X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/b5b3537710ed9f73e1c867e0cc27d50439eaf4cd..9ae15ec3a7007abdd580cd30e4d9e9fbb47628da:/tools/mkcharinfo

diff --git a/tools/mkcharinfo b/tools/mkcharinfo
index 577ffb1..ef1e215 100755
--- a/tools/mkcharinfo
+++ b/tools/mkcharinfo
@@ -3,6 +3,7 @@ use 5.010;
 use strict;
 use warnings;
 use utf8;
+no if $] >= 5.018, warnings => 'experimental::smartmatch';
 
 use open OUT => ':utf8', ':std';
 use Data::Dump 'pp';
@@ -10,9 +11,13 @@ use Data::Dump 'pp';
 our $VERSION = '1.00';
 
 my %info = (
+	# prepare presentational string for some control(lish) entries
 	"\xAD"     => {string => '-'},
 	"\x{200E}" => {string => '→'},
 	"\x{200F}" => {string => '←'},
+	"\x{200B}" => {string => '␣'},
+	"\x{200C}" => {string => '|'}, # ISO-9995-7-081 lookalike (alt: ∣ ⊺ ⟙)
+	"\x{200D}" => {string => '⁀'}, # join (alt: ∤ |ͯ ⨝)
 );
 $info{chr $_} //= {} for 32 .. 126;
 
@@ -32,7 +37,9 @@ eval {
 
 eval {
 	require HTML::Entities;
-	while (my ($char, $entity) = each %HTML::Entities::char2entity) {
+	our %char2entity;
+	HTML::Entities->import('%char2entity');
+	while (my ($char, $entity) = each %char2entity) {
 		$entity =~ /[a-zA-Z]/ or next;  # only actual aliases
 		$info{$char}->{html} = substr($entity, 1, -1);
 	}
@@ -58,6 +65,16 @@ for (keys %diinc) {
 	}
 }
 
+eval {
+	# read introducing unicode versions for known characters
+	my $agemap = do 'unicode-age.inc.pl' or die $@ || $!;
+	for my $chr (keys %info) {
+		my $version = $agemap->{ord $chr} or next;
+		$info{$chr}->{class}->{'u-v'.$version}++
+	}
+	1;
+} or warn "Failed including unicode version data $@";
+
 for my $chr (keys %info) {
 	my $cp = ord $chr;
 	# attempt to get unicode character information
@@ -67,7 +84,7 @@ for my $chr (keys %info) {
 			|| { block => '?', category => 'Xn', name => '', script => '' }
 	} or next;
 
-	$info->{$_} = $info{$chr}->{$_} for qw(di html class string);
+	$info->{$_} = $info{$chr}->{$_} for keys %{ $info{$chr} };
 
 	# categorise by unicode types and writing script
 	$info->{class}->{$_}++ for $info->{category};
@@ -92,7 +109,9 @@ for my $chr (keys %info) {
 				? '<'.$info->{unicode10}.'>'  # the old name was much more useful
 				: sprintf('<control U+%04X>', $cp);  # at least identify by value
 			# show descriptive symbols instead of control chars themselves
-			$info->{string} = $cp < 32 ? chr($cp + 0x2400) : chr(0xFFFD);
+			$info->{string} = $cp < 32   ? chr($cp + 0x2400) :
+			                  $cp == 127 ? chr(0x2421) :
+			                               chr(0xFFFD);
 		}
 	}
 
@@ -115,7 +134,7 @@ for my $cp (sort keys %info) {
 say '}';
 
 sub escapeq {
-	my $_ = shift;
+	local $_ = shift;
 	return 'undef' if not defined;
 	s/(['\\])/\\$1/g;
 	return "'$_'";