package IO::Pager; our $VERSION = "2.10"; #Untouched since 1.03 use 5.008; #At least, for decent perlio, and other modernisms use strict; use warnings; use base qw( Tie::Handle ); use Env qw( PAGER ); use File::Spec; use PerlIO; use Symbol; use overload '+' => "PID", bool=> "PID"; our $SIGPIPE; #use Carp; $SIG{__WARN__} = sub{ print STDERR @_, Carp::longmess(),"\n\n"; }; sub find_pager { # Return the name (or path) of a pager that IO::Pager can use my $io_pager; #Permit explicit use of pure perl pager local $_ = 'IO::Pager::less'; return $_ if (defined($_[0]) && ($_[0] eq $_)) or (defined($PAGER) && ($PAGER eq $_)); # Use File::Which if available (strongly recommended) my $which = eval { require File::Which }; # Look for pager in PAGER first if ($PAGER) { # Strip arguments e.g. 'less --quiet' my ($pager, @options) = (split ' ', $PAGER); $pager = _check_pagers([$pager], $which); $io_pager = join ' ', ($pager, @options) if defined $pager; } # Then search pager amongst usual suspects if (not defined $io_pager) { my @pagers = ('/etc/alternatives/pager', '/usr/local/bin/less', '/usr/bin/less', '/usr/bin/more'); $io_pager = _check_pagers(\@pagers, $which) } # Then check PATH for other pagers if ( (not defined $io_pager) && $which ) { my @pagers = ('less', 'most', 'w3m', 'lv', 'pg', 'more'); $io_pager = _check_pagers(\@pagers, $which ); } # If all else fails, default to more (actually IO::Pager::less first) $io_pager ||= 'more'; return $io_pager; } sub _check_pagers { my ($pagers, $which) = @_; # Return the first pager in the list that is usable. For each given pager, # given a pager name, try to finds its full path with File::Which if possible. # Given a pager path, verify that it exists. my $io_pager = undef; for my $pager (@$pagers) { # Get full path my $loc; if ( $which && (not File::Spec->file_name_is_absolute($pager)) ) { $loc = File::Which::which($pager); } else { $loc = $pager; } # Test that full path is valid (some platforms don't do -x so we use -e) if ( defined($loc) && (-e $loc) ) { $io_pager = $loc; last; } } return $io_pager; } #Should have this as first block for clarity, but not with its use of a sub BEGIN { # Set the $ENV{PAGER} to something reasonable our $oldPAGER = $PAGER || ''; $PAGER = find_pager(); if( ($PAGER =~ 'more' and $oldPAGER ne 'more') or $PAGER eq 'IO::Pager::less' ){ my $io_pager = $PAGER; eval "use IO::Pager::less"; $PAGER = $io_pager if $@ or not defined $PAGER; } } #Factory sub open(*;$@) { # FH, [MODE], [CLASS] my $args = {procedural=>1}; $args->{mode} = splice(@_, 1, 1) if scalar(@_) == 3; $args->{subclass} = pop if scalar(@_) == 2; &new(undef, @_, $args); } #Alternate entrance: drop class but leave FH, subclass sub new(*;$@) { # FH, [MODE], [CLASS] shift; my %args; if( ref($_[-1]) eq 'HASH' ){ %args = %{pop()}; #warn "REMAINDER? (@_)", scalar @_; push(@_, $args{procedural}); } elsif( defined($_[1]) ){ $args{mode} = splice(@_, 1, 1) if $_[1] =~ /^:/; $args{subclass} = pop if exists($_[1]); } #Leave filehandle in @_ for pass by reference to allow gensym $args{subclass} ||= 'IO::Pager::Unbuffered'; $args{subclass} =~ s/^(?!IO::Pager::)/IO::Pager::/; eval "require $args{subclass}" or die "Could not load $args{subclass}: $@\n"; my $token = $args{subclass}->new(@_); if( defined($args{mode}) ){ $args{mode} =~ s/^\|-//; $token->BINMODE($args{mode}); } return $token; } sub _init{ # CLASS, [FH] ## Note reversal of order due to CLASS from new() #Assign by reference if empty scalar given as filehandle $_[1] = gensym() if !defined($_[1]); no strict 'refs'; $_[1] ||= *{select()}; # Are we on a TTY? STDOUT & STDERR are separately bound if ( defined( my $FHn = fileno($_[1]) ) ) { if ( $FHn == fileno(STDOUT) ) { die '!TTY' unless -t $_[1]; } if ( $FHn == fileno(STDERR) ) { die '!TTY' unless -t $_[1]; } } #XXX This allows us to have multiple pseudo-STDOUT #return 0 unless -t