#!/bin/sh -- # A comment mentioning perl
eval 'exec perl -S $0 ${1+"$@"}'
        if 0;

# Usage: dehtml [-f] [<files>]

if ($ARGV[0] eq '-f' ) {
	shift;
	$fmt = 1;
}

$nested_comments = 1;	# try to remove nested <!-- ... <!-- ... --> ... -->

$all = join('', <>);

$all =~ s/\r//g;

if ( $all !~ /\n$/ ) {
	$all .= "\n\n";
}

# remove blocks of unwanted stuff:
if ( $nested_comments ) {
	$all = &remove_nested_html_comments($all);
} else {
	$all =~ s/<!--(.*?)-->//sg;
}
$all =~ s,<SCRIPT\b(.*?)</SCRIPT>,,sig;	
$all =~ s,<STYLE\b(.*?)</STYLE>,,sig;

# simple formating:
$all =~ s/<LI>/\n\t/ig;
$all =~ s/<P>/\n\n/ig;
$all =~ s/<BR>/\n/ig;
$all =~ s,</*(B|EM)>,*,ig;

# remove html tags
$all =~ s/<[^>]*>//g;

# expand some of the &foo; items:
$all =~ s/&(nbsp|#160);?/ /ig;
$all =~ s/&lt;?/</ig;
$all =~ s/&gt;?/>/ig;
$all =~ s/&amp;?/&/ig;
$all =~ s/&(quote|#34);?/°/ig;
$all =~ s/&(cent|#162);?/¢/ig;
$all =~ s/&(pound|#163);?/£/ig;
$all =~ s/&(copy|#169);?/(c)/ig;
$all =~ s/&(reg|#174);?/(R)/ig;
$all =~ s/&(deg|#176);?/°/ig;
$all =~ s/&(trade|#8482|#153);?/[tm]/ig;
$all =~ s/&(middot|#183);?/·/ig;
$all =~ s/&(frac14|#188);?/¼/ig;
$all =~ s/&(frac12|#189);?/½/ig;
$all =~ s/&(frac34|#190);?/¾/ig;
$all =~ s/&(divide|#247);?/÷/ig;
$all =~ s/&(ntilde|#241);?/ñ/ig;

# trim excess white space
$all =~ s/^[ \t]*$//mg;

# limit number of blank lines in a row to 3:
$all =~ s/\n\n(\n)+/\n\n\n/sg;

if ( $fmt && open(FMT, "|fmt") ) {
	print FMT $all;
	close(FMT);
} else {
	print STDOUT $all;
}

exit;

# this is a crude way to remove nested comments:
sub remove_nested_html_comments {
	my ($html) = @_;
	$html =~ s/%/___QUOTE_PERCNT_$$_/og;
	$html =~ s/=/___QUOTE_EQUALS_$$_/og;
	$html =~ s/<\s*!--/%/g;
	$html =~ s/--\s*>/=/g;

	while ( $html =~ /%[^%=]*=/ ) {
		$html =~ s/%[^%=]*=//sg;
	}

	$html =~ s/%/<!--/g;
	$html =~ s/=/-->/g;

	$html =~ s/___QUOTE_PERCNT_$$_/%/og;
	$html =~ s/___QUOTE_EQUALS_$$_/=/og;

	return $html;
}