This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Document File::Glob’s :bsd_glob tag
[perl5.git] / ext / File-Glob / Glob.pm
CommitLineData
72b16652
GS
1package File::Glob;
2
3use strict;
7f39e0ae 4our($VERSION, @ISA, @EXPORT_OK, @EXPORT_FAIL, %EXPORT_TAGS, $DEFAULT_FLAGS);
72b16652 5
da4061d3 6require XSLoader;
22bc907a 7use feature 'switch';
72b16652 8
72f7b9a1 9@ISA = qw(Exporter);
72b16652 10
00c80938
GS
11# NOTE: The glob() export is only here for compatibility with 5.6.0.
12# csh_glob() should not be used directly, unless you know what you're doing.
13
72b16652
GS
14%EXPORT_TAGS = (
15 'glob' => [ qw(
16 GLOB_ABEND
2d5e9e5d 17 GLOB_ALPHASORT
72b16652
GS
18 GLOB_ALTDIRFUNC
19 GLOB_BRACE
220398a0 20 GLOB_CSH
72b16652
GS
21 GLOB_ERR
22 GLOB_ERROR
b8ef571c 23 GLOB_LIMIT
72b16652 24 GLOB_MARK
220398a0 25 GLOB_NOCASE
72b16652
GS
26 GLOB_NOCHECK
27 GLOB_NOMAGIC
28 GLOB_NOSORT
29 GLOB_NOSPACE
30 GLOB_QUOTE
31 GLOB_TILDE
00c80938 32 bsd_glob
f4cbf990 33 glob
72b16652
GS
34 ) ],
35);
f4cbf990
FC
36$EXPORT_TAGS{bsd_glob} = [@{$EXPORT_TAGS{glob}}];
37pop @{$EXPORT_TAGS{bsd_glob}}; # no "glob"
72b16652 38
aa0c903b
NC
39@EXPORT_OK = (@{$EXPORT_TAGS{'glob'}}, 'csh_glob');
40
0b0e6d70 41$VERSION = '1.14';
220398a0
PM
42
43sub import {
7d3fb230 44 require Exporter;
df5a3819
NC
45 local $Exporter::ExportLevel = $Exporter::ExportLevel + 1;
46 Exporter::import(grep {
47 my $passthrough;
48 given ($_) {
22bc907a
NC
49 $DEFAULT_FLAGS &= ~GLOB_NOCASE() when ':case';
50 $DEFAULT_FLAGS |= GLOB_NOCASE() when ':nocase';
51 when (':globally') {
52 no warnings 'redefine';
220398a0
PM
53 *CORE::GLOBAL::glob = \&File::Glob::csh_glob;
54 }
f4cbf990
FC
55 if ($_ eq ':bsd_glob') {
56 no strict; *{caller."::glob"} = \&bsd_glob_override;
57 }
df5a3819 58 $passthrough = 1;
220398a0 59 }
df5a3819
NC
60 $passthrough;
61 } @_);
72b16652
GS
62}
63
da4061d3 64XSLoader::load();
72b16652 65
220398a0 66$DEFAULT_FLAGS = GLOB_CSH();
862f843b 67if ($^O =~ /^(?:MSWin32|VMS|os2|dos|riscos)$/) {
220398a0
PM
68 $DEFAULT_FLAGS |= GLOB_NOCASE();
69}
70
00c80938
GS
71# File::Glob::glob() is deprecated because its prototype is different from
72# CORE::glob() (use bsd_glob() instead)
73sub glob {
e0e8a4dc 74 splice @_, 1; # don't pass PL_glob_index as flags!
