File::Tee - replicate data sent to a Perl stream


  use File::Tee qw(tee);

  # simple usage:
  tee(STDOUT, '>', 'stdout.txt');

  print "hello world\n";
  system "ls";

  # advanced usage:
  my $pid = tee STDERR, { prefix => "err[$$]: ", reopen => 'my.log'};

  print STDERR "foo\n";
  system("cat /bad/path");


This module is able to replicate data written to a Perl stream into another streams. It is the Perl equivalent of the shell utility tee(1).

It is implemeted around fork, creating a new process for every tee'ed stream. That way, there are no problems handling the output generated by external programs run with system or by XS modules that don't go through perlio.


The following function can be imported from this module:

tee $fh, $target, ...

redirects a copy of the data written to $fh to one or several files or streams.

$target, ... is a list of target streams specifications that can be:

  • file names with optional mode specifications:

      tee STDOUT, '>> /tmp/out', '>> /tmp/out2';
      tee STDOUT, '>>', '/tmp/out', '/tmp/out2';

    If the mode specification is a separate argument, it will affect all the file names following and not just the nearest one.

    If mode |- is used as a separate argument, the rest of the arguments are slurped as arguments for the pipe command:

       tee STDERR, '|-', 'grep', '-i', 'error';
       tee STDERR, '| grep -i error'; # equivalent

    Valid modes are >, >>, >&, >>& and |-. The default mode is >>.

    File handles can also be used as targets:

       open my $target1, '>>', '/foo/bar';
       tee STDOUT, $target1, $target2, ...;

    Finally, code references can also be used as targets. The callback will be invoked for every line written to the tee'ed stream with the data in $_. It has to return a true value on success or false if some error happens. Also, note that the callback will be called from a different process.

  • hash references describing the targets

    For instance:

      tee STDOUT, { mode => '>>', open => '/tmp/foo', lock => 1};

    will copy the data sent to STDOUT to /tmp/foo.

    The attributes that can be included inside the hash are:

    open => $file_name
    reopen => $file_name

    sets the target file or stream. It can contain a mode specification and also be an array. For instance:

      tee STDOUT, { open => '>> /tmp/out' };
      tee STDOUT, { reopen => ['>>', '/tmp/out2'] };
      tee STDOUT, { open => '| grep foo > /tmp/out' };

    If reopen is used, the file or stream is reopen for every write operation. The mode will be forced to append after the first write.

    mode => $mode

    Alternative way to specify the mode to open the target file or stream

    lock => $bool

    When true, an exclusive lock is obtained on the target file before writing to it.

    prefix => $txt

    Some text to be prepended to every line sent to the target file.

    For instance:

      tee STDOUT, { prefix => 'OUT: ', lock => 1, mode => '>>', open => '/tmp/out.txt' };
      tee STDERR, { prefix => 'ERR: ', lock => 1, mode => '>>', open => '/tmp/out.txt' };
    preprocess => sub { ... }

    A callback function that can modify the data before it gets sent to the target file.

    For instance:

      sub hexdump {
        my $data = shift;
        my @out;
        while ($data =~ /(.{1,32})/smg) {
            my $line=$1;
            my @c= (( map { sprintf "%02x",$_ } unpack('C*', $line)),
                    (("  ") x 32))[0..31];
            $line=~s/(.)/ my $c=$1; unpack("c",$c)>=32 ? $c : '.' /egms;
            push @out, join(" ", @c, '|', $line), "\n";
        join('', @out);
      tee BINFH, { preprocess => \&hexdump, open => '/tmp/hexout'};
    autoflush => $bool

    Sets autoflush mode for the target streams. Default is on.

    ignore_errors => $bool

    By default, when writting to the targets, any error will close the tee'ed handle. This option allows to change that behaviour.

    process => sub { ... }

    the callback will be called for every line read (see using code references as targets discussion above). This option can not be used at the same time as most other options (open, reopen, lock, autoflush, etc.).

    begin => sub { ... }
    end => sub { ... }

    Those functions are called on the forked process before the first write and when closing the handle respectively.

    For instance:

      my @capture;
      tee STDERR, { process => sub { push @capture, $_ },
                    end => sub { send_mail '', 'stderr capture', "@capture" } };

The funcion returns the PID for the newly created process.

Inside the tee pipe process created, data is readed honouring the input record separator $/.

You could also want to set the tee'ed stream in autoflush mode:

  open $fh, ...;

  my $oldsel = select $fh;
  $| = 1;
  select $fh;

  tee $fh, "> /tmp/log";


Does not work on Windows (patches welcome).

Send bug reports by email or via the CPAN RT web.



IO::Tee is a similar module implemented around tied file handles. Tee allows to launch external processes capturing their output to some files. IO::CaptureOutput allows to capture the output generated from a child process or a subroutine.


Copyright (C) 2007, 2008, 2010, 2011 by Salvador Fandiño (

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.8 or, at your option, any later version of Perl 5 you may have available.