Fix passing of non-ASCII characters as command-line arguments

Base/Testing/Cpp/ctkAppLauncherEnvironmentTest.cpp

@@ -52,24 +52,24 @@ void ctkAppLauncherEnvironmentTester::cleanup()
   QStringList originalEnvKeys = ctkAppLauncherEnvironment::envKeys(this->OriginalEnv);
   foreach(const QString& varName, originalEnvKeys)
     {
-    qputenv(varName.toLatin1(), this->OriginalEnv.value(varName).toLatin1());
+    qputenv(varName.toLocal8Bit(), this->OriginalEnv.value(varName).toLocal8Bit());
     }
 }
 
 // ----------------------------------------------------------------------------
 void ctkAppLauncherEnvironmentTester::setEnv(const QString& name, const QString& value)
 {
-  qputenv(name.toLatin1(), value.toLatin1());
+  qputenv(name.toLocal8Bit(), value.toLocal8Bit());
   this->VariableNames.insert(name);
 }
 
 // ----------------------------------------------------------------------------
 void ctkAppLauncherEnvironmentTester::unsetEnv(const QString& name)
 {
 #if defined(_MSC_VER)
-  qputenv(name.toLatin1(), QString("").toLatin1());
+  qputenv(name.toLocal8Bit(), QString().toLocal8Bit());
 #else
-  unsetenv(name.toLatin1());
+  unsetenv(name.toLocal8Bit());
 #endif
 }
 

Base/ctkAppArguments.cpp

@@ -58,7 +58,7 @@ void ctkChar2DArray::setValues(const QStringList& list)
     {
     QString item = d->List.at(index);
     d->Values[index] = new char[item.size() + 1];
-    qstrcpy(d->Values[index], item.toLatin1().data());
+    qstrcpy(d->Values[index], item.toLocal8Bit().data());
     }
 }
 

Base/ctkAppLauncherEnvironment.cpp

@@ -166,9 +166,9 @@ void ctkAppLauncherEnvironment::updateCurrentEnvironment(const QProcessEnvironme
   foreach(const QString& varName, variablesToUnset)
     {
 #if defined(Q_OS_WIN32)
-    bool success = qputenv(varName.toLatin1(), QString("").toLatin1());
+    bool success = qputenv(varName.toLocal8Bit(), QString().toLocal8Bit());
 #else
-    bool success = unsetenv(varName.toLatin1()) == EXIT_SUCCESS;
+    bool success = unsetenv(varName.toLocal8Bit()) == EXIT_SUCCESS;
 #endif
     if (!success)
       {
@@ -180,7 +180,7 @@ void ctkAppLauncherEnvironment::updateCurrentEnvironment(const QProcessEnvironme
   foreach(const QString& varName, envKeys)
     {
     QString varValue = environment.value(varName);
-    bool success = qputenv(varName.toLatin1(), varValue.toLatin1());
+    bool success = qputenv(varName.toLocal8Bit(), varValue.toLocal8Bit());
     if (!success)
       {
       qWarning() << "Failed to set environment variable"

Base/ctkCommandLineParser.cpp

@@ -819,106 +819,70 @@ void ctkCommandLineParser::setStrictModeEnabled(bool strictMode)
   this->Internal->StrictMode = strictMode;
 }
 
