X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/b4e3ab4c99479faa0b03df7b45e8c69d301e09b8..aaca88ef9f8362a7b895db25ddd51d767502deeb:/tools/convert-unicode.pl diff --git a/tools/convert-unicode.pl b/tools/convert-unicode.pl index 19fde5d..51e27cd 100755 --- a/tools/convert-unicode.pl +++ b/tools/convert-unicode.pl @@ -47,9 +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}++; } } @@ -66,7 +71,7 @@ for my $chr (keys %info) { # 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'; @@ -95,6 +100,7 @@ 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} });