frontend: Use system locale with UTF-8 instead of 'C'

frontend/OBSApp.cpp

@@ -282,7 +282,7 @@ string CurrentDateTimeString()
 	struct tm tstruct;
 	char buf[80];
 	tstruct = *localtime(&now);
-	strftime(buf, sizeof(buf), "%Y-%m-%d, %X", &tstruct);
+	strftime(buf, sizeof(buf), "%Y-%m-%d, %H:%M:%S", &tstruct);
 	return buf;
 }
 
@@ -724,11 +724,6 @@ bool OBSApp::InitLocale()
 
 	locale = lang;
 
-	// set basic default application locale
-	if (!locale.empty()) {
-		QLocale::setDefault(QLocale(QString::fromStdString(locale).replace('-', '_')));
-	}
-
 	string englishPath;
 	if (!GetDataFilePath("locale/" DEFAULT_LANG ".ini", englishPath)) {
 		OBSErrorBox(NULL, "Failed to find locale/" DEFAULT_LANG ".ini");
@@ -768,11 +763,6 @@ bool OBSApp::InitLocale()
 			blog(LOG_INFO, "Using preferred locale '%s'", locale_.c_str());
 			locale = locale_;
 
-			// set application default locale to the new chosen one
-			if (!locale.empty()) {
-				QLocale::setDefault(QLocale(QString::fromStdString(locale).replace('-', '_')));
-			}
-
 			return true;
 		}
 
@@ -914,13 +904,6 @@ OBSApp::OBSApp(int &argc, char **argv, profiler_name_store_t *store)
 {
 	installNativeEventFilter(new OBS::NativeEventFilter);
 
-	/* fix float handling */
-#if defined(Q_OS_UNIX)
-	if (!setlocale(LC_NUMERIC, "C")) {
-		blog(LOG_WARNING, "Failed to set LC_NUMERIC to C locale");
-	}
-#endif
-
 #ifndef _WIN32
 	// Add POSIX signal handlers:
 	// * SIGINT

frontend/OBSApp_Themes.cpp

@@ -384,7 +384,7 @@ static vector<OBSThemeVariable> ParseThemeVariables(const char *themeData)
 
 			/* Look for a suffix and mark variable as size if it exists */
 			while (ch < end) {
-				if (!isdigit(*ch) && !isspace(*ch) && *ch != '.') {
+				if (!isdigit((unsigned char)*ch) && !isspace((unsigned char)*ch) && *ch != '.') {
 					var.suffix = QString::fromUtf8(ch, end - ch);
 					var.type = OBSThemeVariable::Size;
 					break;
@@ -581,14 +581,14 @@ static OBSThemeVariable ParseMathVariable(const QHash<QString, OBSThemeVariable>
 	const QByteArray utf8 = value.toUtf8();
 	const char *data = utf8.constData();
 
-	if (isdigit(*data)) {
+	if (isdigit((unsigned char)*data)) {
 		double f = os_strtod(data);
 		var.type = OBSThemeVariable::Number;
 		var.value = f;
 
 		const char *dataEnd = data + utf8.size();
 		while (data < dataEnd) {
-			if (*data && !isdigit(*data) && *data != '.') {
+			if (*data && !isdigit((unsigned char)*data) && *data != '.') {
 				var.suffix = QString::fromUtf8(data, dataEnd - data);
 				var.type = OBSThemeVariable::Size;
 				break;

frontend/cmake/windows/obs.manifest

@@ -11,6 +11,11 @@
             </requestedPrivileges>
         </security>
     </trustInfo>
+    <application>
+	<windowsSettings>
+	    <activeCodePage xmlns="http://schemas.microsoft.com/SMI/2019/WindowsSettings">UTF-8</activeCodePage>
+        </windowsSettings>
+    </application>
     <compatibility xmlns="urn:schemas-microsoft-com:compatibility.v1">
         <application>
             <!-- Windows 10 and Windows 11 -->

frontend/obs-main.cpp

@@ -45,6 +45,7 @@
 #include <shellapi.h>
 #define WIN32_LEAN_AND_MEAN
 #include <windows.h>
+#include <mbctype.h>
 #else
 #include <signal.h>
 #endif
@@ -77,6 +78,7 @@ bool opt_disable_missing_files_check = false;
 string opt_starting_collection;
 string opt_starting_profile;
 string opt_starting_scene;
+bool using_utf8 = false;
 
 bool restart = false;
 bool restart_safe = false;
@@ -416,6 +418,42 @@ static void create_log_file(fstream &logFile)
 	}
 }
 
