将 TWPhrasesRev.txt 词典文件 check in，不再自动生成，并添加测试保证 TWPhrases / TWPhrasesRev 文件内容匹配

data/CMakeLists.txt

@@ -12,6 +12,7 @@ set(
   TSCharacters
   TSPhrases
   TWPhrases
+  TWPhrasesRev
   TWVariants
   TWVariantsRevPhrases
   HKVariants
@@ -23,7 +24,6 @@ set(
 
 set(
   DICTS_GENERATED
-  TWPhrasesRev
   TWVariantsRev
   HKVariantsRev
   JPVariantsRev
@@ -59,15 +59,6 @@ set(
   ${DICT_REVERSE_BIN} ${DICT_TWVariantsRev_GENERATING_INPUT} TWVariantsRev.txt
 )
 
-set(
-  DICT_TWPhrasesRev_GENERATING_INPUT
-  ${DICT_DIR}/TWPhrases.txt
-)
-set(
-  DICT_TWPhrasesRev_GENERATING_COMMAND
-  ${DICT_REVERSE_BIN} ${DICT_TWPhrasesRev_GENERATING_INPUT} TWPhrasesRev.txt
-)
-
 set(
   DICT_HKVariantsRev_GENERATING_INPUT
   ${DICT_DIR}/HKVariants.txt

data/dictionary/BUILD.bazel

@@ -11,15 +11,13 @@ package(default_visibility = ["//visibility:public"])
     )
     for txt in [
         "TWVariants",
-        "TWPhrases",
         "HKVariants",
         "JPVariants",
     ]
 ]
 
 TEXT_DICTS = glob(["*.txt"]) + [
     "TWVariantsRev.txt",
-    "TWPhrasesRev.txt",
     "HKVariantsRev.txt",
     "JPVariantsRev.txt",
 ]

data/dictionary/DictionaryTest.cpp

@@ -18,6 +18,9 @@
 
 #include "gtest/gtest.h"
 
+#include <unordered_map>
+#include <unordered_set>
+
 #include "src/Lexicon.hpp"
 #include "src/MarisaDict.hpp"
 #include "src/UTF8Util.hpp"
@@ -44,6 +47,21 @@ class DictionaryTest : public ::testing::Test,
 
 std::string DictionaryTest::runfile_dir_;
 
+class DictionaryRunfilesTest : public ::testing::Test {
+protected:
+  static void SetUpTestSuite() {
+    std::string program_filename = ::testing::internal::GetArgvs().front();
+    size_t suffix_pos = program_filename.find(RUNFILE_SUFFIX);
+    ASSERT_NE(suffix_pos, std::string::npos);
+    runfile_dir_ =
+        program_filename.substr(0, suffix_pos + strlen(RUNFILE_SUFFIX));
+  }
+
+  static std::string runfile_dir_;
+};
+
+std::string DictionaryRunfilesTest::runfile_dir_;
+
 INSTANTIATE_TEST_SUITE_P(
     , DictionaryTest,
     ::testing::Values(
@@ -86,4 +104,62 @@ TEST_P(DictionaryTest, BinaryTest) {
   EXPECT_EQ(dict->GetLexicon()->Length(), txt_lexicon->Length());
 }
 
+TEST_F(DictionaryRunfilesTest, TWPhrasesReverseMapping) {
+  const std::string twPhrasesFile =
+      runfile_dir_ + "/data/dictionary/TWPhrases.txt";
+  const std::string twPhrasesRevFile =
+      runfile_dir_ + "/data/dictionary/TWPhrasesRev.txt";
+
+  auto loadLexicon = [](const std::string& path) -> LexiconPtr {
+    FILE* fp = fopen(UTF8Util::GetPlatformString(path).c_str(), "rb");
+    EXPECT_NE(fp, nullptr) << path;
+    if (fp == nullptr) {
+      return LexiconPtr();
+    }
+    return Lexicon::ParseLexiconFromFile(fp);
+  };
+
+  auto buildMap = [](const LexiconPtr& lexicon)
+      -> std::unordered_map<std::string, std::unordered_set<std::string>> {
+    std::unordered_map<std::string, std::unordered_set<std::string>> map;
+    if (!lexicon) {
+      return map;
+    }
+    for (size_t i = 0; i < lexicon->Length(); ++i) {
+      const DictEntry* entry = lexicon->At(i);
+      auto& values = map[entry->Key()];
+      for (const auto& value : entry->Values()) {
+        values.insert(value);
+      }
+    }
+    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;
+    }
+  }
+
+  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;
+    }
+  }
+}
+
 } // namespace opencc