From 264820f2087e7dee13e358ba703d3dd863ed7faa Mon Sep 17 00:00:00 2001
From: toasted-nutbread <toasted-nutbread@users.noreply.github.com>
Date: Sat, 14 Mar 2020 16:11:07 -0400
Subject: Add more unicode code point ranges

---
 ext/bg/js/japanese.js | 26 ++++++++++++++++++++------
 1 file changed, 20 insertions(+), 6 deletions(-)

(limited to 'ext/bg/js')

diff --git a/ext/bg/js/japanese.js b/ext/bg/js/japanese.js
index 182d5b98..4c2df674 100644
--- a/ext/bg/js/japanese.js
+++ b/ext/bg/js/japanese.js
@@ -84,9 +84,24 @@ const jp = (() => {
     const KATAKANA_RANGE = [0x30a0, 0x30ff];
     const KANA_RANGES = [HIRAGANA_RANGE, KATAKANA_RANGE];
 
-    const CJK_COMMON_RANGE = [0x4e00, 0x9fff];
-    const CJK_RARE_RANGE = [0x3400, 0x4dbf];
-    const CJK_RANGES = [CJK_COMMON_RANGE, CJK_RARE_RANGE];
+    const CJK_UNIFIED_IDEOGRAPHS_RANGE = [0x4e00, 0x9fff];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A_RANGE = [0x3400, 0x4dbf];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B_RANGE = [0x20000, 0x2a6df];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C_RANGE = [0x2a700, 0x2b73f];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D_RANGE = [0x2b740, 0x2b81f];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E_RANGE = [0x2b820, 0x2ceaf];
+    const CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F_RANGE = [0x2ceb0, 0x2ebef];
+    const CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT_RANGE = [0x2f800, 0x2fa1f];
+    const CJK_UNIFIED_IDEOGRAPHS_RANGES = [
+        CJK_UNIFIED_IDEOGRAPHS_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_E_RANGE,
+        CJK_UNIFIED_IDEOGRAPHS_EXTENSION_F_RANGE,
+        CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT_RANGE
+    ];
 
     const ITERATION_MARK_CODE_POINT = 0x3005;
 
@@ -95,8 +110,7 @@ const jp = (() => {
         HIRAGANA_RANGE,
         KATAKANA_RANGE,
 
-        CJK_COMMON_RANGE,
-        CJK_RARE_RANGE,
+        ...CJK_UNIFIED_IDEOGRAPHS_RANGES,
 
         [0xff66, 0xff9f], // Halfwidth katakana
 
@@ -119,7 +133,7 @@ const jp = (() => {
     // Character code testing functions
 
     function isCodePointKanji(codePoint) {
-        return isCodePointInRanges(codePoint, CJK_RANGES);
+        return isCodePointInRanges(codePoint, CJK_UNIFIED_IDEOGRAPHS_RANGES);
     }
 
     function isCodePointKana(codePoint) {
-- 
cgit v1.2.3