X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/f1b14de9ebe722dfb26594fe50d66e1c2db6471f..8e8998a5c2e51159db372eed7e4e293df83ccc91:/tools/mkcharinfo

diff --git a/tools/mkcharinfo b/tools/mkcharinfo
index dfcd13d..df5b595 100755
--- a/tools/mkcharinfo
+++ b/tools/mkcharinfo
@@ -1,6 +1,5 @@
 #!/usr/bin/env perl
-use 5.010;
-use strict;
+use 5.014;
 use warnings;
 use utf8;
 no if $] >= 5.018, warnings => 'experimental::smartmatch';
@@ -8,14 +7,14 @@ no if $] >= 5.018, warnings => 'experimental::smartmatch';
 use open OUT => ':utf8', ':std';
 use Data::Dump 'pp';
 
-our $VERSION = '1.00';
+our $VERSION = '1.02';
 
 my %info = (
 	# prepare presentational string for some control(lish) entries
 	"\xAD"     => {string => '-'},
 	"\x{200E}" => {string => '→'},
 	"\x{200F}" => {string => '←'},
-	"\x{200B}" => {string => '␣'},
+	"\x{200B}" => {string => '␣'}, # nbsp: ~ in TeX
 	"\x{200C}" => {string => '|'}, # ISO-9995-7-081 lookalike (alt: ∣ ⊺ ⟙)
 	"\x{200D}" => {string => '⁀'}, # join (alt: ∤ |ͯ ⨝)
 );
@@ -48,8 +47,8 @@ eval {
 
 my %diinc = (
 	'data/digraphs-rfc.inc.pl' => 'u-di',
-	'data/digraphs-shiar.inc.pl' => 'u-prop ex',
-	'data/digraphs-vim.inc.pl' => 'u-prop',
+	'data/digraphs-shiar.inc.pl' => 'u-prop',
+	'data/digraphs-vim.inc.pl' => 'u-vim',
 );
 for (sort keys %diinc) {
 	-e $_ or next;
@@ -58,8 +57,8 @@ for (sort keys %diinc) {
 		my $cp = $di->{$mnem};
 		length $mnem == 2 or next;  # limit to digraphs
 		my $class = $diinc{$_};
-		$info{chr $cp}->{di} //= $mnem;
-		$info{chr $cp}->{class}->{$class}++;
+		$info{$cp}->{di} //= $mnem;
+		$info{$cp}->{class}->{$class}++;
 	}
 }
 
@@ -84,9 +83,9 @@ 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'};
+	# ignore vim flag in addition to rfc support, replace otherwise
+	$info->{class}->{'u-di'} or $info->{class}->{'u-prop'}++
+		if delete $info->{class}->{'u-vim'};
 
 	# categorise by unicode types and writing script
 	$info->{class}->{$_}++ for $info->{category};