IO::Pager - Select a pager (possibly perl-based) & pipe it text if a TTY


  # Select an appropriate pager and set the PAGER environment variable
  use IO::Pager;

  # TIMTOWTDI Object-oriented
    # open()                           # Use all the defaults.
    my $object = new IO::Pager;

    # open FILEHANDLE                  # Unbuffered is default subclass
    my $object = new IO::Pager *STDOUT;

    # open FILEHANDLE,EXPR             # Specify subclass
    my $object = new IO::Pager *STDOUT,  'Unbuffered';

    # Direct subclass instantiation    # FH is optional
    use IO::Pager::Unbuffered;
    my $object = new IO::Pager::Unbuffered  *STDOUT;

    $object->print("OO shiny...\n") while 1;
    print "Some other text sent to STODUT, perhaps from a foreign routine."

    # $object passes out of scope and filehandle is automagically closed

  # TIMTOWTDI Procedural
    # open FILEHANDLE                    # Unbuffered is default subclass
    my $token = IO::Pager::open *STDOUT;

    # open FILEHANDLE,EXPR               # Specify subclass
    my $token = IO::Pager::open *STDOUT,  'Unbuffered';

    # open FILEHANDLE,MODE,EXPR          # En lieu of a separate binmode()
    my $token = IO::Pager::open *STDOUT, '|-:utf8', 'Unbuffered';

    print <<"  HEREDOC" ;
    A bunch of text later

    # $token passes out of scope and filehandle is automagically closed

    # You can also use scalar filehandles...
    my $token = IO::Pager::open(my $FH) or warn($!); XXX
    print $FH "No globs or barewords for us thanks!\n" while 1;


IO::Pager can be used to locate an available pager and set the PAGER environment variable (see "NOTES"). It is also a factory for creating I/O objects such as IO::Pager::Buffered and IO::Pager::Unbuffered.

IO::Pager subclasses are designed to programmatically decide whether or not to pipe a filehandle's output to a program specified in PAGER. Subclasses may implement only the IO handle methods desired and inherit the remainder of those outlined below from IO::Pager. For anything else, YMMV. See the appropriate subclass for implementation specific details.



Almost identical to open, except that you will get an IO::Handle back if there's no TTY to allow for IO::Pager-agnostic programming.


Instantiate a new IO::Pager, which will paginate output sent to FILEHANDLE if interacting with a TTY.

Save the return value to check for errors, use as an object, or for implict close of OO handles when the variable passes out of scope.


You may provide a glob or scalar.

Defaults to currently select()-ed FILEHANDLE.


Specifies which variety of IO::Pager to create. This accepts fully qualified packages IO::Pager::Buffered, or simply the third portion of the package name Buffered for brevity.

Defaults to IO::Pager::Unbuffered.

Returns false and sets $! on failure, same as perl's open.


Call this method on the token returned by open to get the process identifier for the child process i.e; pager; if you need to perform some long term process management e.g; perl's waitpid

You can also access the PID by numifying the instantiation token like so:

  my $child = $token+0;


Explicitly close the filehandle, this stops any redirection of output on FILEHANDLE that may have been warranted.

This does not default to the current filehandle.

Alternatively, you may rely upon the implicit close of lexical handles as they pass out of scope e.g;

     IO::Pager::open local *RIBBIT;
     print RIBBIT "No toad sexing allowed";
  #The filehandle is closed to additional output

     my $token = new IO::Pager::Buffered;
     $token->print("I like trains");
  #The string "I like trains" is flushed to the pager, and the handle closed

binmode( FILEHANDLE, [LAYER] )

Used to set the I/O layer a.k.a. discipline of a filehandle, such as ':utf8' for UTF-8 encoding.


IO::Pager implements a pseudo-IO-layer for capturing output and sending it to a file, similar to tee(1). Although it is limited to one file, this feature is pure-perl and adds no dependencies.

You may indicate what file to store in parentheses, otherwise the default is $$.log. You may also use an implicit (no indicator) or explicit (>) indicator to overwrite an existing file, or an explicit (>>) for appending to a log file. For example:

    binmode(*STDOUT, ':LOG(clobber.log)');

For full tee-style support, use PerlIO::Util like so:

    binmode(*STDOUT, ":tee(TH)");


Used in the eval-until-eof idiom below, IO::Pager will handle broken pipes from deceased children for you in one of two ways. If $ENV{IP_EOF} is false then program flow will pass out of the loop on SIGPIPE, this is the default. If the variable is true, then the program continues running with output for the previously paged filehandle directed to the STDOUT stream; more accurately, the filehandle is reopened to file descriptor 1.

  use IO::Pager::Page; #or whichever you prefer;
    say "Producing prodigious portions of product";
  } until( eof(*STDOUT) );
  print "Cleaning up after our child before terminating."

If using eof() with less, especially when IP_EOF is set, you may want to use the --no-init option by setting $ENV{IP_EOF}='X' to prevent the paged output from being erased when the pager exits.

fileno( FILEHANDLE )

Return the filehandle number of the write-only pipe to the pager.


print() to the filehandle.


printf() to the filehandle.


syswrite() to the filehandle.



Controls IO:Pager behavior when eof is used.


The location of the default pager.


If the location in PAGER is not absolute, PATH may be searched.

See "NOTES" for more information.


IO::Pager may fall back to these binaries in order if PAGER is not executable.

IO::Pager::Perl as tp via IO::Pager::less

See "NOTES" for more information.


The algorithm for determining which pager to use is as follows:

1. Defer to PAGER

If the PAGER environment variable is set, use the pager it identifies, unless this pager is not available.

2. Usual suspects

Try the standard, hardcoded paths in "FILES".

3. File::Which

If File::Which is available, use the first pager possible amongst less, most, w3m, lv, pg and more.

4. Term::Pager via IO::Pager::Perl

You may also set $ENV{PAGER} to Term::Pager to select this extensible, pure perl pager for display.

5. more

Set PAGER to more, and cross our fingers.

Steps 1, 3 and 5 rely upon the PATH environment variable.


You probably want to do something with SIGPIPE eg;

  eval {
    local $SIG{PIPE} = sub { die };
    local $STDOUT = IO::Pager::open(*STDOUT);

    while (1) {
      # Do something

  # Do something else


IO::Pager::Buffered, IO::Pager::Unbuffered, I::Pager::Perl, IO::Pager::Page, IO::Page, Meta::Tool::Less


Jerrad Pierce <>

Florent Angly <>

This module was inspired by Monte Mitzelfelt's IO::Page 0.02


Copyright (C) 2003-2020 Jerrad Pierce

  • Thou shalt not claim ownership of unmodified materials.

  • Thou shalt not claim whole ownership of modified materials.

  • Thou shalt grant the indemnity of the provider of materials.

  • Thou shalt use and dispense freely without other restrictions.

Or, if you prefer:

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.0 or, at your option, any later version of Perl 5 you may have available.