Improve perlfaq5: How can I translate tildes (~) in a filename?

[brian_d_foy]

Want to be a Perl contributor? It's easy. Just fix this FAQ answer. Update: An by update, I mean rewrite it, not tell me how to rewrite it. :)

perlfaq5 has an answer for "How can I translate tildes (~) in a filename?". Looking at the code in the answer, I know it must break in a lot of places since it relies either on getpwnam or environment variables. Who can come up with a good answer that is more robust? I'm especially interested in any module-fu that can be added. You don't have to necessarily keep any of the existing answer. If you replace the answer, you get your name listed right next to it. :)

You can post your answer here or to the mailing list in perlfaq.

Here's the source (although you can get at the svn repository too by looking at the instructions in perlfaq:

=head2 How can I translate tildes (~) in a filename?
X<tilde> X<tilde expansion>

Use the E<lt>E<gt> (C<glob()>) operator, documented in L<perlfunc>.
Versions of Perl older than 5.6 require that you have a shell
installed that groks tildes.  Later versions of Perl have this feature
built in. The C<File::KGlob> module (available from CPAN) gives more
portable glob functionality.

Within Perl, you may use this directly:

    $filename =~ s{
      ^ ~             # find a leading tilde
      (               # save this in $1
          [^/]        # a non-slash character
                *     # repeated 0 or more times (0 means me)
      )
    }{
      $1
          ? (getpwnam($1))[7]
          : ( $ENV{HOME} || $ENV{LOGDIR} )
    }ex;

[halley]

I'm not sure what the actual beef might be with that solution. Here is an explanatory paragraph to introduce the FAQ, which may be all it needs.

On systems which follow Unix shell conventions, a user may
supply a pathname with the tilde (~) character to refer to
the user's home directory (e.g., <c>~/file</c>), or another
user's home directory (e.g., <c>~user/file</c>).  This works
in many Unix shells, but often does not work in scripts or
applications by default.  If your script would like to
support this convenient but optional notation, then the
built-in <c>getpwnam()</c> function and <c>%ENV</c> hash
are the best places to refer, or use the newer globbing
features that support them automatically.

On non-Unix systems, it might be broken. On non-Unix systems, though, the users are less likely to want to use Unix-isms like shell notations that scan /etc/passwd for data. Wrap the whole thing in a $^O if you like.

[Limbic~Region]

brian_d_foy,

• File::HomeDir
• File::Path::Expand
• File::Save::Home

Presumably the edge cases you described are solved in one or more of these modules. I haven't looked myself, but you did ask for module fu. The first one claims to work on any platform for instance.

Update: Added more modules as my search fu improved

[adamk]

Something like this would probably do it (I haven't actually tested it)

require File::HomeDir;
$filename =~ s/^~(\w+)/File::HomeDir->users_home("$1")/e;
$filename =~ s/^~/File::HomeDir->my_home/e;

[andreas1234567]

Works fine for me:

$ cat 691281.pl
use strict;
use warnings;
use File::HomeDir;
use Test::More;
plan tests => 4;

sub rep_tilde {
  my $filename = shift;
  $filename =~ s/^~(\w+)/File::HomeDir->users_home("$1")/e;
  $filename =~ s/^~/File::HomeDir->my_home/e;
  return $filename;
}

is(rep_tilde("~"), "/home/foo", "expect ~ -> /home/foo");
is(rep_tilde("~/"), "/home/foo/", "expect ~/ -> /home/foo/");
is(rep_tilde("~/.feeds"), "/home/foo/.feeds",
  "expect ~/.feeds -> /home/foo/.feeds");
is(
  rep_tilde("~/test/perl/perlmonks/691281/691281.pl"),
  "/home/foo/test/perl/perlmonks/691281/691281.pl",
  "expect ~/test/perl/perlmonks/691281/691281.pl "
    . "-> /home/foo/test/perl/perlmonks/691281/691281.pl"
);

__END__

$ prove 691281.pl
691281....ok
All tests successful.
Files=1, Tests=4,  0 wallclock secs ( 0.05 cusr +  0.00 csys =  0.05 C
+PU)

[pc88mxer]

What's interesting is that the logic used in bsd_glob.c seems to be the reverse: first $HOME is checked and then getpwuid is called.