NAME

File::Random - Perl module for random selecting of a file

SYNOPSIS

  use File::Random qw/random_file/;
 
  my $fname  = random_file();

  my $fname2 = random_file(-dir => $dir);
  
  my $random_gif = random_file(-dir       => $dir,
                               -check     => qr/\.gif$/,
							   -recursive => 1);
							   
  my $no_exe     = random_file(-dir   => $dir,
                               -check => sub {! -x});
							   
  my @jokes_of_the_day = content_of_random_file(-dir => '/usr/lib/jokes');

DESCRIPTION

This module simplifies the routine job of selecting a random file. (As you can find at CGI scripts).

It's done, because it's boring (and errorprone), always to write something like

my @files = (<*.*>);
my $randf = $files[rand @files];

or

 opendir DIR, " ... " or die " ... ";
 my @files = grep {-f ...} (readdir DIR);
 closedir DIR;
 my $randf = $files[rand @files];

It also becomes very boring and very dangerous to write randomly selection for subdirectory searching with special check-routines.

FUNCTION random_file

Returns a randomly selected file(name) from the specified directory If the directory is empty, undef will be returned. There 3 options:

  my $file = random_file(
     -dir         => $dir, 
	 -check       => qr/.../, # or sub { .... }
	 -recursive   => 1        # or 0
  );

-dir

Specifies the directory where file has to come from.

Is the -dir option missing, a random file from the current directory will be used. That means '.' is the default for the -dir option.

-check

With the -check option you can either define a regex every filename has to follow, or a sub routine which gets the filename as argument.

Note, that -check doesn't accept anything else than a regexp or a subroutine. A string like '/.../' won't work. I still work on that.

The default is no checking (undef).

-recursive

Enables, that subdirectories are scanned for files, too. Every file, independent from its position in the file tree, has the same chance to be choosen. Now the relative path from the given subdirectory or the current directory of the randomly choosen file is included to the file name.

Every true value sets recursive behaviour on, every false value switches off. The default if false (undef).

Note, that I programmed the recursive routine very defendly (using File::Find). So switching -recursive on, slowers the program a bit :-)

FUNCTION content_of_random_file

Returns the content as an array of lines of a randomly selected random file. The lines aren't chomped.

This function has the same parameter and a similar behaviour to the random_file method.

EXPORT

None by default.

You can export the function random_file with use File::Random qw/random_file/;, use File::Random qw/content_of_random_file/ or with the more simple use File::Random qw/:all/;.

I didn't want to pollute namespaces as I could imagine, users write methods random_file to create a file with random content. If you think I'm paranoid, please tell me, then I'll take it into the export.

TODO

I think, I'll need to expand the options. Instead of only one directory, it should be possible to take a random file from some directories.

The -check option doesn't except a string looking like regexp. In future versions there should be the possibility of passing a string like '/..../' instead of the regexp qr/.../';

To create some aliases for the params is a good idea, too. I thought to make -d == -dir, -r == -recursive and -c == -check. (Only a lazy programmer is a good programmer).

So I want to make it possible to write:

my $fname = random_file( -dir => '...', -recursive => 1, -check     => '/\.html/' );

or even:

my $fname = random_file( -d => [$dir1, $dir2, $dir3, ...], -r => 1, -c => sub {-M < 7} );

The content_of_random_file method always returns an array, even in scalar context. I think it's useful that it returns the content as one string in scalar context. Also a -firstline option could be useful. Later -randomline option should be implemented, too.

Also speed could be improved, as I tried to write the code very readable, but wasted sometimes a little bit speed. (E.g. missing -check is translated to something like -check => sub{1}) As Functionality and Readability is more important than speed, I'll wait a little bit with speeding up :-)

Using unknown params should bring a warning. At the moment they are ignored.

The next thing, I'll implement is the scalar context of content_of_random_file.

Just have a look some hours later.

BUGS

Oh, I hope none. I still have more test lines than function code. However, it's still BETA code.

Well, but because I want some random data, it's a little bit hard to test. So a test could be wrong, allthough everything is O.K.. To avoid it, I let many tests run, so that the chances for misproofing should be < 0.0000000001% or so. Even it has the disadvantage that the tests need really long :-(

I'm not definitly sure whether my test routines runs on OS, with path seperators different of '/', like in Win with '\\'. Perhaps anybody can try it and tell me the result. [But remember Win* is definitly the greater bug.]

COPYRIGHT

This Program is free software. You can change or redistribute it under the same condition as Perl itself.

Copyright (c) 2002, Janek Schleicher, <bigj@kamelfreund.de>

AUTHOR

Janek Schleicher, <bigj@kamelfreund.de>

SEE ALSO

Tie::Pick Data::Random

cpanm File::Random

perl -MCPAN -e shell
install File::Random