实现 txt 词典注释语法和排序规则

data/dictionary/DictionaryTest.cpp

@@ -135,30 +135,38 @@ TEST_F(DictionaryRunfilesTest, TWPhrasesReverseMapping) {
     return map;
   };
 
-  LexiconPtr twPhrases = loadLexicon(twPhrasesFile);
-  LexiconPtr twPhrasesRev = loadLexicon(twPhrasesRevFile);
-  ASSERT_NE(twPhrases, nullptr);
-  ASSERT_NE(twPhrasesRev, nullptr);
-
-  auto twMap = buildMap(twPhrases);
-  auto twRevMap = buildMap(twPhrasesRev);
-
-  for (const auto& entry : twMap) {
-    const std::string& key = entry.first;
-    for (const auto& value : entry.second) {
-      auto it = twRevMap.find(value);
-      EXPECT_TRUE(it != twRevMap.end() && it->second.count(key) > 0)
-          << "Missing reverse mapping: " << key << " -> " << value;
+  try {
+    LexiconPtr twPhrases = loadLexicon(twPhrasesFile);
+    LexiconPtr twPhrasesRev = loadLexicon(twPhrasesRevFile);
+    ASSERT_NE(twPhrases, nullptr);
+    ASSERT_NE(twPhrasesRev, nullptr);
+
+    auto twMap = buildMap(twPhrases);
+    auto twRevMap = buildMap(twPhrasesRev);
+
+    for (const auto& entry : twMap) {
+      const std::string& key = entry.first;
+      for (const auto& value : entry.second) {
+        auto it = twRevMap.find(value);
+        EXPECT_TRUE(it != twRevMap.end() && it->second.count(key) > 0)
+            << "Missing reverse mapping: " << key << " -> " << value;
+      }
     }
-  }
 
-  for (const auto& entry : twRevMap) {
-    const std::string& key = entry.first;
-    for (const auto& value : entry.second) {
-      auto it = twMap.find(value);
-      EXPECT_TRUE(it != twMap.end() && it->second.count(key) > 0)
-          << "Missing reverse mapping: " << key << " -> " << value;
+    for (const auto& entry : twRevMap) {
+      const std::string& key = entry.first;
+      for (const auto& value : entry.second) {
+        auto it = twMap.find(value);
+        EXPECT_TRUE(it != twMap.end() && it->second.count(key) > 0)
+            << "Missing reverse mapping: " << key << " -> " << value;
+      }
     }
+  } catch (const Exception& ex) {
+    FAIL() << "Exception: " << ex.what();
+  } catch (const std::exception& ex) {
+    FAIL() << "std::exception: " << ex.what();
+  } catch (...) {
+    FAIL() << "Unknown exception thrown during reverse mapping check.";
   }
 }
 

data/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

data/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

data/dictionary/JPShinjitaiCharacters.txt

@@ -1,3 +1,10 @@
+# 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

data/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

data/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

data/dictionary/STCharacters.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: STCharacters.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, s2t.json, s2tw.json, s2twp.json
+
 㐷	傌
 㐹	㑶 㐹
 㐽	偑

data/dictionary/STPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: STPhrases.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, s2t.json, s2tw.json, s2twp.json
+
 㓦划	㓦劃
 一丝不挂	一絲不掛
 一了心愿	一了心願

data/dictionary/TSCharacters.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TSCharacters.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, t2s.json, tw2s.json, tw2sp.json
+
 㑮	𫝈
 㑯	㑔
 㑳	㑇

data/dictionary/TSPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TSPhrases.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, t2s.json, tw2s.json, tw2sp.json
+
 一目瞭然	一目了然
 上鍊	上链
 不瞭解	不了解

data/dictionary/TWPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TWPhrases.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: s2twp.json (via TWPhrases.ocd2)
+
 PN結	PN接面
 SQL注入	SQL隱碼攻擊
 SQL注入攻擊	SQL隱碼攻擊

data/dictionary/TWPhrasesRev.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TWPhrasesRev.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: tw2sp.json (via TWPhrasesRev.ocd2)
+
 PN接面	PN結
 SQL隱碼攻擊	SQL注入 SQL注入攻擊
 三極體	三極管

data/dictionary/TWVariants.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TWVariants.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: s2tw.json, s2twp.json, t2tw.json
+
 僞	偽
 啓	啟
 喫	吃

data/dictionary/TWVariantsRevPhrases.txt

@@ -1,3 +1,10 @@
+# Open Chinese Convert (OpenCC) Dictionary
+# File: TWVariantsRevPhrases.txt
+# Format: key	value(s) (values separated by spaces)
+# License: Apache-2.0 (see LICENSE)
+# Source: https://github.com/ByVoid/OpenCC
+# Used in configs: tw2s.json, tw2sp.json, tw2t.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