+#if defined (_WIN32)
 // --------------------------------------------------------------------------
 void ctkCommandLineParser::convertWindowsCommandLineToUnixArguments(
-  const char *cmd_line, int *argc, char ***argv)
+  PWSTR cmd_line, int* argc, char*** argv)
 {
   if (!cmd_line || !argc || !argv)
     {
     return;
     }
-
-  // A space delimites an argument except when it is inside a quote
-
-  (*argc) = 1;
-
-  size_t cmd_line_len = strlen(cmd_line);
-
-  size_t i;
-  for (i = 0; i < cmd_line_len; i++)
-    {
-    while (isspace(cmd_line[i]) && i < cmd_line_len)
-      {
-      i++;
-      }
-    if (i < cmd_line_len)
-      {
-      if (cmd_line[i] == '\"')
-        {
-        i++;
-        while (cmd_line[i] != '\"' && i < cmd_line_len)
-          {
-          i++;
-          }
-        (*argc)++;
-        }
-      else
-        {
-        while (!isspace(cmd_line[i]) && i < cmd_line_len)
-          {
-          i++;
-          }
-        (*argc)++;
-        }
-      }
-    }
-
-  (*argv) = new char* [(*argc) + 1];
-  (*argv)[(*argc)] = NULL;
-
-  // Set the first arg to be the exec name
-
-  (*argv)[0] = new char [1024];
-#ifdef _WIN32
-  ::GetModuleFileName(0, (*argv)[0], 1024);
-#else
-  (*argv)[0][0] = '\0';
-#endif
-
-  // Allocate the others
-
-  int j;
-  for (j = 1; j < (*argc); j++)
+  *argc = 0;
+  *argv = nullptr;
+
+  // Split the command line to separate arguments
+  int numArgs = 0;
+  LPWSTR* wideArgs = nullptr;
+  // If cmd_line is an empty string then CommandLineToArgvW function returns the path to the current executable file,
+  // so we need to check if the string is empty.
+  if (lstrlenW(cmd_line) > 0)
+  {
+    wideArgs = CommandLineToArgvW(cmd_line, &numArgs);
+    if (wideArgs == nullptr)
     {
-    (*argv)[j] = new char [cmd_line_len + 10];
+      return;
     }
+  }
 
-  // Grab the args
-
-  size_t pos;
-  int argc_idx = 1;
+  // Allocate space for pointers in argv
+  (*argc) = numArgs + 1; // +1 because the first argument is the executable name
+  (*argv) = new char* [numArgs + 1];
+  for (int i = 0; i < numArgs + 1; i++)
+  {
+    (*argv)[i] = nullptr;
+  }
 
-  for (i = 0; i < cmd_line_len; i++)
+  if (wideArgs)
+  {
+    // Convert each argument to UTF8 and save it in argv
+    for (int i = 0; i < numArgs; ++i)
     {
-    while (isspace(cmd_line[i]) && i < cmd_line_len)
+      BOOL lpUsedDefaultChar = false;
+      // Get length
+      int utf8length = WideCharToMultiByte(CP_UTF8, 0, wideArgs[i], -1, NULL, 0, NULL, &lpUsedDefaultChar);
+      char* utf8buffer = new char[utf8length + 1];
+      int retval = WideCharToMultiByte(CP_UTF8, 0, wideArgs[i], -1, utf8buffer, utf8length, NULL, &lpUsedDefaultChar);
+      if (!SUCCEEDED(retval))
       {
-      i++;
-      }
-    if (i < cmd_line_len)
-      {
-      if (cmd_line[i] == '\"')
-        {
-        i++;
-        pos = i;
-        while (cmd_line[i] != '\"' && i < cmd_line_len)
-          {
-          i++;
-          }
-        memcpy((*argv)[argc_idx], &cmd_line[pos], i - pos);
-        (*argv)[argc_idx][i - pos] = '\0';
-        argc_idx++;
-        }
-      else
-        {
-        pos = i;
-        while (!isspace(cmd_line[i]) && i < cmd_line_len)
-          {
-          i++;
-          }
-        memcpy((*argv)[argc_idx], &cmd_line[pos], i - pos);
-        (*argv)[argc_idx][i - pos] = '\0';
-        argc_idx++;
-        }
+        // set to empty string in case of encoding error
+        utf8buffer[0] = '\0';
+        continue;
       }
+      utf8buffer[utf8length] = '\0'; // Make sure the string is null-terminated
+      (*argv)[i + 1] = utf8buffer; // +1 because the first argument is the executable name
     }
+    LocalFree(wideArgs);
+  }
+
+  // Get the application name
+  wchar_t wideBuffer[MAX_PATH];
+  DWORD wideLength = GetModuleFileNameW(NULL, wideBuffer, MAX_PATH);
+  // Convert the wide string to UTF-8
+  int utf8length = WideCharToMultiByte(CP_UTF8, 0, wideBuffer, wideLength, NULL, 0, NULL, NULL);
+  char* utf8buffer = new char[utf8length + 1];
+  WideCharToMultiByte(CP_UTF8, 0, wideBuffer, wideLength, utf8buffer, utf8length, NULL, NULL);
+  utf8buffer[utf8length] = '\0'; // Make sure the string is null-terminated
+  (*argv)[0] = utf8buffer;
  }
+#endif

