Allow 4 bytes UTF-8 (UCS-4 range 00010000-001FFFFF)

tatsuo-ishii · tatsuo-ishii · commit 4c35ec53a9e8 · 2007-03-23T13:51:30.000Z
This is necessary to support JIS X 0213 &lt;--&gt; UTF-8 conversion.
diff --git a/src/backend/utils/mb/Unicode/ucs2utf.pl b/src/backend/utils/mb/Unicode/ucs2utf.pl
@@ -1,6 +1,8 @@
 #
-# $PostgreSQL: pgsql/src/backend/utils/mb/Unicode/ucs2utf.pl,v 1.2 2003/11/29 22:40:01 pgsql Exp $
-# convert UCS-2 to UTF-8
+# Copyright (c) 2001-2007, PostgreSQL Global Development Group
+#
+# $PostgreSQL: pgsql/src/backend/utils/mb/Unicode/ucs2utf.pl,v 1.3 2007/03/23 13:51:30 ishii Exp $
+# convert UCS-4 to UTF-8
 #
 sub ucs2utf {
 	local($ucs) = @_;
@@ -10,11 +12,16 @@ sub ucs2utf {
 		$utf = $ucs;
 	} elsif ($ucs > 0x007f && $ucs <= 0x07ff) {
 		$utf = (($ucs & 0x003f) | 0x80) | ((($ucs >> 6) | 0xc0) << 8);
-	} else {
+	} elsif ($ucs > 0x07ff && $ucs <= 0xffff) {
 		$utf = ((($ucs >> 12) | 0xe0) << 16) | 
 			(((($ucs & 0x0fc0) >> 6) | 0x80) << 8) |
 				(($ucs & 0x003f) | 0x80);
-	}
+	} else {
+	   $utf = ((($ucs >> 18) | 0xf0) << 24) |
+		   (((($ucs & 0x3ffff) >> 12) | 0x80) << 16) |		  
+		   (((($ucs & 0x0fc0) >> 6) | 0x80) << 8) |
+		   (($ucs & 0x003f) | 0x80);
+    }
 	return($utf);
 }
 1;
