are ignored.
353.Pp
354The historical
355.Dv GLOB_QUOTE
356flag is no longer supported.
357Per
358.St -p1003.2-92 ,
359backslash escaping of special characters is the default behaviour;
360it may be disabled by specifying the
361.Dv GLOB_NOESCAPE
362flag.
363.Sh RETURN VALUES
364On successful completion,
365.Fn glob
366returns zero.
367In addition the fields of
368.Fa pglob
369contain the values described below:
370.Bl -tag -width GLOB_NOCHECK
371.It Fa gl_pathc
372contains the total number of matched pathnames so far.
373This includes other matches from previous invocations of
374.Fn glob
375if
376.Dv GLOB_APPEND
377was specified.
378.It Fa gl_matchc
379contains the number of matched pathnames in the current invocation of
380.Fn glob .
381.It Fa gl_flags
382contains a copy of the
383.Fa flags
384parameter with the bit
385.Dv GLOB_MAGCHAR
386set if
387.Fa pattern
388contained any of the special characters ``*'', ``?'' or ``['', cleared
389if not.
390.It Fa gl_pathv
391contains a pointer to a
392.Dv NULL Ns -terminated
393list of matched pathnames.
394However, if
395.Fa gl_pathc
396is zero, the contents of
397.Fa gl_pathv
398are undefined.
399.El
400.Pp
401If
402.Fn glob
403terminates due to an error, it sets
404.Va errno
405and returns one of the following non-zero constants, which are defined
406in the include file
407.In glob.h :
408.Bl -tag -width GLOB_ABORTEDXXX
409.It Dv GLOB_ABORTED
410The scan was stopped because an error was encountered and either
411.Dv GLOB_ERR
412was set or
413.Fa (*errfunc)()
414returned non-zero.
415.It Dv GLOB_NOMATCH
416The pattern does not match any existing pathname, and
417.Dv GLOB_NOCHECK
418was not set in
419.Dv flags .
420.It Dv GLOB_NOSPACE
421An attempt to allocate memory failed, or if
422.Va errno
423was 0
424.Li GLOB_LIMIT
425was specified in the flags and
426.Li ARG_MAX
427patterns were matched.
428.El
429.Pp
430The historical
431.Dv GLOB_ABEND
432return constant is no longer supported.
433Portable applications should use the
434.Dv GLOB_ABORTED
435constant instead.
436.Pp
437The arguments
438.Fa pglob\-\*[Gt]gl_pathc
439and
440.Fa pglob\-\*[Gt]gl_pathv
441are still set as specified above.
442.Sh ENVIRONMENT
443.Bl -tag -width HOME -compact
444.It Ev HOME
445If defined, used as the home directory of the current user in
446tilde expansions.
447.El
448.Sh EXAMPLES
449A rough equivalent of
450.Ql "ls -l *.c *.h"
451can be obtained with the
452following code:
453.Bd -literal -offset indent
454glob_t g;
455
456g.gl_offs = 2;
457glob("*.c", GLOB_DOOFFS, NULL, \*[Am]g);
458glob("*.h", GLOB_DOOFFS | GLOB_APPEND, NULL, \*[Am]g);
459g.gl_pathv[0] = "ls";
460g.gl_pathv[1] = "-l";
461execvp("ls", g.gl_pathv);
462.Ed
463.Sh SEE ALSO
464.Xr sh 1 ,
465.Xr fnmatch 3 ,
466.Xr regexp 3
467.Sh STANDARDS
468The
469.Fn glob
470function is expected to be
471.St -p1003.2
472compatible with the exception
473that the flags
474.Dv GLOB_ALTDIRFUNC ,
475.Dv GLOB_BRACE ,
476.Dv GLOB_MAGCHAR ,
477.Dv GLOB_NOMAGIC ,
478.Dv GLOB_TILDE ,
479and
480.Dv GLOB_LIMIT
481and the fields
482.Fa gl_matchc
483and
484.Fa gl_flags
485should not be used by applications striving for strict
486.Tn POSIX
487conformance.
488.Sh HISTORY
489The
490.Fn glob
491and
492.Fn globfree
493functions first appeared in
494.Bx 4.4 .
495The
496.Fn glob_pattern_p
497function is modelled after the one found in glibc.
498.Sh BUGS
499Patterns longer than
500.Dv MAXPATHLEN
501may cause unchecked errors.
502.Pp
503The
504.Fn glob
505function may fail and set
506.Va errno
507for any of the errors specified for the library routines
508.Xr stat 2 ,
509.Xr closedir 3 ,
510.Xr opendir 3 ,
511.Xr readdir 3 ,
512.Xr malloc 3 ,
513and
514.Xr free 3 .
515