From 9445c538bedc82733b7b1c6eb9df7bf9cb54ed64 Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Wed, 21 Feb 2024 21:01:16 +0100
Subject: [PATCH] digraphs/xorg: json include with character info

Similar to primary [vim] data so precedence can be preserved as alias class
(ignored for now).
---
 Makefile              |  4 ++--
 digraphs.plp          |  4 ++--
 tools/mkdigraphs-xorg | 21 +++++++++++++++------
 3 files changed, 19 insertions(+), 10 deletions(-)

diff --git a/Makefile b/Makefile
index be94112..df6cbe5 100644
--- a/Makefile
+++ b/Makefile
@@ -30,7 +30,7 @@ word/put.js: $(download)
 word/%.min.js: word/%.js
 	uglifyjs -m '' $< -o $@
 
-data: data/digraphs.json data/unicode-cover.inc.pl data/countries.inc.pl data/browser data/termcol-xcolor.inc.pl data/digraphs-xorg.inc.pl data/unicode-sampler word
+data: data/digraphs.json data/unicode-cover.inc.pl data/countries.inc.pl data/browser data/termcol-xcolor.inc.pl data/digraphs-xorg.json data/unicode-sampler word
 
 data/DerivedAge.txt: $(download)
 	tools/wget-ifmodified http://www.unicode.org/Public/UNIDATA/$(@F) $@
@@ -53,7 +53,7 @@ data/keysymdef.h: $(download)
 	tools/wget-ifmodified http://gitlab.freedesktop.org/xorg/proto/xorgproto/-/raw/master/include/X11/$(@F) $@
 data/keysymdef.json: tools/mkxkeysymdef data/keysymdef.h
 	$(call cmdsave,$^)
-data/digraphs-xorg.inc.pl: tools/mkdigraphs-xorg data/xorg-compose data/keysymdef.json
+data/digraphs-xorg.json: tools/mkdigraphs-xorg data/xorg-compose data/keysymdef.json
 	$(call cmdsave,$^)
 
 data/digraphs-vim.inc.pl: tools/mkdigraphs-vim
diff --git a/digraphs.plp b/digraphs.plp
index 4ee4d1f..6b1a5c2 100644
--- a/digraphs.plp
+++ b/digraphs.plp
@@ -58,8 +58,8 @@ my @columns = !exists $get{split} ? \@chars2 :
 	([@chars2[0, 1, 3, 4, 6]], [@chars2[2, 5, 7]]);
 
 if ($mode) {
-	my $xorg = Data('data/digraphs-xorg');
-	$_ = [ord $_] for values %{$xorg};
+	my $xorg = Data('digraphs-xorg');
+	$_->[3] = undef for values %{$xorg};  # reset alias classes
 	$xorg->{$_}->[2] = # class = compatibility
 		!$di->{key}->{$_} ? 'l2' :  # free
 		$di->{key}->{$_}->[0] != $xorg->{$_}->[0] ? 'l1' :  # conflict
diff --git a/tools/mkdigraphs-xorg b/tools/mkdigraphs-xorg
index 90b0d86..ec0e862 100755
--- a/tools/mkdigraphs-xorg
+++ b/tools/mkdigraphs-xorg
@@ -15,9 +15,11 @@ my $symname = eval {
 	return decode_json(readline $keysymh);
 } or die "Could not read keysym definitions: $@\n";
 
-say "# automatically generated by $0";
-say '+{';
+# optionally get unicode character information
+my $uninfo = do './data/unicode-char.inc.pl'
+	or warn "could not include unicode details: ", $@ // $!;
 
+my %table;
 while ($_ = readline) {
 	my ($mnem, $chr, $trail) = m/\A <Multi_key> \h (.*?) \h+ : \h "([^"]+)" \h* (.*)/
 		or next;
@@ -27,12 +29,19 @@ while ($_ = 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
+	$table{$mnem} = [
+		ord $chr,
+		$uninfo->{$chr}->[1] // '',  # name
+		0,  # comparison
+		$alias ? 'l0 ex' :
+		($uninfo->{$chr}->[0] // '') =~ s/ u-di| u-prop| ex//gr,  # class
+		$uninfo->{$chr}->[4] // (),  # string
+	];
 }
 
-say '}';
+print JSON->new->canonical->indent->encode(\%table);
 
 __END__
 
-- 
2.30.0