X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/3903ac75fe6b199a3d69558ed9ecf1235f9b6bf5..aaca88ef9f8362a7b895db25ddd51d767502deeb:/tools/convert-unicode.pl

diff --git a/tools/convert-unicode.pl b/tools/convert-unicode.pl
index 348878c..51e27cd 100755
--- a/tools/convert-unicode.pl
+++ b/tools/convert-unicode.pl
@@ -30,6 +30,15 @@ eval {
 	1;
 } or warn "Failed reading unicode tables: $@";
 
+eval {
+	require HTML::Entities;
+	while (my ($char, $entity) = each %HTML::Entities::char2entity) {
+		$entity =~ /[a-zA-Z]/ or next;  # only actual aliases
+		$info{$char}->{html} = substr($entity, 1, -1);
+	}
+	1;
+} or warn "Failed importing html entities: $@";
+
 my %diinc = (
 	'digraphs.inc.pl' => 'u-di',
 );
@@ -38,9 +47,14 @@ for (keys %diinc) {
 	my $di = do $_ or die "Error reading digraphs file $_: ", $@ || $!;
 	while (my ($mnem, $cp) = each %$di) {
 		length $mnem == 2 or next;  # limit to digraphs
-		$cp = chr $cp->[0] if ref $cp;  # old style array
+		my $class = $diinc{$_};
+		if (ref $cp) {
+			# old style array
+			$class = 'u-prop' if $cp->[2] and $cp->[2] =~ m/\bXz\b/;
+			$cp = chr $cp->[0];
+		}
 		$info{$cp}->{di} //= $mnem;
-		$info{$cp}->{class}->{$_}++ for $diinc{$_};
+		$info{$cp}->{class}->{$class}++;
 	}
 }
 
@@ -53,11 +67,11 @@ for my $chr (keys %info) {
 			|| { block => '?', category => 'Xn', name => '', script => '' }
 	} or next;
 
-	$info->{$_} = $info{$chr}->{$_} for qw(di class string);
+	$info->{$_} = $info{$chr}->{$_} for qw(di html class string);
 
 	# categorise by unicode types and writing script
 	$info->{class}->{$_}++ for $info->{category};
-	$info->{class}->{"u-$_"}++ for $info->{script} || ();
+	$info->{class}->{$_}++ for $info->{script} || ();
 
 	# add custom categories for certain blocks
 	$info->{class}->{Xa}++ if $info->{block} eq 'Basic Latin';
@@ -86,11 +100,12 @@ for my $chr (keys %info) {
 }
 
 # output perl code of hash
+say 'use utf8;';
 say '+{';
 for my $cp (sort keys %info) {
 	$info{$cp}->{classstr} = join(' ', sort keys %{ $info{$cp}->{class} });
 	# convert info hashes into arrays of strings to output in display order
-	my $row = [ map { $info{$cp}->{$_} } qw/classstr name di string/ ];
+	my $row = [ map { $info{$cp}->{$_} } qw/classstr name di html string/ ];
 	# strip off trailing missing values (especially string may be unknown)
 	defined $row->[-1] ? last : pop @$row for 1 .. @$row;
 	# final line (assume safe within single quotes)