Merge pull request #399 from zonble/master

Introduces input macros
openvanilla · Dec 11, 2023 · 2ca3ab7 · 2ca3ab7
2 parents 9e47fc3 + 7e19809
commit 2ca3ab7
Show file tree

Hide file tree

Showing 9 changed files with 404 additions and 5 deletions.
diff --git a/McBopomofo.xcodeproj/project.pbxproj b/McBopomofo.xcodeproj/project.pbxproj
@@ -46,6 +46,7 @@
 		D427F7B4279086DC004A2160 /* InputSourceHelper in Frameworks */ = {isa = PBXBuildFile; productRef = D427F7B3279086DC004A2160 /* InputSourceHelper */; };
 		D427F7B6279086F6004A2160 /* InputSourceHelper in Frameworks */ = {isa = PBXBuildFile; productRef = D427F7B5279086F6004A2160 /* InputSourceHelper */; };
 		D427F7C127908EFC004A2160 /* OpenCCBridge in Frameworks */ = {isa = PBXBuildFile; productRef = D427F7C027908EFC004A2160 /* OpenCCBridge */; };
+		D43FC40B2B23788400ED5A1C /* InputMacro.swift in Sources */ = {isa = PBXBuildFile; fileRef = D43FC40A2B23788400ED5A1C /* InputMacro.swift */; };
 		D44FB74527915565003C80A6 /* Preferences.swift in Sources */ = {isa = PBXBuildFile; fileRef = D44FB74427915555003C80A6 /* Preferences.swift */; };
 		D44FB74A2791B829003C80A6 /* VXHanConvert in Frameworks */ = {isa = PBXBuildFile; productRef = D44FB7492791B829003C80A6 /* VXHanConvert */; };
 		D44FB74D2792189A003C80A6 /* PhraseReplacementMap.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D44FB74B2792189A003C80A6 /* PhraseReplacementMap.cpp */; };
@@ -167,6 +168,7 @@
 		D427F7AC27907B7E004A2160 /* NotifierUI */ = {isa = PBXFileReference; lastKnownFileType = wrapper; name = NotifierUI; path = Packages/NotifierUI; sourceTree = "<group>"; };
 		D427F7B2279086B5004A2160 /* InputSourceHelper */ = {isa = PBXFileReference; lastKnownFileType = wrapper; name = InputSourceHelper; path = Packages/InputSourceHelper; sourceTree = "<group>"; };
 		D427F7BF27908EAC004A2160 /* OpenCCBridge */ = {isa = PBXFileReference; lastKnownFileType = wrapper; name = OpenCCBridge; path = Packages/OpenCCBridge; sourceTree = "<group>"; };
+		D43FC40A2B23788400ED5A1C /* InputMacro.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = InputMacro.swift; sourceTree = "<group>"; };
 		D44FB74427915555003C80A6 /* Preferences.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Preferences.swift; sourceTree = "<group>"; };
 		D44FB7482791B346003C80A6 /* VXHanConvert */ = {isa = PBXFileReference; lastKnownFileType = wrapper; name = VXHanConvert; path = Packages/VXHanConvert; sourceTree = "<group>"; };
 		D44FB74B2792189A003C80A6 /* PhraseReplacementMap.cpp */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = PhraseReplacementMap.cpp; sourceTree = "<group>"; };
@@ -278,6 +280,7 @@
 				D4E569DB27A34CC100AC2CEF /* KeyHandler.mm */,
 				6AE215102A2849BB005A6A02 /* UTF8Helper.cpp */,
 				6AE2150F2A2849BB005A6A02 /* UTF8Helper.h */,
+				D43FC40A2B23788400ED5A1C /* InputMacro.swift */,
 				D456576D279E4F7B00DF6BC9 /* KeyHandlerInput.swift */,
 				D461B791279DAC010070E734 /* InputState.swift */,
 				D427F76B278CA1BA004A2160 /* AppDelegate.swift */,
@@ -671,6 +674,7 @@
 				6A0D4F4515FC0EB100ABF4B3 /* Mandarin.cpp in Sources */,
 				6ACC3D452793701600F1B140 /* ParselessLM.cpp in Sources */,
 				D41355DE278EA3ED005E5CBD /* UserPhrasesLM.cpp in Sources */,
+				D43FC40B2B23788400ED5A1C /* InputMacro.swift in Sources */,
 				6AE215112A2849BB005A6A02 /* UTF8Helper.cpp in Sources */,
 				6ACC3D3F27914F2400F1B140 /* KeyValueBlobReader.cpp in Sources */,
 				B058C5272AC9DF51002EDD66 /* ServiceProvider.swift in Sources */,

