File Coverage

blib/lib/EBook/Ishmael/TextBrowserDump.pm

Criterion	Covered	Total	%
statement	23	42	54.7
branch	0	12	0.0
condition	0	12	0.0
subroutine	8	9	88.8
pod	1	1	100.0
total	32	76	42.1

package EBook::Ishmael::TextBrowserDump;
use 5.016;
our $VERSION = '2.03';
use strict;
use warnings;

use Exporter 'import';
our @EXPORT = qw(browser_dump);

use Encode qw(decode);
use List::Util qw(first);

use File::Which;

use EBook::Ishmael::ShellQuote qw(safe_qx);

our $CAN_DUMP = 0;

# TODO: Is there some sort of post-processing we can do to get rid of s?

# From links
my $WIDTH_MAX = 512;
my $WIDTH_MIN = 10;

my @ORDER = qw(
    lynx
    links
    elinks
    w3m
    chawan
    pandoc
    queequeg
);

my %Browsers = (
    'lynx' => {
        Bins  => [ qw(lynx) ],
        Bin   => undef,
        Opts  => [ qw(-dump -force_html -nolist -display_charset=utf8) ],
        Width => '-width %d',
        Xhtml => [ qw(-xhtml_parsing) ],
    },
    'links' => {
        Bins  => [ qw(links links2) ],
        Bin   => undef,
        Opts  => [ qw(-dump -force-htm -codepage utf8) ],
        Width => '-width %d',
        Xhtml => [],
    },
    'elinks' => {
        Bins  => [ qw(elinks) ],
        Bin   => undef,
        Opts  => [ qw(-dump -force-html -no-home -no-references -no-numbering
                      -dump-charset utf8) ],
        Width => '-dump-width %d',
        Xhtml => [],
    },
    'w3m' => {
        Bins  => [ qw(w3m) ],
        Bin   => undef,
        Opts  => [ qw(-dump -T text/html -O utf8) ],
        Width => '-cols %d',
        Xhtml => [],
    },
    'chawan' => {
        Bins  => [ qw(cha) ],
        Bin   => undef,
        Opts  => [ qw(-d -I utf8 -O utf8 -T text/html) ],
        Width => "-o 'display.columns=%d'",
        Xhtml => [],
    },
    'pandoc' => {
        Bins  => [ qw(pandoc) ],
        Bin   => undef,
        Opts  => [ qw(-f html -t plain -o -) ],
        Width => "--columns %d",
        Xhtml => [],
    },
    'queequeg' => {
        Bins  => [ qw(queequeg) ],
        Bin   => undef,
        Opts  => [ qw(-e utf8) ],
        Width => '-w %d',
        Xhtml => [],
    },
);

my $Default = undef;
for my $k (@ORDER) {
    my $bin = first { which $_ } @{ $Browsers{ $k }->{Bins} };
    next unless defined $bin;
    $Browsers{ $k }->{Bin} = $bin;
    $Default //= $k;
}

$CAN_DUMP = defined $Default;

unless ($CAN_DUMP) {
    warn "No valid text browser was found installed on your system, you " .
         "will be unable to use any feature requiring a text browser\n";
}

sub browser_dump {

    unless (defined $Default) {
        die "Cannot format HTML; no valid program was found on your system\n";
    }

    my $file  = shift;
    my $param = shift // {};

    my $browser = $param->{browser} // $Default;
    my $xhtml   = $param->{xhtml}   // 0;
    my $width   = $param->{width}   // 80;

    unless (exists $Browsers{ $browser }) {
        die <<"HERE";
'$browser' is not a valid program; the following is a list of valid programs:
    lynx
    links
    elinks
    w3m
    chawan
    pandoc
    queequeg
HERE
    }

    unless (defined $Browsers{ $browser }->{Bin}) {
        die "'$browser' is not installed on your system\n";
    }

    unless ($width >= $WIDTH_MIN and $width <= $WIDTH_MAX) {
        die "Width cannot be greater than $WIDTH_MAX or less than $WIDTH_MIN\n";
    }

    my $dump = safe_qx(
        $Browsers{ $browser }->{Bin},
        split(/\s+/, sprintf($Browsers{ $browser }->{Width}, $width)),
        @{ $Browsers{ $browser }->{Opts} },
        ($xhtml ? join(" ", @{ $Browsers{ $browser }->{Xhtml} }) : ()),
        $file
    );
    unless ($? >> 8 == 0) {
        die "Failed to dump $file with $Browsers{ $browser }->{Bin}\n";
    }

    # We can't use 'open IN => ":encoding(UTF-8)"' because it was broken prior
    # to 5.34, so we must manually decode :-/.
    return decode('UTF-8', $dump);

}

1;

=head1 NAME

EBook::Ishmael::TextBrowserDump - Format HTML through different installed programs

=head1 SYNOPSIS

  use EBook::Ishmael::TextBrowserDump;

  my $dump = browser_dump($file);

=head1 DESCRIPTION

B is a module for dumping the contents of
HTML files to formatted text, via programs like L. For
L user documentation, you should consult its manual (this is
developer documentation).

B requires at least one of the following
programs to be installed:

=over 4

=item L

=item L

=item L

=item L

=item chawan

=item L

=item L

=back

=head1 SUBROUTINES

=head2 $dump = browser_dump($file, $opt_ref)

Subroutine that dumps the formatted text contents of C<$file> via a text
web browser.

C can also be given a hash ref of options.

=over 4

=item browser

The specific browser you would like to use for the dumping. See above for a list
of valid browsers. If not specified, defaults to the first browser
C finds installed on your system.

=item xhtml

Bool specifying whether the input file is XHTML or not. Defaults to C<0>.

=item width

Specify the width of the formatted text. Defaults to C<80>.

=back

=head1 GLOBAL VARIABLES

=head2 $EBook::Ishmael::TextBrowserDump::CAN_DUMP

Bool stating whether this module is able to dump or not (is a valid
text browser installed or not).

=head1 AUTHOR

Written by Samuel Young, [e-mail].

This project's source can be found on its
L. Comments and pull
requests are welcome!

=head1 COPYRIGHT

Copyright (C) 2025-2026 Samuel Young

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

=head1 SEE ALSO

L, L, L, L, L, L,
L,

=cut