digraphs: parse vim :digraphs to indicate extensions

[sheet.git] / tools / mkcharinfo

diff --git a/tools/mkcharinfo b/tools/mkcharinfo
index b58f0c373f6215e828e4ad787683d1cfc23cd114..2c006c5d34fd202cc8aec19648953516796116be 100755 (executable)
--- a/tools/mkcharinfo
+++ b/tools/mkcharinfo
@@ -48,9 +48,10 @@ eval {
 
 my %diinc = (
        'data/digraphs-rfc.inc.pl' => 'u-di',
-       'data/digraphs-shiar.inc.pl' => 'u-prop Xz',
+       'data/digraphs-shiar.inc.pl' => 'u-prop ex',
+       'data/digraphs-vim.inc.pl' => 'u-prop',
 );
-for (keys %diinc) {
+for (sort keys %diinc) {
        -e $_ or next;
        my $di = do $_ or die "Error reading digraphs file $_: ", $@ || $!;
        while (my ($mnem, $cp) = each %$di) {
@@ -82,6 +83,10 @@ for my $chr (keys %info) {
 
        $info->{$_} = $info{$chr}->{$_} for keys %{ $info{$chr} };
 
+       # official digraphs either lose vim flag or gain experimental
+       delete $info->{class}->{'u-prop'} or $info->{class}->{ex}++
+               if $info->{class}->{'u-di'};
+
        # categorise by unicode types and writing script
        $info->{class}->{$_}++ for $info->{category};
        $info->{class}->{$_}++ for $info->{script} || ();
@@ -90,15 +95,15 @@ for my $chr (keys %info) {
        $info->{class}->{Xa}++ if $info->{block} eq 'Basic Latin';
        $info->{class}->{Xl}++ if $info->{block} eq 'Latin-1 Supplement';
 
-       given ($cp) {
-               when ($info->{string}) {
+       {
+               if ($info->{string}) {
                        # keep predefined presentational string
                }
-               when ($info->{combining}) {
+               elsif ($info->{combining}) {
                        # overlay combining accents
                        $info->{string} = chr(9676) . $chr;
                }
-               when (($cp & ~0b1001_1111) == 0 or $cp == 127) {
+               elsif (($cp & ~0b1001_1111) == 0 or $cp == 127) {
                        # control characters (first 32 chars from 0 and 128)
                        # rename to something more descriptive
                        $info->{name} = $info->{unicode10}