diff --git a/Source/Data/Macros.txt b/Source/Data/Macros.txt
@@ -0,0 +1,39 @@
+今天日期 ㄐㄧㄣ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TODAY_SHORT ㄐㄧㄣ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TODAY_MEDIUM ㄐㄧㄣ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TODAY_MEDIUM_ROC ㄐㄧㄣ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TODAY_MEDIUM_CHINESE ㄐㄧㄣ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+昨天日期 ㄗㄨㄛˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_YESTERDAY_SHORT ㄗㄨㄛˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_YESTERDAY_MEDIUM ㄗㄨㄛˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_YESTERDAY_MEDIUM_ROC ㄗㄨㄛˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_YESTERDAY_MEDIUM_CHINESE ㄗㄨㄛˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+明天日期 ㄇㄧㄥˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TOMORROW_SHORT ㄇㄧㄥˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TOMORROW_MEDIUM ㄇㄧㄥˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TOMORROW_MEDIUM_ROC ㄇㄧㄥˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+MACRO@DATE_TOMORROW_MEDIUM_CHINESE ㄇㄧㄥˊ-ㄊㄧㄢ-ㄖˋ-ㄑㄧˊ -8
+現在時刻 ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄎㄜˋ -8
+MACRO@TIME_NOW_SHORT ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄎㄜˋ -8
+MACRO@TIME_NOW_MEDIUM ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄎㄜˋ -8
+現在時間 ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄐㄧㄢ -8
+MACRO@TIME_NOW_SHORT ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄐㄧㄢ -8
+MACRO@TIME_NOW_MEDIUM ㄒㄧㄢˋ-ㄗㄞˋ-ㄕˊ-ㄐㄧㄢ -8
+目前時區 ㄇㄨˋ-ㄑㄧㄢˊ-ㄕˊ-ㄑㄩ -8
+MACRO@TIMEZONE_STANDARD ㄇㄨˋ-ㄑㄧㄢˊ-ㄕˊ-ㄑㄩ -8
+MACRO@TIMEZONE_GENERIC_SHORT ㄇㄨˋ-ㄑㄧㄢˊ-ㄕˊ-ㄑㄩ -8
+所在時區 ㄙㄨㄛˇ-ㄗㄞˋ-ㄕˊ-ㄑㄩ -8
+MACRO@TIMEZONE_STANDARD ㄙㄨㄛˇ-ㄗㄞˋ-ㄕˊ-ㄑㄩ -8
+MACRO@TIMEZONE_GENERIC_SHORT ㄙㄨㄛˇ-ㄗㄞˋ-ㄕˊ-ㄑㄩ -8
+今年干支 ㄐㄧㄣ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+MACRO@THIS_YEAR_GANZHI ㄐㄧㄣ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+去年干支 ㄑㄩˋ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+MACRO@LAST_YEAR_GANZHI ㄑㄩˋ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+明年干支 ㄇㄧㄥˊ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+MACRO@NEXT_YEAR_GANZHI ㄇㄧㄥˊ-ㄋㄧㄢˊ-ㄍㄢ-ㄓ -8
+今年生肖 ㄐㄧㄣ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
+MACRO@THIS_YEAR_CHINESE_ZODIAC ㄐㄧㄣ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
+去年生肖 ㄑㄩˋ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
+MACRO@LAST_YEAR_CHINESE_ZODIAC ㄑㄩˋ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
+明年生肖 ㄇㄧㄥˊ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
+MACRO@NEXT_YEAR_CHINESE_ZODIAC ㄇㄧㄥˊ-ㄋㄧㄢˊ-ㄕㄥ-ㄒㄧㄠˋ -8
diff --git a/Source/Data/Makefile b/Source/Data/Makefile
@@ -8,9 +8,9 @@ data-plain-bpmf.txt: bin/cook-plain-bpmf.py BPMFBase.txt BPMFPunctuations.txt
 	bin/cook-plain-bpmf.py BPMFBase.txt BPMFPunctuations.txt data-plain-bpmf.txt
 
 data.txt: bin/cook.py BPMFBase.txt BPMFMappings.txt BPMFPunctuations.txt \
-          PhraseFreq.txt phrase.occ Symbols.txt \
-          heterophony1.list heterophony2.list  heterophony3.list
-	bin/cook.py PhraseFreq.txt BPMFMappings.txt BPMFBase.txt BPMFPunctuations.txt Symbols.txt $@
+          PhraseFreq.txt phrase.occ Symbols.txt Macros.txt\
+          heterophony1.list heterophony2.list heterophony3.list
+	bin/cook.py PhraseFreq.txt BPMFMappings.txt BPMFBase.txt BPMFPunctuations.txt Symbols.txt Macros.txt $@
 
