X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/f7debc650090d344b01da353f44a49dcf029303a..25c42c7d1649e85339b165f205e620b8eadf6e7b:/keyboard.eng.inc.pl

diff --git a/keyboard.eng.inc.pl b/keyboard.eng.inc.pl
index 0e6a8cb..4b72d08 100644
--- a/keyboard.eng.inc.pl
+++ b/keyboard.eng.inc.pl
@@ -1,35 +1,42 @@
+use strict;
+use warnings;
 use Unicode::Normalize qw( NFKD );
 use Text::Unidecode qw( unidecode );
 
+our %rows;
+my %g; # present group classes
 {
-	rows => [0],
+	rows => [0, 1],
 	def => {
 		'' => {
 			map {
-				my @row = @{$_};
+				my $v = $rows{$_};
 				my $class = (
-					  !defined $row[2] || $row[0] eq $row[2] ? 1 # identical
-					: $row[2] =~ /\A\p{Mn}+\z/ ? 8 # combining accent
-					: NFKD($row[2]) =~ $row[0] ? 2 # decomposed equivalent
-					: $row[2] =~ /^\p{Latin}/ ? 4 # latin script
-					: unidecode($row[2]) =~ /^\W*\Q$row[0]/ ? 5 # transliterated
-					: 7
+					  !defined $v || $_ eq $v ? 1 # identical
+					: $v =~ /\A\p{Mn}+\z/ ? 9 # combining accent
+					: $v =~ /\A[\p{Sk}\p{Lm}]+\z/ ? 8 # modifier symbol
+					: $v =~ /\A[\pM\pP]+\z/ ? 7 # mark
+					: NFKD($v) =~ /\Q$_/ ? 2 # decomposed equivalent
+					: unidecode($v) =~ /\Q$_\E+/i ? 4 # transliterated
+					: $v =~ /^\p{Latin}/ ? 5 # latin script
+					: 6
 				);
-				$row[0] => "g$class"
-			} @rows
+				$g{$class} = 1;
+				$_ => "g$class"
+			} keys %rows
 		},
 	},
-	key => {
-		map {
-			$_->[0] => "$_->[1]<br>$_->[2]"
-		} @rows
-	},
+	key => \%rows,
 	flag => {
 		g1 => ['unaltered', "same results as without modifier"],
 		g2 => ['accented', "decomposes to the original letter with a combining accent"],
-		g4 => ['latin', "a different (accented) latin letter"],
-		g5 => ['similar', "transliterates (mostly) into the unmodified letter"],
-		g7 => ['other', "symbol not directly deducible from key"],
-		g8 => ['combining', "diacritical mark to be combined with a following character"],
+		g4 => ['similar', "transliterates (mostly) into the unmodified letter"],
+		g5 => ['latin', "a different (accented) latin letter"],
+		g6 => ['symbol', "other character not directly deducible from key"],
+		g7 => ['punctuation', "(punctuation) mark"],
+			!$g{9} ? () : (
+		g8 => ['mark', "modifier letter or mark (spacing diacritic)"],
+		g9 => ['combining', "diacritical mark to be combined with a following character"],
+			),
 	},
 }