X-Git-Url: http://git.shiar.net/sheet.git/blobdiff_plain/7b13ef7af2e128af5a6246761c26a2058a48e954..800cd11bec46c84a8187b9180f0a7684e7be7d1c:/tools/mkdigraphs-xorg

diff --git a/tools/mkdigraphs-xorg b/tools/mkdigraphs-xorg
index 90b0d86..67c60a6 100755
--- a/tools/mkdigraphs-xorg
+++ b/tools/mkdigraphs-xorg
@@ -6,6 +6,7 @@ use open IO => ':encoding(utf-8)', ':std';
 use re '/msx';
 use JSON 'decode_json';
 use Data::Dump 'pp';
+use Shiar_Sheet::FormatChar;
 
 our $VERSION = '1.01';
 
@@ -15,9 +16,13 @@ my $symname = eval {
 	return decode_json(readline $keysymh);
 } or die "Could not read keysym definitions: $@\n";
 
-say "# automatically generated by $0";
-say '+{';
+my $vidi = eval {
+	open my $jsfh, '<', 'data/digraphs.json' or die $!;
+	local $/;
+	return JSON->new->decode(readline $jsfh);
+} or warn "Could not read comparison digraphs: $@\n";
 
+my %table;
 while ($_ = readline) {
 	my ($mnem, $chr, $trail) = m/\A <Multi_key> \h (.*?) \h+ : \h "([^"]+)" \h* (.*)/
 		or next;
@@ -27,12 +32,43 @@ while ($_ = readline) {
 		$mnem =~ s{<([^>]+)> \h?}{$symname->{$1} // die "reference to unknown keysym $1\n"}eg;
 		1;
 	} or warn($@), next;
-	$mnem !~ m/[^\x20-\x7F]/ or next;  # skip unicode
-#	(state $seen = {})->{$chr}++ and next;
-	printf "%s => %s,\n", pp($mnem), pp($chr);
+	$mnem =~ m/\A [\x20-\x7F]{2} \z/ or next;  # only interested in two ascii
+	my $alias = (state $seen = {})->{$chr}++;  # assume first is preferred
+	my $cp = ord $chr;
+	my $uninfo = Shiar_Sheet::FormatChar->glyph_info($cp);
+	my $comparison = (
+		!$vidi->{key}->{$mnem} ? 'l3' :  # free
+		$vidi->{key}->{$mnem}->[0] != $cp ? 'l1' :  # conflict
+		$vidi->{key}->{$mnem}->[2] eq 'l5' ? 'l5' :  # rfc
+		'l4'  # any
+	);
+	$table{$mnem} = [
+		$cp,
+		$uninfo->[1] // '',  # name
+		$comparison,
+		$alias ? 'l0 ex' : $uninfo->[0] // '',  # class
+		$uninfo->[4] // (),  # string
+	];
 }
 
-say '}';
+print JSON->new->canonical->indent->encode({
+	title => 'X.Org',
+	key   => \%table,
+	intro => join("\n",
+		'Character mnemonics following compose key ⎄:',
+		'in the X Window System (Shift+AltGr by default).',
+		'Differences from <a href="/digraphs">RFC-1345</a> are indicated.',
+		'Also see <a href="/unicode">common Unicode</a>.',
+	),
+	keywords => [qw( xorg x11 x )],
+	flag  => {
+		'l5' => "matching RFC-1345",
+		'l4' => "matching proposal",
+		'l3' => "unique to Xorg",
+		'l1' => "conflict",
+		'l0 ex' => "duplicate",
+	},
+});
 
 __END__
 
@@ -43,13 +79,13 @@ mkdigraphs-xorg - Output Xorg compose sequences
 =head1 SYNOPSIS
 
 
-    mkdigraphs-xorg /usr/share/X11/locale/en_US.UTF-8/Compose >digraphs-xorg.inc.pl
-    perl -e'$di = do "digraphs-xorg.inc.pl"; print chr $di->{AT}'
+    mkdigraphs-xorg /usr/share/X11/locale/en_US.UTF-8/Compose |
+    jq -r '.key."AT"[0]' | perl -nE 'say chr' # @
 
 =head1 DESCRIPTION
 
 Extracts Multi_key definitions from X11/Xorg Compose.pre include file.
-If successful, Perl code is output resulting in a hash
+If successful, a JSON object is output containing a digraphs list in C<key>
 with Unicode code points keyed by mnemonics.
 Any errors and warnings are given at STDERR.