00c80938
GS
75 goto &bsd_glob;
76}
77
72b16652
GS
781;
79__END__
80
81=head1 NAME
82
83File::Glob - Perl extension for BSD glob routine
84
85=head1 SYNOPSIS
86
5144542d 87 use File::Glob ':bsd_glob';
9d70ac1b 88
00c80938
GS
89 @list = bsd_glob('*.[ch]');
90 $homedir = bsd_glob('~gnat', GLOB_TILDE | GLOB_ERR);
9d70ac1b 91
72b16652
GS
92 if (GLOB_ERROR) {
93 # an error occurred reading $homedir
94 }
95
00c80938 96 ## override the core glob (CORE::glob() does this automatically
11fe14b1 97 ## by default anyway, since v5.6.0)
220398a0 98 use File::Glob ':globally';
6bd08436 99 my @sources = <*.{c,h,y}>;
220398a0
PM
100
101 ## override the core glob, forcing case sensitivity
102 use File::Glob qw(:globally :case);
6bd08436 103 my @sources = <*.{c,h,y}>;
220398a0
PM
104
105 ## override the core glob forcing case insensitivity
106 use File::Glob qw(:globally :nocase);
6bd08436 107 my @sources = <*.{c,h,y}>;
9d70ac1b 108
6bd08436
SS
109 ## glob on all files in home directory
110 use File::Glob ':globally';
111 my @sources = <~gnat/*>;
72b16652
GS
112
113=head1 DESCRIPTION
114
9d70ac1b
RGS
115The glob angle-bracket operator C<< <> >> is a pathname generator that
116implements the rules for file name pattern matching used by Unix-like shells
117such as the Bourne shell or C shell.
6bd08436 118
00c80938
GS
119File::Glob::bsd_glob() implements the FreeBSD glob(3) routine, which is
120a superset of the POSIX glob() (described in IEEE Std 1003.2 "POSIX.2").
121bsd_glob() takes a mandatory C<pattern> argument, and an optional
72b16652
GS
122C<flags> argument, and returns a list of filenames matching the
123pattern, with interpretation of the pattern modified by the C<flags>
00c80938
GS
124variable.
125
126Since v5.6.0, Perl's CORE::glob() is implemented in terms of bsd_glob().
127Note that they don't share the same prototype--CORE::glob() only accepts
128a single argument. Due to historical reasons, CORE::glob() will also
129split its argument on whitespace, treating it as multiple patterns,
5144542d
FC
130whereas bsd_glob() considers them as one pattern. But see C<:bsd_glob>
131under L</EXPORTS>, below.
00c80938 132
6bd08436
SS
133=head2 META CHARACTERS
134
9d70ac1b
RGS
135 \ Quote the next metacharacter
136 [] Character class
137 {} Multiple pattern
138 * Match any string of characters
139 ? Match any single character
140 ~ User name home directory
141
142The metanotation C<a{b,c,d}e> is a shorthand for C<abe ace ade>. Left to
143right order is preserved, with results of matches being sorted separately
144at a low level to preserve this order. As a special case C<{>, C<}>, and
145C<{}> are passed undisturbed.
6bd08436 146
5144542d
FC
147=head2 EXPORTS
148
149The C<:bsd_glob> export tag exports bsd_glob() and the constants listed
150below. It also overrides glob() in the calling package with one that
151behaves like bsd_glob() with regard to spaces (the space is treated as part
152of a file name), but supports iteration in scalar context; i.e., it
153preserves the core function's feature of returning the next item each time
154it is called.
155
156The C<:glob> tag, now discouraged, is the old version of C<:bsd_glob>. It
157exports the same constants and functions, but its glob() override does not
158support iteration; it returns the last file name in scalar context. That
159means this will loop forever:
160
161 use File::Glob ':glob';
162 while (my $file = <* copy.txt>) {
163 ...
164 }
165
166The bsd_glob() function and the constants below can be exported
167individually.
168
6bd08436
SS
169=head2 POSIX FLAGS
170
00c80938 171The POSIX defined flags for bsd_glob() are:
72b16652
GS
172
173=over 4
174
175=item C<GLOB_ERR>
176
00c80938
GS
177Force bsd_glob() to return an error when it encounters a directory it
178cannot open or read. Ordinarily bsd_glob() continues to find matches.
72b16652 179
b8ef571c
JH
180=item C<GLOB_LIMIT>
181
182Make bsd_glob() return an error (GLOB_NOSPACE) when the pattern expands
183to a size bigger than the system constant C<ARG_MAX> (usually found in
184limits.h). If your system does not define this constant, bsd_glob() uses
185C<sysconf(_SC_ARG_MAX)> or C<_POSIX_ARG_MAX> where available (in that
186order). You can inspect these values using the standard C<POSIX>
187extension.
188
72b16652
GS
189=item C<GLOB_MARK>
190
191Each pathname that is a directory that matches the pattern has a slash
192appended.
193
220398a0
PM
194=item C<GLOB_NOCASE>
195
196By default, file names are assumed to be case sensitive; this flag
00c80938 197makes bsd_glob() treat case differences as not significant.