+static bool set_utf8_locale(void)
+{
+	// Use system locale with UTF-8 codepage (available from Windows 10 version 1803)
+	bool usingUTF8 = !!setlocale(LC_ALL, ".UTF-8");
+
+	/*
+	Fallback to minimal C locale
+	Could use "" for system defaults, but the Windows default ANSI codepages (such as .125x or .9xx)
+	mismatch with UTF-8 that is assumed by many parts of the codebase. "C" only covers ASCII, so no mismatches.
+	*/
+	usingUTF8 = usingUTF8 || !!setlocale(LC_ALL, "C.UTF-8");
+
+#ifdef _WIN32
+	usingUTF8 = usingUTF8 && (_setmbcp(CP_UTF8) == 0);
+#endif
+	if (!usingUTF8) {
+		setlocale(LC_ALL, "C");
+	}
+
+	// fix float handling
+	setlocale(LC_NUMERIC, "C");
+
+	// Copy C runtime locale for C++
+	std::locale defaultLocale(setlocale(LC_ALL, nullptr));
+	std::locale::global(defaultLocale);
+
+	/*
+	system() is already the QLocale default, but just to be explicit about the intention.
+	Unlike CRT and C++ locales above, QLocale doesn't support customization of locale categories and codepages.
+	Ie. We can't enforce decimal point to be a dot and at the same time use user's preferred locale.
+	*/
+	QLocale::setDefault(QLocale::system());
+
+	return usingUTF8;
+}
+
 static auto ProfilerNameStoreRelease = [](profiler_name_store_t *store) {
 	profiler_name_store_free(store);
 };
@@ -533,6 +571,12 @@ static int run_program(fstream &logFile, int argc, char *argv[])
 	qputenv("QT_NO_SUBTRACTOPAQUESIBLINGS", "1");
 
 	OBSApp program(argc, argv, profilerNameStore.get());
+
+#if defined(Q_OS_UNIX)
+	// OBSApp (QApplication) constructor overwrote the locale for unix, so set it again
+	set_utf8_locale();
+#endif
+
 	try {
 		QAccessible::installFactory(accessibleFactory);
 		QFontDatabase::addApplicationFont(":/fonts/OpenSans-Regular.ttf");
@@ -683,6 +727,15 @@ static int run_program(fstream &logFile, int argc, char *argv[])
 			blog(LOG_INFO, "Command Line Arguments: %s", stor.str().c_str());
 		}
 
+		if (!using_utf8) {
+			blog(LOG_DEBUG, "UTF-8 locale is not available on this OS version. "
+					"Skipping regional formatting and sorting.");
+		}
+
+		// Use collation (sorting rules) as indicator of regional settings
+		blog(LOG_DEBUG, "Locale: %s", setlocale(LC_COLLATE, nullptr));
+		blog(LOG_DEBUG, "App language: %s", program.GetLocale());
+
 		if (!program.OBSInit()) {
 			return 0;
 		}
@@ -903,6 +956,7 @@ static void set_process_mitigation_policies()
 
 int main(int argc, char *argv[])
 {
+	using_utf8 = set_utf8_locale();
 #ifndef _WIN32
 	using SignalHandlerCallback = decltype(OBSApp::sigIntSignalHandler);
 

libobs/obs-win-crash-handler.c

@@ -245,7 +245,7 @@ static inline void write_header(struct exception_handler_data *data)
 	time_t now = time(0);
 	struct tm ts;
 	ts = *localtime(&now);
-	strftime(date_time, sizeof(date_time), "%Y-%m-%d, %X", &ts);
+	strftime(date_time, sizeof(date_time), "%Y-%m-%d, %H:%M:%S", &ts);
 
 	const char *obs_bitness;
 	if (sizeof(void *) == 8)

libobs/obs.h

@@ -382,8 +382,11 @@ EXPORT const char *obs_get_version_string(void);
  * and safely copies argv/argc from main(). Subsequent calls do nothing.
  *
  * @param  argc  The count of command line arguments, from main()
- * @param  argv  An array of command line arguments, copied from main() and ends
+ * @param  argv  An array of command line arguments in UTF-8, copied from main() and ends
  *               with NULL.
+ * 
+ *		 On Windows this requires that active code page is UTF-8.
+ *		 For example as a declaration in obs.manifest.
  */
 EXPORT void obs_set_cmdline_args(int argc, const char *const *argv);
 

libobs/util/cf-lexer.c

@@ -71,7 +71,7 @@ static inline void cf_convert_from_escape_literal(char **p_dst, const char **p_s
 
