manned/lib/ManUtils/ManUtils.pm
Yorhel ccecb28835 Drop grog(1) in favor of 'groff -mandoc' + two regexes
This provides an almost 2x speedup in man page rendering time and
removes some heuristics to work around bad guesses by grog(1).

Funnily enough, this also fixes rendering of obscure man pages that
happen to use 'grap' macros; grog detected those correctly but my groff
installation doesn't actually support rendering that.

No doubt I broke rendering of other pages, will have to see.
2024-04-28 15:08:00 +02:00

64 lines
1.5 KiB
Perl

package ManUtils;
use strict;
use warnings;
use AE;
use AnyEvent::Util;
use Encode 'decode_utf8', 'encode_utf8';
our $VERSION = '0.01';
require XSLoader;
XSLoader::load('ManUtils', $VERSION);
# Usage: $cv = fmt($input, \$output, \@errors)
# $cv = AnyEvent condition variable, fired when done.
# $input = UTF-8 encoded manual page source
# $output = variable that will hold the output when done
# @errors = list of warnings/errors while running groff
sub fmt {
my($input, $output, $errors) = @_;
$$output = '';
@$errors = ();
my @cmd = 'groff';
push @cmd, '-t' if $input =~ /^\.TS/m;
push @cmd, '-e' if $input =~ /^\.EQ/m;
# $MANWIDTH works by using the following groff options: -rLL=100n -rLT=100n
push @cmd, qw/-mandoc -Tutf8 -DUTF-8 -P-c -rLL=80n -rLT=80n -/;
$input =
# Disable hyphenation, since that screws up man page references. :-(
".hy 0\n.de hy\n..\n"
# Emulate man-db's --nj option
.".na\n.de ad\n..\n"
.encode_utf8($input);
my $groff = run_cmd \@cmd,
'<' => \$input,
'>' => $output,
'2>' => sub { if($_[0]) { chomp(my $e = $_[0]); push @$errors, "groff: $e" } };
my $cv = AE::cv;
$groff->cb(sub {
decode_utf8 $$output;
$$output =~ s/[\t\s\r\n]+$//;
$cv->send;
});
$cv
}
# Blocking version of fmt(). Returns the formatted man page, errors are
# forwarded to warn().
sub fmt_block {
my $c = shift;
my $cv = fmt $c, \my $out, \my @err;
$cv->recv;
#warn "$_\n" for @err;
$out;
}
1;