Make the Unicode data more useful (include U+XXXX)

1 files changed, 11 insertions, 8 deletions

diff --git a/.local/bin/uni b/.local/bin/uni
index 66b9950..72f9b70 100755
--- a/.local/bin/uni
+++ b/.local/bin/uni
@@ -5,14 +5,17 @@ set -e NOTIFY_SHORT `basename $args[0]
 
 func setup {
 	curl 'https://www.unicode.org/Public/UNIDATA/UnicodeData.txt'
-	| sed -E '
-		s/;[^;]*//2g
-		s/\<(.)([A-Z]*)/\1\L\2/2g
-		/^[^;]*;</d
-		/Compatibility/d
-		/Variation Selector/d
-		s/[^;]*/\\u&/
-		s/^\\u([^;]{5})/\\U000\1/
+	| awk '
+	BEGIN             { FS = ";" }
+	$2 == "<control>" { $2 = $11 }
+	$2 !~ /(First|Last)>$/ {
+		while (match($2, /[A-Z]{2,}([^)]|$)/)) {
+			car = substr($2, RSTART, 1)
+			cdr = substr($2, RSTART + 1, RLENGTH - 1)
+			sub(/[A-Z]{2,}([^)]|$)/, car tolower(cdr), $2)
+		}
+		printf "\\u%s;U+%s %s\n", $1, $1, $2
+	}
 	' >$DATA
 }