#!/usr/pkg/bin/perl -w

print "Re-escaping chars for UTF-8...\n";

$cntr = 0;

while (glob "*.html")
{ chmod (0666, $_);
  $was_name = $_;

  open(WAS, "< $_") or die "Can't open file '$_' for reading: $!";
  open(NEW, "> fix.html") or die "Can't open new file for writing: $!";

  $cntr++;

  while (defined($linio = <WAS>))
  { chomp $linio;
	
	  # C with circumflex
	  $linio =~ s/Ĉ/&#264;/g;
	  $linio =~ s/ĉ/&#265;/g;
		
    # G with circumflex
	  $linio =~ s/Ĝ/&#284;/g;
	  $linio =~ s/ĝ/&#285;/g;
		
    # H with circumflex
	  $linio =~ s/Ĥ/&#292;/g;
	  $linio =~ s/ĥ/&#293;/g;
		
    # J with circumflex
	  $linio =~ s/Ĵ/&#308;/g;
	  $linio =~ s/ĵ/&#309;/g;
		
    # S with circumflex
	  $linio =~ s/Ŝ/&#348;/g;
	  $linio =~ s/ŝ/&#349;/g;
		
    # U with breve
	  $linio =~ s/Ŭ/&#/g;
	  $linio =~ s/ŭ/&#/g;
    
    # single quotes
	  $linio =~ s/‘/&#8216;/g;
	  $linio =~ s/’/&#8217;/g;
		
    # double quotes
	  $linio =~ s/“/&#8220;/g;
	  $linio =~ s/”/&#8221;/g;
		
    # n- and m-dashes
	  $linio =~ s/–/&#8211;/g;
	  $linio =~ s/—/&#8212;/g;
	  $linio =~ s/―/&#8213;/g;
		
    # ellipsis
	  $linio =~ s/ΰ/&#8163;/g;
		
    # Left, up, down and right arrows
	  $linio =~ s/←/&#8592/g;
	  $linio =~ s/↑/&#8593/g;
	  $linio =~ s/↓/&#8595/g;
	  $linio =~ s/→/&#8594/g;

    print NEW "$linio \n";
  }

  close(WAS);
  close(NEW);

  rename("fix.html", "$was_name");
  unlink "fix.html";

}

print "Done. In all, $cntr files were processed.\n";