220398a0 198
72b16652
GS
199=item C<GLOB_NOCHECK>
200
00c80938 201If the pattern does not match any pathname, then bsd_glob() returns a list
72b16652
GS
202consisting of only the pattern. If C<GLOB_QUOTE> is set, its effect
203is present in the pattern returned.
204
205=item C<GLOB_NOSORT>
206
207By default, the pathnames are sorted in ascending ASCII order; this
00c80938 208flag prevents that sorting (speeding up bsd_glob()).
72b16652
GS
209
210=back
211
212The FreeBSD extensions to the POSIX standard are the following flags:
213
214=over 4
215
216=item C<GLOB_BRACE>
217
a45bd81d 218Pre-process the string to expand C<{pat,pat,...}> strings like csh(1).
72b16652
GS
219The pattern '{}' is left unexpanded for historical reasons (and csh(1)
220does the same thing to ease typing of find(1) patterns).
221
222=item C<GLOB_NOMAGIC>
223
224Same as C<GLOB_NOCHECK> but it only returns the pattern if it does not
225contain any of the special characters "*", "?" or "[". C<NOMAGIC> is
226provided to simplify implementing the historic csh(1) globbing
227behaviour and should probably not be used anywhere else.
228
229=item C<GLOB_QUOTE>
230
231Use the backslash ('\') character for quoting: every occurrence of a
232backslash followed by a character in the pattern is replaced by that
233character, avoiding any special interpretation of the character.
220398a0 234(But see below for exceptions on DOSISH systems).
72b16652
GS
235
236=item C<GLOB_TILDE>
237
238Expand patterns that start with '~' to user name home directories.
239
240=item C<GLOB_CSH>
241
242For convenience, C<GLOB_CSH> is a synonym for
2d5e9e5d 243C<GLOB_BRACE | GLOB_NOMAGIC | GLOB_QUOTE | GLOB_TILDE | GLOB_ALPHASORT>.
72b16652
GS
244
245=back
246
247The POSIX provided C<GLOB_APPEND>, C<GLOB_DOOFFS>, and the FreeBSD
248extensions C<GLOB_ALTDIRFUNC>, and C<GLOB_MAGCHAR> flags have not been
249implemented in the Perl version because they involve more complex
250interaction with the underlying C structures.
251
2d5e9e5d
JH
252The following flag has been added in the Perl implementation for
253csh compatibility:
254
255=over 4
256
257=item C<GLOB_ALPHASORT>
258
259If C<GLOB_NOSORT> is not in effect, sort filenames is alphabetical
260order (case does not matter) rather than in ASCII order.
261
262=back
263
72b16652
GS
264=head1 DIAGNOSTICS
265
00c80938 266bsd_glob() returns a list of matching paths, possibly zero length. If an
72b16652
GS
267error occurred, &File::Glob::GLOB_ERROR will be non-zero and C<$!> will be
268set. &File::Glob::GLOB_ERROR is guaranteed to be zero if no error occurred,
269or one of the following values otherwise:
270
271=over 4
272
273=item C<GLOB_NOSPACE>
274
275An attempt to allocate memory failed.
276
277=item C<GLOB_ABEND>
278
279The glob was stopped because an error was encountered.
280
281=back
282
00c80938
GS
283In the case where bsd_glob() has found some matching paths, but is
284interrupted by an error, it will return a list of filenames B<and>
72b16652
GS
285set &File::Glob::ERROR.
286
00c80938
GS
287Note that bsd_glob() deviates from POSIX and FreeBSD glob(3) behaviour
288by not considering C<ENOENT> and C<ENOTDIR> as errors - bsd_glob() will
72b16652
GS
289continue processing despite those errors, unless the C<GLOB_ERR> flag is
290set.
291
292Be aware that all filenames returned from File::Glob are tainted.
293
294=head1 NOTES
295
296=over 4
297
298=item *
299
9d70ac1b
RGS
300If you want to use multiple patterns, e.g. C<bsd_glob("a* b*")>, you should
301probably throw them in a set as in C<bsd_glob("{a*,b*}")>. This is because
150b260b
GS
302the argument to bsd_glob() isn't subjected to parsing by the C shell.
303Remember that you can use a backslash to escape things.
72b16652
GS
304
305=item *
306
220398a0
PM
307On DOSISH systems, backslash is a valid directory separator character.
