Question

The current product I work on is a Windows Service written in C++ and going forward all new functionality will have unit tests written for it. But this creates an interesting problem (for me at least) we do a lot of Win32 calls for various things and behave accordingly, so for the unit tests to be complete it would be nice to test a variety of outputs, not just the current system state.

My question is what is the best way to mock the results of the Win32 calls? I have thought about two different methods:

1) Put all the Win32 calls used into function pointers and pass them into the functions or classes (depending on how many times they are hit) that use them and use this to get mock results.

2) Have lots of #ifdef UNITTEST everywhere and if it is calling my own special methods or have the normal ones called if not.

Am I completely off base here, or missing a fundamental piece of knowledge?

Answer 1

With respect to (2), most Win32 functions which take string parameters already have their common form defined as a macro, e.g. from WinUser.h:

WINUSERAPI
int
WINAPI
MessageBoxA(
    __in_opt HWND hWnd,
    __in_opt LPCSTR lpText,
    __in_opt LPCSTR lpCaption,
    __in UINT uType);
WINUSERAPI
int
WINAPI
MessageBoxW(
    __in_opt HWND hWnd,
    __in_opt LPCWSTR lpText,
    __in_opt LPCWSTR lpCaption,
    __in UINT uType);
#ifdef UNICODE
#define MessageBox  MessageBoxW
#else
#define MessageBox  MessageBoxA
#endif // !UNICODE

You could certainly add a header to your project which redefines the API functions you want to simulate:

#ifdef UNITTEST
#undef MessageBox
#define MessageBox UnitTestMessageBox
#endif

By redefining the name, you can avoid scattering lots of conditional compilation throughout your source code.