#!/usr/bin/perl
# Warren Block
# ts=4

use strict;
use warnings;

use Getopt::Std;
use File::Basename;

# convert leading tabs to 8 spaces, ignoring special sections

our ($opt_h, $opt_d, $opt_r);

my $version = '$Rev$';

my $ignore = 0;
my @ignoreblock_tags;
my $ignoreblockstart;
my $ignoreblockend;

my $prog = basename($0);

sub usage {
	$version =~ tr/A-Z:$/a-z/d;
	print <<USAGE;
$prog $version
usage: $prog [-h|-d|-r] file

    Detab and retab DocBook SGML files without harming special
    blocks like <programlisting>.

    -h  show summary of command line options and exit

	-d  detab: convert leading tabs to blocks of 8 spaces

    -r  retab: convert leading blocks of 8 spaces to tabs
USAGE
	exit 0;
}

sub initialize {
	# create regex for sgml block start and end
	@ignoreblock_tags = qw/ literallayout screen programlisting /;
	$ignoreblockstart = '(?:<!--';
	for my $tag (@ignoreblock_tags) {
		$ignoreblockstart .= "|<$tag.*?>";
	}
	$ignoreblockstart .= ')';
	$ignoreblockend = '(?:-->';
	for my $tag (@ignoreblock_tags) {
		$ignoreblockend .= "|<\/$tag>";
	}
	$ignoreblockend .= ')';
}

sub tabinate {
	my $txt = shift;
	return $txt if $ignore;
	if ( $opt_d ) {
		$txt =~ s/\t/        /g;
	} elsif ( $opt_r ) {
		$txt =~ s/( {8})/\t/g;
	}
	return $txt;
}

getopts('hdr');
usage() if $opt_h || ($#ARGV < 0);
# usage() unless $opt_d || $opt_r;

# main
initialize();

while (<>) {
	# don't chomp, leave linefeeds
	if ( /^\s*$/ ) {
  		print;
  		next;
	}

	if ( /^(\s*)(\s*.*)$/ ) {
		my $indent = $1;
		my $text   = $2;
	} else {
		# no indent
		print
	}

	if ( /($ignoreblockstart)/ ) {
		# operate on this line, but start ignoring
		print tabinate($_);
		$ignore = 1;
		next;
	}

	if ( /($ignoreblockend)/ ) {
		# don't operate on this line, but start paying attention
		print;
		$ignore = 0;
		next;
	}

	print tabinate($_);
}