🔒 Locked

[.net] Unicode callbacks

Started by Headkaze Jul 17, 2008 at 5:32 PM 3 replies 1k views

Original Post

Jul 17, 2008 at 5:32 PM

I can't seem to find any information on forcing a callback from a C++ dll to C# to use Unicode strings. I use a StringBuilder type for the function prototype but it doesn't seem to know it's a PWCHAR that is being returned. Is there any way to explicity tell it to expect a Unicode string? Here is the callback type in C++

typedef int (__stdcall *WMP_ERROR)(long ErrorCode, PWCHAR ErrorDesc);

and here it is defined in C#

public delegate void WMP_ERROR(uint ErrorCode, StringBuilder ErrorDesc);

[MarshalAs(UnmanagedType.FunctionPtr)]
WMP_ERROR pError = null;

When I output ErrorDesc.ToString() it will only print the first letter of the string (as it would if it expects an Ansi string.

benryves

Jul 17, 2008 at 8:58 PM

You could try this:

public delegate void WMP_ERROR(	uint ErrorCode,	[MarshalAs(UnmanagedType.LPWStr)] StringBuilder ErrorDesc);

Hope that fixes it!

Headkaze

Jul 17, 2008 at 9:16 PM

Yes it did, thanks! :)

Anon Mike

Jul 18, 2008 at 12:37 PM

Could this be happening because you aren't using PWSTR (or LPWSTR)? PWCHAR is technically a pointer to a single character, not a string.

Headkaze

Jul 19, 2008 at 7:16 AM

The actual issue with it displaying one letter is not because I was using PWCHAR because at the end of the day all I'm passing from the C++ dll to C# is a pointer value which changes from PWCHAR to LPWSTR when it's marshalled back to C#.

The problem was because C# defaults to an ANSI string pointer when using StringBuilder and a callback so when you try to output the string it outputs one letter because the second byte of the UNICODE string was NULL terminating the ANSI string. I believe UNICODE strings use two NULL's for the terminator.

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.