Qt: work around Qt5's font choice for Chinese

On Windows there are currently two fonts used.

The first, does the Menu, QTreeView and Tooltips
Second is Everything else which is a default font.

From inspecting QApplication::font() at runtime
Windows 10 English: QFont(MS Shell Dlg 2,8.25,-1,5,50,0,0,0,0,0)
Windows 11 Japanese:        MS UI Gothic,9   ,-1,5,50,0,0,0,0,0
Windows 11 Traditional Chinese: PMingLiU,9   ,-1,5,50,0,0,0,0,0
Windows 11 Simplified Chinese:    SimSun,9   ,-1,5,50,0,0,0,0,0
Windows 11 Korean:                 Gulim,9   ,-1,5,50,0,0,0,0,0

I initially investigated dynamically changing the font when
the UI language is English, but this was getting quite messy

Qt6 makes changes to default font in some situations, so this
PR is being narrowed in scope to only effect Chinese font choices.
This change only effects rendering of Latin/Cyrillic characters.
master
Kyle Kienapfel 2022-10-01 15:27:23 +07:00
parent 2a752bbd64
commit 1dba5fab62
1 changed files with 16 additions and 0 deletions

@ -261,6 +261,18 @@ static QString PrettyProductName() {
return QSysInfo::prettyProductName();
}
#ifdef _WIN32
static void OverrideWindowsFont() {
// Qt5 chooses these fonts on Windows and they have fairly ugly alphanumeric/cyrllic characters
// Asking to use "MS Shell Dlg 2" gives better other chars while leaving the Chinese Characters.
const QString startup_font = QApplication::font().family();
const QStringList ugly_fonts = {QStringLiteral("SimSun"), QStringLiteral("PMingLiU")};
if (ugly_fonts.contains(startup_font)) {
QApplication::setFont(QFont(QStringLiteral("MS Shell Dlg 2"), 9, QFont::Normal));
}
}
#endif
bool GMainWindow::CheckDarkMode() {
#ifdef __linux__
const QPalette test_palette(qApp->palette());
@ -4134,6 +4146,10 @@ int main(int argc, char* argv[]) {
QCoreApplication::setAttribute(Qt::AA_DontCheckOpenGLContextThreadAffinity);
QApplication app(argc, argv);
#ifdef _WIN32
OverrideWindowsFont();
#endif
// Workaround for QTBUG-85409, for Suzhou numerals the number 1 is actually \u3021
// so we can see if we get \u3008 instead
// TL;DR all other number formats are consecutive in unicode code points