STDOUT; return ($_[0], $_[1]); } # Methods required for implementing a tied filehandle class sub TIEHANDLE { my ($class, $tied_fh) = @_; unless ( $PAGER ){ die "The PAGER environment variable is not defined, you may need to set it manually."; } my($real_fh, $child, $dupe_fh); # XXX What about localized GLOBs?! # if( $tied_fh =~ /\*(?:\w+::)?STD(?:OUT|ERR)$/ ){ # open($dupe_fh, '>&', $tied_fh) or warn "Unable to dupe $tied_fh"; # } do{ no warnings; $child = CORE::open($real_fh, '|-', $PAGER) }; if ( $child ){ my @oLayers = PerlIO::get_layers($tied_fh, details=>1, output=>1); my $layers = ''; for(my $i=0;$i<$#oLayers;$i+=3){ #An extra base layer requires more keystrokes to exit next if $oLayers[$i] =~ /unix|stdio/ && !defined($oLayers[+1]); $layers .= ":$oLayers[$i]"; $layers .= '(' . ($oLayers[$i+1]) . ')' if defined($oLayers[$i+1]); } CORE::binmode($real_fh, $layers); } else{ die "Could not pipe to PAGER ('$PAGER'): $!\n"; } return bless { 'real_fh' => $real_fh, # 'dupe_fh' => $dupe_fh, 'tied_fh' => "$tied_fh", #Avoid self-reference leak 'child' => $child, 'pager' => $PAGER, }, $class; } sub BINMODE { my ($self, $layer) = @_; if( $layer =~ /^:LOG\((>{0,2})(.*)\)$/ ){ CORE::open($self->{LOG}, $1||'>', $2||"$$.log") or die $!; } else{ CORE::binmode($self->{real_fh}, $layer||':raw'); } } sub WNOHANG(); sub EOF { my $self = shift; unless( defined($SIGPIPE) ){ eval 'use POSIX ":sys_wait_h";'; $SIGPIPE = 0; } $SIG{PIPE} = sub { $SIGPIPE = 1 unless $ENV{IP_EOF}; CORE::close($self->{real_fh}); waitpid($self->{child}, WNOHANG); CORE::open($self->{real_fh}, '>&1'); close($self->{LOG}); }; return $SIGPIPE; } sub PRINT { my ($self, @args) = @_; CORE::print {$self->{LOG}} @args if exists($self->{LOG}); CORE::print {$self->{real_fh}} @args or die "Could not print to PAGER: $!\n"; } sub PRINTF { my ($self, $format, @args) = @_; $self->PRINT(sprintf($format, @args)); } sub say { my ($self, @args) = @_; $args[-1] .= "\n"; $self->PRINT(@args); } sub WRITE { my ($self, $scalar, $length, $offset) = @_; $self->PRINT(substr($scalar, $offset||0, $length)); } sub TELL { #Buffered classes provide their own, and others may use this in another way return undef; } sub FILENO { CORE::fileno($_[0]->{real_fh}); } sub CLOSE { my ($self) = @_; CORE::close($self->{real_fh}); # untie($self->{tied_fh}); # *{$self->{tied_fh}} = *{$self->{dupe_fh}}; } { no warnings 'once'; *DESTROY = \&CLOSE; } #Non-IO methods sub PID{ my ($self) = @_; return $self->{child}; } #Provide lowercase aliases for accessors foreach my $method ( qw(BINMODE CLOSE EOF PRINT PRINTF TELL WRITE PID) ){ no strict 'refs'; *{lc($method)} = \&{$method}; } 1; __END__ =pod =head1 NAME IO::Pager - Select a pager (possibly perl-based) & pipe it text if a TTY =head1 SYNOPSIS # 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 HEREDOC # $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; } =head1 DESCRIPTION IO::Pager can be used to locate an available pager and set the I environment variable (see L). It is also a factory for creating I/O objects such as L and L. IO::Pager subclasses are designed to programmatically decide whether or not to pipe a filehandle's output to a program specified in I. 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. =head1 METHODS =head2 new( FILEHANDLE, [MODE], [SUBCLASS] ) Almost identical to open, except that you will get an L back if there's no TTY to allow for IO::Pager-agnostic programming. =head2 open( FILEHANDLE, [MODE], [SUBCLASS] ) 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. =over =item FILEHANDLE You may provide a glob or scalar. Defaults to currently select()-ed F. =item SUBCLASS Specifies which variety of IO::Pager to create. This accepts fully qualified packages I, or simply the third portion of the package name I for brevity. Defaults to L. Returns false and sets I<$!