digraphs: reuse unicode character details

[sheet.git] / tools / mkcharinfo

diff --git a/tools/mkcharinfo b/tools/mkcharinfo
index ef1e2150c535892c30afe789580a549f759c2203..b58f0c373f6215e828e4ad787683d1cfc23cd114 100755 (executable)
--- a/tools/mkcharinfo
+++ b/tools/mkcharinfo
@@ -47,7 +47,8 @@ eval {
 } or warn "Failed importing html entities: $@";
 
 my %diinc = (
-       'digraphs.inc.pl' => 'u-di',
+       'data/digraphs-rfc.inc.pl' => 'u-di',
+       'data/digraphs-shiar.inc.pl' => 'u-prop Xz',
 );
 for (keys %diinc) {
        -e $_ or next;
@@ -55,13 +56,8 @@ for (keys %diinc) {
        while (my ($mnem, $cp) = each %$di) {
                length $mnem == 2 or next;  # limit to digraphs
                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}->{$class}++;
+               $info{chr $cp}->{di} //= $mnem;
+               $info{chr $cp}->{class}->{$class}++;
        }
 }
 
@@ -119,6 +115,7 @@ for my $chr (keys %info) {
 }
 
 # output perl code of hash
+say "# automatically generated by $0";
 say 'use utf8;';
 say '+{';
 for my $cp (sort keys %info) {