digraphs: mark latin/ascii characters

[sheet.git] / rfc1345convert

diff --git a/rfc1345convert b/rfc1345convert
index 8b29b8048b821846b23a9886f22beb3a49754a66..6edfa7445083ea4588365224c05927102eb6f0a2 100644 (file)
--- a/rfc1345convert
+++ b/rfc1345convert
@@ -47,18 +47,24 @@ for (@t) {
 }
 
 # optionally get unicode character information
-my %charinfo = eval {
+my %info = eval {
        require Unicode::UCD;
        map { $_ => Unicode::UCD::charinfo($di{$_}) } keys %di;
 };
 
+# add custom categories for certain blocks
+for (values %info) {
+       $_->{category} .= ' Xa' if $_->{block} eq 'Basic Latin';
+       $_->{category} .= ' Xl' if $_->{block} eq 'Latin-1 Supplement';
+}
+
 # output perl code of hash
 # (assume no backslashes or curlies, so we can just q{} w/o escaping)
 print "{\n";
 printf "q{%s}=>[%s],\n", $_, join(',',
        $di{$_},       # glyph code point
-       $charinfo{$_}  # optional additional arguments
-               ? map {"'$_'"} @{ $charinfo{$_} }{qw/name category script/}
+       $info{$_}  # optional additional arguments
+               ? map {"'$_'"} @{ $info{$_} }{qw/name category script/}
                : ()
 ) for sort keys %di;
 print "}\n";