> on failure, same as perl's C. =back =head2 PID Call this method on the token returned by C 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 C You can also access the PID by numifying the instantiation token like so: my $child = $token+0; =head2 close( FILEHANDLE ) Explicitly close the filehandle, this stops any redirection of output on FILEHANDLE that may have been warranted. I. 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 =head2 binmode( FILEHANDLE, [LAYER] ) Used to set the I/O layer a.k.a. discipline of a filehandle, such as C<':utf8'> for UTF-8 encoding. =head3 :LOG([>>FILE]) IO::Pager implements a pseudo-IO-layer for capturing output and sending it to a file, similar to L. 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 C<$$.log>. You may also use an implicit (no indicator) or explicit (I>) indicator to overwrite an existing file, or an explicit (IE>) for appending to a log file. For example: binmode(*STDOUT, ':LOG(clobber.log)'); ... $STDOUT->binmode(':LOG(>>noclobber.log)'); For full tee-style support, use L like so: binmode(*STDOUT, ":tee(TH)"); #OR $STDOUT->binmode(':tee(TH)'); =head2 eof( FILEHANDLE ) Used in the eval-until-eof idiom below, I will handle broken pipes from deceased children for you in one of two ways. If I<$ENV{IP_EOF}> is false then program flow will pass out of the loop on I, this is the default. If the variable is true, then the program continues running with output for the previously paged filehandle directed to the I stream; more accurately, the filehandle is reopened to file descriptor 1. use IO::Pager::Page; #or whichever you prefer; ... eval{ say "Producing prodigious portions of product"; ... } until( eof(*STDOUT) ); print "Cleaning up after our child before terminating." If using eof() with L, especially when IP_EOF is set, you may want to use the I<--no-init> option by setting I<$ENV{IP_EOF}='X'> to prevent the paged output from being erased when the pager exits. =head2 fileno( FILEHANDLE ) Return the filehandle number of the write-only pipe to the pager. =head2 print( FILEHANDLE LIST ) print() to the filehandle. =head2 printf( FILEHANDLE FORMAT, LIST ) printf() to the filehandle. =head2 syswrite( FILEHANDLE, SCALAR, [LENGTH], [OFFSET] ) syswrite() to the filehandle. =head1 ENVIRONMENT =over =item IP_EOF Controls IO:Pager behavior when C is used. =item PAGER The location of the default pager. =item PATH If the location in PAGER is not absolute, PATH may be searched. See L for more information. =back =head1 FILES IO::Pager may fall back to these binaries in order if I is not executable. =over =item /etc/alternatives/pager =item /usr/local/bin/less =item /usr/bin/less =item L as C via L =item /usr/bin/more =back See L for more information. =head1 NOTES The algorithm for determining which pager to use is as follows: =over =item 1. Defer to I If the I environment variable is set, use the pager it identifies, unless this pager is not available. =item 2. Usual suspects Try the standard, hardcoded paths in L. =item 3. File::Which If File::Which is available, use the first pager possible amongst C, C, C, C, C and L. =item 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. =item 5. more Set I to C, and cross our fingers. =back Steps 1, 3 and 5 rely upon the I environment variable. =head1 CAVEATS 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 =head1 SEE ALSO L, L, L, L, L, L =head1 AUTHOR Jerrad Pierce Florent Angly This module was inspired by Monte Mitzelfelt's IO::Page 0.02 =head1 COPYRIGHT AND LICENSE Copyright (C) 2003-2020 Jerrad Pierce =over =item * Thou shalt not claim ownership of unmodified materials. =item * Thou shalt not claim whole ownership of modified materials. =item * Thou shalt grant the indemnity of the provider of materials. =item * Thou shalt use and dispense freely without other restrictions. =back 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. =cut