X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/dd5b69ff6ef39a7b8521bac32f2a1aa8ad327729..9314086776a6cd9afc7349082d3acdafa970db09:/tools/mkcharinfo

diff --git a/tools/mkcharinfo b/tools/mkcharinfo
index b673105..27c11d9 100755
--- a/tools/mkcharinfo
+++ b/tools/mkcharinfo
@@ -61,6 +61,16 @@ for (keys %diinc) {
 	}
 }
 
+eval {
+	# read introducing unicode versions for known characters
+	my $agemap = do 'unicode-age.inc.pl' or die $@ || $!;
+	for my $chr (keys %info) {
+		my $version = $agemap->{ord $chr} or next;
+		$info{$chr}->{class}->{'u-v'.$version}++
+	}
+	1;
+} or warn "Failed including unicode version data $@";
+
 for my $chr (keys %info) {
 	my $cp = ord $chr;
 	# attempt to get unicode character information
@@ -70,7 +80,7 @@ for my $chr (keys %info) {
 			|| { block => '?', category => 'Xn', name => '', script => '' }
 	} or next;
 
-	$info->{$_} = $info{$chr}->{$_} for qw(di html class string);
+	$info->{$_} = $info{$chr}->{$_} for keys %{ $info{$chr} };
 
 	# categorise by unicode types and writing script
 	$info->{class}->{$_}++ for $info->{category};
@@ -95,7 +105,9 @@ for my $chr (keys %info) {
 				? '<'.$info->{unicode10}.'>'  # the old name was much more useful
 				: sprintf('<control U+%04X>', $cp);  # at least identify by value
 			# show descriptive symbols instead of control chars themselves
-			$info->{string} = $cp < 32 ? chr($cp + 0x2400) : chr(0xFFFD);
+			$info->{string} = $cp < 32   ? chr($cp + 0x2400) :
+			                  $cp == 127 ? chr(0x2421) :
+			                               chr(0xFFFD);
 		}
 	}