Filenames truncate to only show first character - c++

Question

I'm following this guide from MSDN on how to list the files in a directory (i'm using the current directory). In my case I need to put the information in the message part of my packet (char array of size 1016) to send it to the client. When I print packet.message on both the client and server only the first character of the filenames are shown. What's wrong? Here's a snippet of the relevant section of code:
WIN32_FIND_DATA f;
HANDLE h = FindFirstFile(TEXT("./*.*"), &f);
string file;
int size_needed;
do
{
sprintf(packet.message,"%s", &f.cFileName);
//Send packet
} while(FindNextFile(h, &f));

Answer 1

This is commonly caused by a wide character string being mistakenly treated as an ASCII string. The build is targeting UNICODE and cFileName contains a wide character string, but sprintf() is assuming it is an ASCII string.
FindFirstFile() will be mapped to either FindFirstFileA() or FindFirstFileW() depending if the build is or is not targeting UNICODE.
A solution would be to use FindFirstFileA() and ASCII strings explicitly.
Note that the & is unrequired in the sprintf():
sprintf(packet.message, "%s", f.cFileName);
As the application is consuming strings that are outside of its control (i.e file names) I would recommend using the safer _snprintf() to avoid buffer overruns:
/* From your comment on the question 'packet.message' is a 'char[1016]'
so 'sizeof()' will function correctly. */
if (_snprintf(packet.message, sizeof(packet.message), "%s", f.cFileName) > 0)
{
}

Answer 2

You're using the Unicode version of FindFirstFile, almost guaranteed, Either invoke the narrow version or change the format specifier of your print. Personally I would do the former:
WIN32_FIND_DATAA f;
HANDLE h = FindFirstFileA("./*.*", &f);
string file;
int size_needed;
do
{
sprintf(packet.message,"%s", f.cFileName);
//Send packet
} while(FindNextFileA(h, &f));
FindClose(h);
Alternatively, you can compile with MBCS or regular characters.

Answer 3

As others have mentioned, you are calling the Unicode version of FindFirstFile() and are passing Unicode data to the Ansi sprintf() function. The %s specifier expects Ansi input. You have a few choices to address the issue in your code:
continue using sprintf(), but change the %s specifier to %ls so it will accept Unicode input and convert it to Ansi when writing to your message buffer:
sprintf(packet.message, "%ls", f.cFileName);
This is not ideal, though, because it will use the Ansi encoding of the local machine, which may be different than the Ansi encoding used by the receiving machine.
change your message buffer to use TCHAR instead of char, and then switch to either wsprintf() or _stprintf() instead of sprintf(). Like FindFirstFile(), they will match whatever character format that TCHAR and TEXT() use:
TCHAR message[1016];
wsprintf(packet.message, TEXT("%s"), f.cFileName);
Or:
#include <tchar.h>
_TCHAR message[1016];
_stprintf(packet.message, _T("%s"), f.cFileName);
if you must use a char buffer, then you should accept Unicode data from the API and convert it to UTF-8 for transmission, and then the receiver can convert it back to Unicode and use it as needed.
WIN32_FIND_DATAW f;
HANDLE h = FindFirstFileW(L"./*.*", &f);
if (h)
{
do
{
WideCharToMultiByte(CP_UTF8, 0, f.cFileName, lstrlenW(f.cFileName), packet.message, sizeof(packet.message), NULL, NULL);
//Send packet
} while(FindNextFile(h, &f));
FindClose(h);
}