#!/usr/bin/perl -w -- # -*- Perl -*-

my @filenames = ();
my $opts = "";

my $tmpStdout = "/tmp/tidy.stdout.$$";
my $tmpStderr = "/tmp/tidy.stderr.$$";

while (@ARGV) {
    $_ = shift @ARGV;
    if (@filenames || !/^-/) {
	push (@filenames, $_);
    } else {
	$opts .= " " if $opts ne "";
	$opts .= $_;

	if ($_ eq '--doctype') {
	    $_ = shift @ARGV;
	    $opts .= " $_";
	}
    }
}

foreach my $file (@filenames) {
    my $content = "";

    system("/usr/bin/tidy -wrap 512 $opts $file > $tmpStdout 2>$tmpStderr");
    if (-s $tmpStderr != 0) {
	print STDERR "Tidy $file...\n";
	open (F, $tmpStderr);
	while (<F>) {
	    print $_;
	}
	close (F);
    }

    if (-s $tmpStdout == 0) {
	# we must have done an "in-place" tidy
	open (F, $file);
	read (F, $content, -s $file);
	close (F);
    } else {
	open (F, $tmpStdout);
	read (F, $content, -s $tmpStdout);
	close (F);
    }

    $content = cleanupContent($content);

    open (F, ">$file");
    print F $content;
    close (F);
}

unlink $tmpStdout;
unlink $tmpStderr;

sub cleanupContent {
    my $content = shift;

    $content =~ s/(<a\s[^>]+>)\s+/$1/sg;
    $content =~ s/\s+(<\/a>)/$1/sg;

    my $newContent = "";

    # workaround a bug in tidy...it removes &nbsp; sometimes!
    while ($content
	     =~ /^(.*?)(<div class=([\'\"])literallayout\3.*?>)(.*?)(<\/div>.*)$/is) {
	my $pre = $1;
	my $starttag = $2;
	my $litlayout = $4;
	my $post = $5;

	#print "====$litlayout====\n";
	#print "------------------\n";
	#print "====", addnbsp($litlayout), "====\n";

	$newContent .= $pre . $starttag . addnbsp($litlayout);
	$content = $post;
    }

    $content = $newContent . $content;
    $newContent = "";

    while ($content
	   =~ /^(.*?)(<script[^>]+text\/javascript.*?>\s*\/\/<!\[CDATA\[)(.*?)(\/\/\]\]\>.*)$/is) {
	my $pre = $1;
	my $scripttag = $2;
	my $script = $3;
	my $post = $4;

	# belt and suspenders is both unnecessary ... and WRONG
	$script =~ s/\&lt;/</sg;
	$script =~ s/\&gt;/>/sg;
	$script =~ s/\&amp;/\&/sg;

	$newContent .= $pre . $scripttag . $script;

	$content = $post;
    }

    return $newContent . $content;
}

sub addnbsp {
    my $content = shift;
    my $esc = "";

    while ($content ne "") {
	if ($content =~ /^(.*?)(<.*?>)(.*)$/s) {
	    my $text = $1;
	    my $tag = $2;
	    $content = $3;
	    $text =~ s/ /&\#160;/g;
	    $esc .= $text . $tag;
	} else {
	    my $text = $content;
	    $content = "";
	    $text =~ s/ /&\#160;/g;
	    $esc .= $text;
	}
    }

    return $esc;
}