Sync OpenCC dictionaries, configs, and upstream testcases

helper/readme.txt

@@ -1,13 +1,3 @@
-merge.py
-    for merging dictionary files into a single file (requires Python3)
-
-    merge 'TWPhrasesIT.txt', 'TWPhrasesName.txt', and 'TWPhrasesOther.txt'
-    into a single file 'TWPhrases.txt'
-
-    Run:
-        python3 merge.py
-
-
 reverse.py
     for reversing dictionary keys and values (requires Python3)
 

opencc/config/hk2t.json

@@ -1,5 +1,5 @@
 {
-  "name": "Traditional Chinese (Hong Kong variant) to Traditional Chinese",
+  "name": "Traditional Chinese (Hong Kong variant) to Traditional Chinese (OpenCC Standard)",
   "segmentation": {
     "type": "mmseg",
     "dict": {

opencc/config/tw2t.json

@@ -1,5 +1,5 @@
 {
-  "name": "Traditional Chinese (Taiwan standard) to Traditional Chinese",
+  "name": "Traditional Chinese (Taiwan Standard) to Traditional Chinese (OpenCC Standard)",
   "segmentation": {
     "type": "mmseg",
     "dict": {

opencc/dictionary/HKVariants.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: HKVariants.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: s2hk.json, t2hk.json
+
 僞	偽
 兌	兑
 叄	叁

opencc/dictionary/HKVariantsRevPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: HKVariantsRevPhrases.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: hk2s.json, hk2t.json
+
 一口吃個	一口喫個
 一口吃成	一口喫成
 一家三口	一家三口
@@ -71,6 +78,7 @@
 懸巖	懸巖
 懸巖峭壁	懸巖峭壁
 懸胄	懸冑
+控制台	控制台
 攀巖	攀巖
 支胄	支胄
 教胄	教胄

opencc/dictionary/JPShinjitaiCharacters.txt

@@ -1,7 +1,14 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: JPShinjitaiCharacters.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: jp2t.json
+
 両	兩 輛
 弁	辨 辯 瓣 辦 弁
 御	御 禦
 欠	缺 欠
+浜	濱 浜
 糸	絲 糸
 芸	藝 芸
-浜	濱 浜

opencc/dictionary/JPShinjitaiPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: JPShinjitaiPhrases.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: jp2t.json
+
 一獲千金	一攫千金
 丁寧	叮嚀
 丁重	鄭重

opencc/dictionary/JPVariants.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: JPVariants.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: t2jp.json
+
 乘	乗
 亂	乱
 亙	亘
@@ -10,8 +17,10 @@
 價	価
 儉	倹
 兒	児
+兔	兎
 內	内
 兩	両
+冰	氷
 剎	刹
 剩	剰
 劍	剣
@@ -49,7 +58,6 @@
 壹	壱
 壽	寿
 奧	奥
-奬	奨
 妝	粧
 孃	嬢
 學	学
@@ -158,12 +166,12 @@
 淨	浄
 淺	浅
 渴	渇
-潑	溌
 溪	渓
 溫	温
 溼	湿
 滯	滞
 滿	満
+潑	溌
 潛	潜
 澀	渋
 澤	沢
@@ -185,6 +193,7 @@
 犧	犠
 狀	状
 狹	狭
+獎	奨
 獨	独
 獵	猟
 獸	獣
@@ -203,6 +212,7 @@
 盜	盗
 盡	尽
 碎	砕
+礪	砺
 祕	秘
 祿	禄
 禦	御
@@ -229,10 +239,10 @@
 縣	県
 縱	縦
 總	総
-繫	繋
 繡	繍
 繩	縄
 繪	絵
+繫	繋
 繼	継
 續	続
 纔	才
@@ -257,7 +267,7 @@
 舊	旧
 舍	舎
 荔	茘
-莊	荘
+莊	荘 庄
 莖	茎
 菸	煙
 萊	莱
@@ -312,9 +322,9 @@
 鄉	郷
 酢	醋
 醉	酔
-醱	醗
 醫	医
 醬	醤
+醱	醗
 釀	醸
 釋	釈
 鋪	舗