308In this case, use of backslash as a quoting character (via GLOB_QUOTE)
309interferes with the use of backslash as a directory separator. The
310best (simplest, most portable) solution is to use forward slashes for
311directory separators, and backslashes for quoting. However, this does
312not match "normal practice" on these systems. As a concession to user
313expectation, therefore, backslashes (under GLOB_QUOTE) only quote the
314glob metacharacters '[', ']', '{', '}', '-', '~', and backslash itself.
315All other backslashes are passed through unchanged.
316
317=item *
318
72b16652
GS
319Win32 users should use the real slash. If you really want to use
320backslashes, consider using Sarathy's File::DosGlob, which comes with
321the standard Perl distribution.
322
7369a524
CN
323=item *
324
325Mac OS (Classic) users should note a few differences. Since
326Mac OS is not Unix, when the glob code encounters a tilde glob (e.g.
be708cc0 327~user) and the C<GLOB_TILDE> flag is used, it simply returns that
7369a524
CN
328pattern without doing any expansion.
329
330Glob on Mac OS is case-insensitive by default (if you don't use any
331flags). If you specify any flags at all and still want glob
332to be case-insensitive, you must include C<GLOB_NOCASE> in the flags.
333
334The path separator is ':' (aka colon), not '/' (aka slash). Mac OS users
335should be careful about specifying relative pathnames. While a full path
336always begins with a volume name, a relative pathname should always
337begin with a ':'. If specifying a volume name only, a trailing ':' is
338required.
339
be708cc0
JH
340The specification of pathnames in glob patterns adheres to the usual Mac
341OS conventions: The path separator is a colon ':', not a slash '/'. A
342full path always begins with a volume name. A relative pathname on Mac
343OS must always begin with a ':', except when specifying a file or
344directory name in the current working directory, where the leading colon
345is optional. If specifying a volume name only, a trailing ':' is
346required. Due to these rules, a glob like E<lt>*:E<gt> will find all
347mounted volumes, while a glob like E<lt>*E<gt> or E<lt>:*E<gt> will find
348all files and directories in the current directory.
349
350Note that updirs in the glob pattern are resolved before the matching begins,
351i.e. a pattern like "*HD:t?p::a*" will be matched as "*HD:a*". Note also,
352that a single trailing ':' in the pattern is ignored (unless it's a volume
353name pattern like "*HD:"), i.e. a glob like E<lt>:*:E<gt> will find both
354directories I<and> files (and not, as one might expect, only directories).
355You can, however, use the C<GLOB_MARK> flag to distinguish (without a file
356test) directory names from file names.
357
358If the C<GLOB_MARK> flag is set, all directory paths will have a ':' appended.
359Since a directory like 'lib:' is I<not> a valid I<relative> path on Mac OS,
360both a leading and a trailing colon will be added, when the directory name in
361question doesn't contain any colons (e.g. 'lib' becomes ':lib:').
362
a45bd81d
GS
363=back
364
6bd08436
SS
365=head1 SEE ALSO
366
367L<perlfunc/glob>, glob(3)
368
72b16652
GS
369=head1 AUTHOR
370
0e950d83 371The Perl interface was written by Nathan Torkington E<lt>gnat@frii.comE<gt>,
72b16652 372and is released under the artistic license. Further modifications were
7369a524
CN
373made by Greg Bacon E<lt>gbacon@cs.uah.eduE<gt>, Gurusamy Sarathy
374E<lt>gsar@activestate.comE<gt>, and Thomas Wegner
375E<lt>wegner_thomas@yahoo.comE<gt>. The C glob code has the
72b16652
GS
376following copyright:
377
0e950d83
GS
378 Copyright (c) 1989, 1993 The Regents of the University of California.
379 All rights reserved.
3cb6de81 380
0e950d83
GS
381 This code is derived from software contributed to Berkeley by
382 Guido van Rossum.
383
384 Redistribution and use in source and binary forms, with or without
385 modification, are permitted provided that the following conditions
386 are met:
387
388 1. Redistributions of source code must retain the above copyright
389 notice, this list of conditions and the following disclaimer.
390 2. Redistributions in binary form must reproduce the above copyright
391 notice, this list of conditions and the following disclaimer in the
392 documentation and/or other materials provided with the distribution.
393 3. Neither the name of the University nor the names of its contributors
394 may be used to endorse or promote products derived from this software
395 without specific prior written permission.
396
397 THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
398 ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
399 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
400 ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
401 FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
402 DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
403 OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
404 HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
405 LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
406 OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
407 SUCH DAMAGE.
72b16652
GS
408
409=cut