compile-manual.pl

#!/usr/bin/perl

# Cheezy perl script to produce the typeset Unix manual, as ken intended.
#
# Gets a list of every manpage from man(1), and compile them into per-section PDF files in the final/ subdir
# Pulls the intro page to the front of each section, if there is one.
# Also leaves the individual PDF pages in the various man* subdirs
#
# Assumptions: you have groff and ghostscript installed. I think you will need to run makewhatis, too.
# 
# You probably want to make sure the man-pages package is installed to get section 2 & 3 fleshed out
#
# Takes about an hour to run on my average CentOS 6 system, producing 18145 pages (45MB) of PDF output!
#
# Twitter: @anotherhowie November 2016


mkdir("final");

# man will tell us all the manpages it knows about
open(PAGES, "man -aw  \\* |");
while(<PAGES>) {
	chomp; chomp;
	@parts = split(/\//);
	$section = $parts[-2];

	$new_path = join("/", @parts[-2,-1]);
	$new_path =~ s/\.gz/.pdf/;

	if (! exists($sections{$section})) {
		$sections{$section} = {};
		mkdir $section;
	}
	$sections{$section}{$_} = $new_path;
}
close(PAGES);

foreach $section (sort keys %sections) {
	print "## Section $section\n";
	@pages = sort keys %{$sections{$section}};
	foreach $page (@pages) {
		$new_path = $sections{$section}{$page};
		# print "# $page => $new_path\n";
		$command = "cat $page | groff -Tps -mandoc - | ps2pdf - $new_path";
		if ($page =~ /\.gz$/) {
			$command = "z".$command;
		}
		if (! -f $new_path) {
			print "$command\n";
			system($command);
		}
	}
	# end of section collating
	@outputs = map {$sections{$section}{$_}} @pages;

	# move the intro to the front of the section, if there is one
	($intro) = grep { m/\/intro\./} @outputs;
	@page_order =  grep { ! m/\/intro\./} @outputs;
	if ($intro) {
		unshift(@page_order, $intro);
	}

	system("gs", "-dBATCH", "-dNOPAUSE", "-sDEVICE=pdfwrite", "-sOutputFile=final/${section}.pdf", @page_order);
}