digraphs: page footer

[sheet.git] / rfc1345convert

diff --git a/rfc1345convert b/rfc1345convert
index c1b3861aa65dc77137bba89a191723dd28a15279..412bac14cbbd7cc50afc017301d627f158733dbe 100644 (file)
--- a/rfc1345convert
+++ b/rfc1345convert
@@ -46,9 +46,45 @@ for (@t) {
        $di{$mnem} = hex $char;
 }
 
+# personal addendums
+my @extra;
+if (-r 'shiar.inc.txt') {
+       open my $include, '<:utf8', 'shiar.inc.txt';
+       for (readline $include) {
+               m{^([!"%'-Z_a-z]{2}) (.)} or next;
+               warn("$1 already defined"), next if defined $di{$1};
+               $di{$1} = ord $2;
+               push @extra, $1;
+       }
+}
+warn $@ if $@;
+
+# optionally get unicode character information
+my %info = eval {
+       require Unicode::UCD;
+       map {
+               $_ => Unicode::UCD::charinfo($di{$_})
+                       || { block => '?', category => 'Xn', name => '', script => '' }
+       } keys %di;
+};
+
+# add custom categories for certain blocks
+for (values %info) {
+       $_->{category} .= ' Xa' if $_->{block} eq 'Basic Latin';
+       $_->{category} .= ' Xl' if $_->{block} eq 'Latin-1 Supplement';
+}
+
+# mark unofficial extras as such
+$info{$_}->{category} .= ' Xz' for @extra;
+
 # output perl code of hash
 # (assume no backslashes or curlies, so we can just q{} w/o escaping)
 print "{\n";
-print "q{$_}=>$di{$_},\n" for sort keys %di;
+printf "q{%s}=>[%s],\n", $_, join(',',
+       $di{$_},       # glyph code point
+       $info{$_}  # optional additional arguments
+               ? map {"'$_'"} @{ $info{$_} }{qw/name category script/}
+               : ()
+) for sort keys %di;
 print "}\n";