How to correctly retrieve Unicode text from a control? - c++

Question

Here is an example of Unicode: I avoided to use a win32 application for for brevity sake:
In main I created an edit control, a button to retrieve text from it and adds it to the listbox when pressed. So I used an object of struct MSG and blocked in a while loop peeking the messages from the message queue.
int main(){
// An edit control where I can add any unicode text to.
HWND hEdit = CreateWindowExW(WS_EX_CLIENTEDGE,
L"Edit", L"你好！ 你好吗？", WS_BORDER | WS_VISIBLE | WS_SYSMENU,
200, 100, 250, 70, 0, 0, GetModuleHandle(NULL), NULL);
// A listobx to receive the content of the edit control when pressing the button get text.
HWND hLstBx = CreateWindowExW(WS_EX_CLIENTEDGE,
L"Listbox", NULL, WS_BORDER | WS_VISIBLE | WS_OVERLAPPEDWINDOW,
500, 100, 170, 400, 0, 0, GetModuleHandle(NULL), NULL);
// A button when pressed adds the content of the edit to the listbox.
HWND hGetText = CreateWindowExW(WS_EX_CLIENTEDGE,
L"Button", L"中文", WS_BORDER | WS_VISIBLE,
450, 300, 100, 70, 0, 0, GetModuleHandle(NULL), NULL);
// msg struct to pass to GetMessage to receive the messages from the queue.
MSG msg;
// blocking and getting messages from the queue.
while (GetMessageW(&msg, 0, 0, 0)) {
// some virtual keys translation.
TranslateMessage(&msg);
// sending the message to the specified window.
DispatchMessageW(&msg);
// Now after the messages sent to the target Window I check which control the message has been passed to, So if it is the button then:
if (msg.message == WM_LBUTTONDOWN &&
msg.hwnd == hGetText) {
std::wstring wstrBuff;
int txtLen = SendMessageW(hEdit, WM_GETTEXTLENGTH, 0, 0);
// SendMessageW(hEdit, WM_GETTEXT, txtLen + 1, (LPARAM)wstrBuff.c_str());
// MessageBoxW(0, wstrBuff.c_str(), 0, 0);
wchar_t lpTxt[MAX_PATH];
SendMessageW(hEdit, WM_GETTEXT, MAX_PATH, (LPARAM)lpTxt);
SendMessageW(hLstBx, LB_ADDSTRING, 0, (LPARAM)lpTxt);
MessageBoxW(0, lpTxt, L"你好！ 你好吗？", 0);
//delete[]lpTxt;
}
}
std::wcout << std::endl << std::endl;
std::wcin.get();
return 0;
}
Every thing works fine except: If I un-comment the lines above I get run-time error facing the assertion message showing me the txtLen and the size of the text of the edit control. Is this because there's some string overlapping?
If I enter a small text it works fine but with a text about 14 characters I get the error.
Also is that the right way to pass std::wstring.c_str() to SendMessageW() to get the text?
One last question: How to correctly and effectively retrieve Unicode text from a control? How to use LPWSTR with dynamic memory: I don't want to exhaust stack.
NB: I saved the source file as utf8 /BOM otherwise I get unreadable characters. Thanks to the members who helped me about that.

Answer 1

Sending (LPARAM)wstrBuff.c_str() will return a pointer to read-only buffer with a single null symbol, not a buffer of txtLen + 1 symbols. If you are using latest VS (with C++17 standard support) you can modify your code to supply a proper pointer:
std::wstring wstrBuff;
wstrBuff.resize(txtLen + 1);
const LRESULT copied_symbols_count = SendMessageW
(
hEdit
, WM_GETTEXT
, static_cast<WPARAM>(wstrBuff.size())
, reinterpret_cast<LPARAM>(wstrBuff.data())
);
if(copied_symbols_count == txtLen)
{
assert(L'\0' == wstrBuff.back());
wstrBuff.pop_back(); // get rid of terminating null
}
else
{
wstrBuff.clear(); // something went wrong...
}
Note that C++17 standard adds non-const-qualified wstring::data() method that can be safely used to obtain a pointer to writable buffer.