From 21a79fea6555b094deb710f76b63afd2e08498b6 Mon Sep 17 00:00:00 2001
From: Joe Cheng <joe@joecheng.com>
Date: Wed, 23 Mar 2011 12:57:00 -0700
Subject: [PATCH] Respect full width property of all East Asian characters

Common Japanese punctuation (comma, period) was being treated as half-width, causing cursor positioning to be off. The new code is comprehensive for code points in the Basic Multilingual Plane.
See https://gist.github.com/883802 for the script I used to generate the code.
---
 lib/ace/edit_session.js | 65 ++++++++++++++++++++++++++---------------
 1 file changed, 41 insertions(+), 24 deletions(-)

diff --git a/lib/ace/edit_session.js b/lib/ace/edit_session.js
index 6c33c449..a296b694 100644
--- a/lib/ace/edit_session.js
+++ b/lib/ace/edit_session.js
@@ -962,14 +962,8 @@ var EditSession = function(text, mode) {
 			else if(c == 32) {
 			    arr.push(SPACE);
 			}
-    		// CJK characters
-            else if (
-                c >= 0x3040 && c <= 0x309F || // Hiragana
-                c >= 0x30A0 && c <= 0x30FF || // Katakana
-                c >= 0x4E00 && c <= 0x9FFF || // Single CJK ideographs
-                c >= 0xF900 && c <= 0xFAFF ||
-                c >= 0x3400 && c <= 0x4DBF
-            ) {
+    		// full width characters
+            else if (isFullWidth(c)) {
                 arr.push(CHAR, CHAR_EXT);
             } else {
                 arr.push(CHAR);
@@ -995,14 +989,8 @@ var EditSession = function(text, mode) {
 			if (c == 9) {
 				screenColumn += tabSize;
 			}
-    		// CJK characters
-            else if (
-                c >= 0x3040 && c <= 0x309F || // Hiragana
-                c >= 0x30A0 && c <= 0x30FF || // Katakana
-                c >= 0x4E00 && c <= 0x9FFF || // Single CJK ideographs
-                c >= 0xF900 && c <= 0xFAFF ||
-                c >= 0x3400 && c <= 0x4DBF
-            ) {
+    		// full width characters
+            else if (isFullWidth(c)) {
             	screenColumn += 2;
 			} else {
 				screenColumn += 1;
@@ -1149,14 +1137,8 @@ var EditSession = function(text, mode) {
                         docColumn -= 1;
                     }
                 }
-                // CJK characters
-                else if (
-                    c >= 0x3040 && c <= 0x309F || // Hiragana
-                    c >= 0x30A0 && c <= 0x30FF || // Katakana
-                    c >= 0x4E00 && c <= 0x9FFF || // Single CJK ideographs
-                    c >= 0xF900 && c <= 0xFAFF ||
-                    c >= 0x3400 && c <= 0x4DBF
-                ) {
+                // full width characters
+                else if (isFullWidth(c)) {
                     if (remaining >= 2) {
                         remaining -= 2;
                     } else {
@@ -1292,6 +1274,41 @@ var EditSession = function(text, mode) {
         }
         return screenRows;
     }
+    
+    function isFullWidth(c) {
+        return c >= 0x1100 && c <= 0x115F ||
+               c >= 0x11A3 && c <= 0x11A7 ||
+               c >= 0x11FA && c <= 0x11FF ||
+               c >= 0x2329 && c <= 0x232A ||
+               c >= 0x2E80 && c <= 0x2E99 ||
+               c >= 0x2E9B && c <= 0x2EF3 ||
+               c >= 0x2F00 && c <= 0x2FD5 ||
+               c >= 0x2FF0 && c <= 0x2FFB ||
+               c >= 0x3000 && c <= 0x303E ||
+               c >= 0x3041 && c <= 0x3096 ||
+               c >= 0x3099 && c <= 0x30FF ||
+               c >= 0x3105 && c <= 0x312D ||
+               c >= 0x3131 && c <= 0x318E ||
+               c >= 0x3190 && c <= 0x31BA ||
+               c >= 0x31C0 && c <= 0x31E3 ||
+               c >= 0x31F0 && c <= 0x321E ||
+               c >= 0x3220 && c <= 0x3247 ||
+               c >= 0x3250 && c <= 0x32FE ||
+               c >= 0x3300 && c <= 0x4DBF ||
+               c >= 0x4E00 && c <= 0xA48C ||
+               c >= 0xA490 && c <= 0xA4C6 ||
+               c >= 0xA960 && c <= 0xA97C ||
+               c >= 0xAC00 && c <= 0xD7A3 ||
+               c >= 0xD7B0 && c <= 0xD7C6 ||
+               c >= 0xD7CB && c <= 0xD7FB ||
+               c >= 0xF900 && c <= 0xFAFF ||
+               c >= 0xFE10 && c <= 0xFE19 ||
+               c >= 0xFE30 && c <= 0xFE52 ||
+               c >= 0xFE54 && c <= 0xFE66 ||
+               c >= 0xFE68 && c <= 0xFE6B ||
+               c >= 0xFF01 && c <= 0xFF60 ||
+               c >= 0xFFE0 && c <= 0xFFE6;
+    }
 
 }).call(EditSession.prototype);