 	/* oct */
 	default:
-		if (isdigit(*src)) {
+		if (isdigit((unsigned char)*src)) {
 			*(dst++) = (char)strtoul(src, NULL, 8);
 			src += 3;
 		}

libobs/util/dstr.c

@@ -43,8 +43,8 @@ int astrcmpi(const char *str1, const char *str2)
 		str2 = astrblank;
 
 	do {
-		char ch1 = (char)toupper(*str1);
-		char ch2 = (char)toupper(*str2);
+		unsigned char ch1 = (unsigned char)toupper((unsigned char)*str1);
+		unsigned char ch2 = (unsigned char)toupper((unsigned char)*str2);
 
 		if (ch1 < ch2)
 			return -1;
@@ -85,8 +85,8 @@ int astrcmp_n(const char *str1, const char *str2, size_t n)
 		str2 = astrblank;
 
 	do {
-		char ch1 = *str1;
-		char ch2 = *str2;
+		unsigned char ch1 = (unsigned char)*str1;
+		unsigned char ch2 = (unsigned char)*str2;
 
 		if (ch1 < ch2)
 			return -1;
@@ -129,8 +129,8 @@ int astrcmpi_n(const char *str1, const char *str2, size_t n)
 		str2 = astrblank;
 
 	do {
-		char ch1 = (char)toupper(*str1);
-		char ch2 = (char)toupper(*str2);
+		unsigned char ch1 = (unsigned char)toupper((unsigned char)*str1);
+		unsigned char ch2 = (unsigned char)toupper((unsigned char)*str2);
 
 		if (ch1 < ch2)
 			return -1;

libobs/util/lexer.c

@@ -29,7 +29,8 @@ int strref_cmp(const struct strref *str1, const char *str2)
 		str2 = astrblank;
 
 	do {
-		char ch1, ch2;
+		unsigned char ch1;
+		unsigned char ch2;
 
 		ch1 = (i < str1->len) ? str1->array[i] : 0;
 		ch2 = *str2;
@@ -53,10 +54,11 @@ int strref_cmpi(const struct strref *str1, const char *str2)
 		str2 = astrblank;
 
 	do {
-		char ch1, ch2;
+		unsigned char ch1;
+		unsigned char ch2;
 
-		ch1 = (i < str1->len) ? (char)toupper(str1->array[i]) : 0;
-		ch2 = (char)toupper(*str2);
+		ch1 = (i < str1->len) ? (unsigned char)toupper((unsigned char)str1->array[i]) : 0;
+		ch2 = (unsigned char)toupper((unsigned char)*str2);
 
 		if (ch1 < ch2)
 			return -1;
@@ -77,7 +79,8 @@ int strref_cmp_strref(const struct strref *str1, const struct strref *str2)
 		return -1;
 
 	do {
-		char ch1, ch2;
+		unsigned char ch1;
+		unsigned char ch2;
 
 		ch1 = (i < str1->len) ? str1->array[i] : 0;
 		ch2 = (i < str2->len) ? str2->array[i] : 0;
@@ -103,10 +106,11 @@ int strref_cmpi_strref(const struct strref *str1, const struct strref *str2)
 		return -1;
 
 	do {
-		char ch1, ch2;
+		unsigned char ch1;
+		unsigned char ch2;
 
-		ch1 = (i < str1->len) ? (char)toupper(str1->array[i]) : 0;
-		ch2 = (i < str2->len) ? (char)toupper(str2->array[i]) : 0;
+		ch1 = (i < str1->len) ? (unsigned char)toupper((unsigned char)str1->array[i]) : 0;
+		ch2 = (i < str2->len) ? (unsigned char)toupper((unsigned char)str2->array[i]) : 0;
 
 		if (ch1 < ch2)
 			return -1;

plugins/text-freetype2/find-font.c

@@ -333,8 +333,8 @@ static inline size_t get_rating(struct font_path_info *info, struct dstr *cmp)
 	size_t num = 0;
 
 	do {
-		char ch1 = (char)toupper(*src);
-		char ch2 = (char)toupper(*dst);
+		unsigned char ch1 = (unsigned char)toupper((unsigned char)*src);
+		unsigned char ch2 = (unsigned char)toupper((unsigned char)*dst);
 
 		if (ch1 != ch2)
 			break;