From 92dc64088ef6dfdb273cf9cca397177d5c7e68c4 Mon Sep 17 00:00:00 2001
From: Mischa POSLAWSKY <perl@shiar.org>
Date: Wed, 21 Feb 2024 20:40:44 +0100
Subject: [PATCH] digraphs: encode data include directly as json

---
 Makefile            |  6 +++---
 digraphs.plp        |  2 +-
 tools/mkdigraphlist | 28 ++++++++++++----------------
 3 files changed, 16 insertions(+), 20 deletions(-)

diff --git a/Makefile b/Makefile
index f9c0565..be94112 100644
--- a/Makefile
+++ b/Makefile
@@ -30,7 +30,7 @@ word/put.js: $(download)
 word/%.min.js: word/%.js
 	uglifyjs -m '' $< -o $@
 
-data: data/digraphs.inc.pl 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.inc.pl data/unicode-sampler word
 
 data/DerivedAge.txt: $(download)
 	tools/wget-ifmodified http://www.unicode.org/Public/UNIDATA/$(@F) $@
@@ -64,7 +64,7 @@ data/digraphs-plan9.txt:
 data/digraphs-plan9.inc.pl: tools/mkdigraphs-plan9 data/digraphs-plan9.txt
 	$(call cmdsave,$^)
 
-data/digraphs.inc.pl: tools/mkdigraphlist data/digraphs-rfc.inc.pl data/digraphs-vim.inc.pl data/digraphs-shiar.inc.pl data/unicode-char.inc.pl
+data/digraphs.json: tools/mkdigraphlist data/digraphs-rfc.inc.pl data/digraphs-vim.inc.pl data/digraphs-shiar.inc.pl data/unicode-char.inc.pl
 	$(call cmdsave,$<)
 
 data/unicode-char.inc.pl: tools/mkcharinfo data/digraphs-rfc.inc.pl data/digraphs-shiar.inc.pl data/unicode-age.inc.pl unicode-table.inc.pl
@@ -156,7 +156,7 @@ data/browser/usage-wm.inc.pl: tools/mkusage-wikimedia data/browser/usage-wm.tsv
 	$(call cmdsave,$^)
 
 clean:
-	-rm data/digraphs.inc.pl
+	-rm data/digraphs.json
 	-rm data/unicode-char.inc.pl
 	-rm data/unicode-age.inc.pl
 	-rm -rf data/font/
diff --git a/digraphs.plp b/digraphs.plp
index 2f57b36..ccb1987 100644
--- a/digraphs.plp
+++ b/digraphs.plp
@@ -38,7 +38,7 @@ say '<p class="aside">Unofficial <span class="u-l2">proposals</span>',
 :>
 
 <:
-my $di = Data('data/digraphs');
+my $di = Data('digraphs');
 
 if (exists $get{v}) {
 	# show characters for inverted mnemonics (vim alternatives)
diff --git a/tools/mkdigraphlist b/tools/mkdigraphlist
index 3bc6ff4..5356987 100755
--- a/tools/mkdigraphlist
+++ b/tools/mkdigraphlist
@@ -5,8 +5,9 @@ use warnings;
 use utf8;
 
 use open OUT => ':encoding(utf-8)', ':std';
+use JSON ();
 
-our $VERSION = '1.06';
+our $VERSION = '1.07';
 
 # import and combine various digraph data
 push @INC, 'data';
@@ -24,18 +25,14 @@ my $di = { %{$vim // {}}, %{$rfc}, %{$extra // {}} };
 my $uninfo = do 'unicode-char.inc.pl'
 	or warn "could not include unicode details: ", $@ // $!;
 
-# output perl code of hash
-# (assume no backslashes or curlies, so we can just q{} w/o escaping)
-print "# automatically generated by $0\n";
-print "use utf8;\n";
-print "+{\n";
-printf '(map {$_=>0} qw{%s}),'."\n", join(' ',
+# output json map of character info
+my %table;
+$table{$_} = 0 for (
 	grep { !defined $di->{$_} }
 	map { substr($_, 1, 1).substr($_, 0, 1) } sort keys %{$di}
 );
-printf "q{%s}=>[%s],\n", s/(?=[\\}])/\\/gr, join(',',
+$table{$_} = [
 	ord $di->{$_},   # original code point
-	map {"'$_'"}
 	$uninfo->{ $di->{$_} }->[1] // '',  # name
 	join(' ',
 		$rfc->{$_}
@@ -45,8 +42,8 @@ printf "q{%s}=>[%s],\n", s/(?=[\\}])/\\/gr, join(',',
 	),
 	($uninfo->{ $di->{$_} }->[0] // '') =~ s/ u-di| u-prop| ex//gr,  # class
 	$uninfo->{ $di->{$_} }->[4] // (),  # string
-) for sort keys %{$di};
-print "}\n";
+] for sort keys %{$di};
+print JSON->new->ascii->canonical->encode(\%table);
 
 __END__
 
@@ -56,8 +53,7 @@ mkdigraphlist - Output character list of combined digraph data
 
 =head1 SYNOPSIS
 
-    mkdigraphlist >digraphs.inc.pl
-    perl -e'$di = do "digraphs.inc.pl"; print chr $di->{DO}->[0]'
+    mkdigraphlist | jq -r '."DO"[0]' | perl -nE 'say chr' # $
 
 =head1 DESCRIPTION
 
@@ -72,9 +68,9 @@ is appended: character name, usage classes, unicode classes,
 and replacement output string.
 For example:
 
- +{
-   AE => [198, 'LATIN CAPITAL LETTER AE', 'u-di', 'Latin Lu Xl u-v11'],
-   EA => 'AE',
+  {
+   "AE" => [198, "LATIN CAPITAL LETTER AE", "u-di", "Latin Lu Xl u-v11"],
+   "EA" => "AE",
   }
 
 =head1 AUTHOR
-- 
2.30.0