rm mecab

Author: wolfgitpr

scripts/vcpkg-manifest/vcpkg.json

@@ -3,7 +3,6 @@
   "dependencies": [
     "sdl2",
     "ffmpeg-fake",
-    "mecab",
     "fftw3",
     "yaml-cpp",
     "cpp-pinyin",

src/CMakeLists.txt

@@ -2,14 +2,9 @@ if (NOT DEFINED BUILD_TESTS)
     set(BUILD_TESTS on)
 endif ()
 
-find_package(QT NAMES Qt6 Qt5 COMPONENTS Core Widgets Svg REQUIRED)
+find_package(QT NAMES Qt6 COMPONENTS Core Widgets Svg REQUIRED)
 find_package(Qt${QT_VERSION_MAJOR} COMPONENTS Core Widgets Svg REQUIRED)
 
-if (QT_VERSION_MAJOR GREATER_EQUAL 6)
-    message(STATUS "Qt6 detected, adding Core5Compat module")
-    find_package(Qt${QT_VERSION_MAJOR} COMPONENTS Core5Compat REQUIRED)
-endif ()
-
 add_custom_target(DeployedTargets)
 
 add_subdirectory(apps)

src/apps/MinLabel/CMakeLists.txt

@@ -7,14 +7,12 @@ set(CMAKE_AUTOUIC ON)
 file(GLOB_RECURSE _src *.h *.cpp)
 add_executable(${PROJECT_NAME} ${_src} res.qrc res/qss.qrc)
 
-find_package(mecab CONFIG REQUIRED)
 find_package(cpp-pinyin CONFIG REQUIRED)
 find_package(cpp-kana CONFIG REQUIRED)
 
 target_link_libraries(${PROJECT_NAME} PRIVATE
         Qt${QT_VERSION_MAJOR}::Core
         Qt${QT_VERSION_MAJOR}::Widgets
-        mecab::mecab
         cpp-pinyin::cpp-pinyin
         cpp-kana::cpp-kana
         QMCore SDLPlayback FFmpegDecoder
@@ -39,7 +37,6 @@ if (APPLE)
 endif ()
 
 set(DIRECTORIES_TO_COPY
-        mecabDict
         qss
 )
 

src/apps/MinLabel/gui/TextWidget.cpp

@@ -5,13 +5,11 @@
 #include <QDebug>
 #include <QLineEdit>
 #include <QRegularExpression>
-#include <QTextCodec>
 
 #include <cpp-kana/Kana.h>
 
 TextWidget::TextWidget(QWidget *parent)
-    : QWidget(parent), g2p_man(new Pinyin::Pinyin()), g2p_canton(new Pinyin::Jyutping()),
-      mecabYomi(mecabInit("mecabDict", "yomi")), mecabWakati(mecabInit("mecabDict", "wakati")) {
+    : QWidget(parent), g2p_man(new Pinyin::Pinyin()), g2p_canton(new Pinyin::Jyutping()) {
     wordsText = new QLineEdit();
     wordsText->setPlaceholderText("Enter mandarin here...");
 
@@ -88,19 +86,18 @@ TextWidget::TextWidget(QWidget *parent)
     connect(replaceAction, &QAction::triggered, this, &TextWidget::_q_replaceButtonClicked);
 }
 
-TextWidget::~TextWidget(){};
+TextWidget::~TextWidget() = default;
 
 QString TextWidget::sentence() const {
     QString words = wordsText->text();
-    words.replace("\r\n", " ");
-    words.replace("\n", " ");
+    words.replace(QRegularExpression(R"([\r\n]+)"), " ");
+    words = words.simplified();
     return words;
 }
 
 void TextWidget::_q_pasteButtonClicked() const {
     const auto board = QApplication::clipboard();
-    const QString text = board->text();
-    if (!text.isEmpty()) {
+    if (const QString text = board->text(); !text.isEmpty()) {
         wordsText->setText(text);
     }
 }
@@ -124,7 +121,7 @@ void TextWidget::_q_replaceButtonClicked() const {
             break;
         }
         case 1: {
-            str = QString::fromUtf8(Kana::kanaToRomaji(mecabConvert(jpInput).toUtf8().toStdString(),
+            str = QString::fromUtf8(Kana::kanaToRomaji(jpInput.toUtf8().toStdString(),
                                                        Kana::Error::Default, doubleConsonant->isChecked())
                                         .toStdStr()
                                         .c_str());
@@ -157,7 +154,7 @@ void TextWidget::_q_appendButtonClicked() const {
             break;
         }
         case 1: {
-            str = QString::fromUtf8(Kana::kanaToRomaji(mecabConvert(jpInput).toUtf8().toStdString(),
+            str = QString::fromUtf8(Kana::kanaToRomaji(jpInput.toUtf8().toStdString(),
                                                        Kana::Error::Default, doubleConsonant->isChecked())
                                         .toStdStr()
                                         .c_str());
@@ -197,26 +194,4 @@ void TextWidget::_q_onLanguageComboIndexChanged() {
     for (QCheckBox *control : optionMap[selectedLanguage]) {
         control->show();
     }
-}
-
-MeCab::Tagger *TextWidget::mecabInit(const QString &path, const QString &format) {
-    const QString args = "-O" + format + " -d " + path + " -r" + path + "/mecabrc";
-    return MeCab::createTagger(args.toUtf8());
-}
-
-QString TextWidget::mecabConvert(const QString &input) const {
-    const QTextCodec *codec = QTextCodec::codecForName("GBK");
-    const QByteArray mecabRes = mecabWakati->parse(codec->fromUnicode(input));
-
-    QStringList out;
-    foreach (auto &it, mecabRes.split(' ')) {
-        QString res = codec->toUnicode(mecabYomi->parse(it));
-        const QStringList item = res.split("\t");
-        if (item.size() > 1) {
-            out.append(item[1]);
-        } else if (!item.empty() && item[0] != "") {
-            out.append(item[0]);
-        }
-    }
-    return out.join(" ");
 }

src/apps/MinLabel/gui/TextWidget.h

@@ -12,8 +12,6 @@
 #include <cpp-pinyin/Jyutping.h>
 #include <cpp-pinyin/Pinyin.h>
 
-#include <mecab/mecab.h>
-
 class TextWidget final : public QWidget {
     Q_OBJECT
 public:
@@ -50,11 +48,7 @@ class TextWidget final : public QWidget {
     QScopedPointer<Pinyin::Jyutping> g2p_canton;
 
 private:
-    MeCab::Tagger *mecabYomi;
-    MeCab::Tagger *mecabWakati;
-    [[nodiscard]] QString sentence() const;
-    static MeCab::Tagger *mecabInit(const QString &path = "mecabDict", const QString &format = "wakati");
-    [[nodiscard]] QString mecabConvert(const QString &input) const;
+    QString sentence() const;
 
     void _q_pasteButtonClicked() const;
     void _q_replaceButtonClicked() const;