 PhraseFreq.txt: bin/buildFreq.py phrase.occ exclusion.txt
 	bin/buildFreq.py

diff --git a/Source/Data/Symbols.txt b/Source/Data/Symbols.txt
@@ -18,8 +18,8 @@
 ☢ ㄈㄨˊ-ㄕㄜˋ-ㄒㄧㄥˋ -8
 ≥ ㄉㄚˋ-ㄩˊ-ㄉㄥˇ-ㄩˊ -8
 🉐 ㄉㄜˊ -8
-▼ ㄉㄠˋ-ㄙㄢ-ㄐㄧㄠˇ -8
 ℡ ㄉㄧㄢˋ-ㄏㄨㄚˋ -8
+▼ ㄉㄠˋ-ㄙㄢ-ㄐㄧㄠˇ -8
 ㏒ ㄉㄨㄟˋ-ㄕㄨˋ -8
 ♏ ㄊㄧㄢ-ㄒㄧㄝ -8
 ♎ ㄊㄧㄢ-ㄔㄥˋ -8

diff --git a/Source/Data/bin/cook.py b/Source/Data/bin/cook.py
@@ -200,6 +200,12 @@
             assert len(row) == 3, row
             output.append(tuple(row))
 
+    with open(sys.argv[6]) as macro_file:
+        for line in macro_file:
+            row = line.rstrip().split(" ")
+            assert len(row) == 3, row
+            output.append(tuple(row))
+
     output = convert_vks_rows_to_sorted_kvs_rows(output)
     with open(sys.argv[-1], "w") as fout:
         fout.write(HEADER)

diff --git a/Source/Engine/McBopomofoLM.cpp b/Source/Engine/McBopomofoLM.cpp
@@ -176,6 +176,10 @@ void McBopomofoLM::setExternalConverter(std::function<std::string(std::string)>
     m_externalConverter = externalConverter;
 }
 
+void McBopomofoLM::setMacroConverter(std::function<std::string(std::string)> macroConverter) {
+    m_macroConverter = macroConverter;
+}
+
 std::vector<Formosa::Gramambular2::LanguageModel::Unigram> McBopomofoLM::filterAndTransformUnigrams(const std::vector<Formosa::Gramambular2::LanguageModel::Unigram> unigrams, const std::unordered_set<std::string>& excludedValues, std::unordered_set<std::string>& insertedValues)
 {
     std::vector<Formosa::Gramambular2::LanguageModel::Unigram> results;
@@ -189,17 +193,22 @@ std::vector<Formosa::Gramambular2::LanguageModel::Unigram> McBopomofoLM::filterA
         }
 
         std::string value = originalValue;
+
         if (m_phraseReplacementEnabled) {
             std::string replacement = m_phraseReplacement.valueForKey(value);
             if (!replacement.empty()) {
                 value = replacement;
             }
         }
+        if (m_macroConverter) {
+            std::string replacement = m_macroConverter(value);
+            value = replacement;
+        }
         if (m_externalConverterEnabled && m_externalConverter) {
             std::string replacement = m_externalConverter(value);
             value = replacement;
         }
-        if (insertedValues.find(value) == insertedValues.end()) {
+        if (!value.empty() && insertedValues.find(value) == insertedValues.end()) {
             results.emplace_back(value, unigram.score());
             insertedValues.insert(value);
         }

diff --git a/Source/Engine/McBopomofoLM.h b/Source/Engine/McBopomofoLM.h
@@ -101,6 +101,8 @@ class McBopomofoLM : public Formosa::Gramambular2::LanguageModel {
     bool externalConverterEnabled() const;
     /// Sets a lambda to let the values of unigrams could be converted by it.
     void setExternalConverter(std::function<std::string(std::string)> externalConverter);
+    /// Sets a lambda to convert the macros.
+    void setMacroConverter(std::function<std::string(std::string)> macroConverter);
 
     const std::vector<std::string> associatedPhrasesForKey(const std::string& key);
     bool hasAssociatedPhrasesForKey(const std::string& key);
@@ -128,6 +130,7 @@ class McBopomofoLM : public Formosa::Gramambular2::LanguageModel {
     bool m_phraseReplacementEnabled;
     bool m_externalConverterEnabled;
     std::function<std::string(std::string)> m_externalConverter;
+    std::function<std::string(std::string)> m_macroConverter;
 };
 };