Base/ctkCommandLineParser.h

@@ -1,6 +1,10 @@
 #ifndef __ctkCommandLineParser_h
 #define __ctkCommandLineParser_h
 
+#if defined (_WIN32)
+#include <windows.h>
+#endif
+
 // Qt includes
 #include <QString>
 #include <QStringList>
@@ -383,14 +387,18 @@ class /*CTK_CORE_EXPORT*/ ctkCommandLineParser
     */
   void setStrictModeEnabled(bool strictMode);
 
+#if defined (_WIN32)
   /**
    * Convert windows-style arguments given as a command-line string
    * into more traditional argc/argv arguments.
+   * If an argument is failed to be retrieved (for example, due to character encoding error)
+   * then the corresponding argv pointer is set to point to an empty string.
    *
    * @note argv[0] will be assigned the executable name using the ::GetModuleFileName function.
    */
   static void convertWindowsCommandLineToUnixArguments(
-    const char *cmd_line, int *argc, char ***argv);
+    PWSTR cmd_line, int* argc, char*** argv);
+#endif
 
 private:
   class ctkInternal;

Base/ctkTest.h

@@ -97,8 +97,8 @@ static void mouseEvent(QTest::MouseAction action, QWidget *widget, Qt::MouseButt
     {
     static const char *mouseActionNames[] =
         { "MousePress", "MouseRelease", "MouseClick", "MouseDClick", "MouseMove" };
-    QString warning = QString::fromLatin1("Mouse event \"%1\" not accepted by receiving widget");
-    QTest::qWarn(warning.arg(QString::fromLatin1(mouseActionNames[static_cast<int>(action)])).toLatin1());
+    QString warning = QLatin1String("Mouse event \"%1\" not accepted by receiving widget");
+    QTest::qWarn(warning.arg(QLatin1String(mouseActionNames[static_cast<int>(action)])).toLocal8Bit());
     }
 }
 

Main.cpp

@@ -186,15 +186,23 @@ int appLauncherMain(int argc, char** argv)
 
 // --------------------------------------------------------------------------
 #ifndef CTKAPPLAUNCHER_WITHOUT_CONSOLE_IO_SUPPORT
+// NOTE: On Windows, wmain should be used to allow non-ASCII characters in command-line arguments
 int main(int argc, char *argv[])
 {
+  // Uncomment the next two lines to stop at application start (to give a chance to connect with a debugger)
+  // std::cout << "Attach debugger and hit Enter" << std::endl;
+  // std::cin.get();
+
   return appLauncherMain(argc, argv);
 }
 #elif defined Q_OS_WIN32
-int __stdcall WinMain(HINSTANCE hInstance,
+int __stdcall wWinMain(HINSTANCE hInstance,
                       HINSTANCE hPrevInstance,
-                      LPSTR lpCmdLine, int nShowCmd)
+                      PWSTR lpCmdLine, int nShowCmd)
 {
+  // Uncomment the next line to stop at application start (to give a chance to connect with a debugger)
+  // int msgboxID = MessageBox(NULL, "Attach your debugger", "Debug", MB_ICONWARNING);
+
   Q_UNUSED(hInstance);
   Q_UNUSED(hPrevInstance);
   Q_UNUSED(nShowCmd);