Char* in C++ 6 to C# .Net [modified]

[Lost User]

The default marshalling type for C# strings is Unicode (2 bytes per character) as described here[^]. You will need to modify your C++ code to handle Unicode characters, or add a front end converter from WCHAR to CHAR.

I must get a clever new signature for 2011.

[KeithF]

Just an update on this, I found that if i do the following in C#

[DllImport("mydll.dll", CharSet = CharSet.Auto)]
public static extern int MyMethod(
[param: MarshalAs(UnmanagedType.VBByRefStr)]
ref string strA,
[param: MarshalAs(UnmanagedType.VBByRefStr)]
ref string strB,
[param: MarshalAs(UnmanagedType.VBByRefStr)]
ref string strC);

That my messagebox in MyMethod in the dll shows the first character of the strings eg passing 1, 234, 5 the results are A=1,B=2,C=5

[KeithF]

Thanks Richard, I cant modify the dll as its from a third party and they just sent me on a samepl of the code they use how would I go about adding a front end converter from WCHAR to CHAR? Thanks In Advance

[Lost User]

Just write your own C/C++ function that interfaces to the C# code and takes the three strings and converts them from Unicode to char using the WideCharToMultiByte()[^] function. You can then pass the converted strings to the DLL routines.

I must get a clever new signature for 2011.

[Lost User]

That sounds about right as the Unicode characters for these values are the same as ANSI. However, because the Unicode are 2-byte characters, and the second byte in these cases will be zero, you effectively see a 1 character string in each parameter.

I must get a clever new signature for 2011.

[KeithF]

Hi Richard, Thanks for all the help. After playing around with a few other things i found that this works

[DllImport("MyDll.dll", CharSet = CharSet.Unicode)]
public static extern int MyMethod(
[param: MarshalAs(UnmanagedType.LPStr)]
StringBuilder strA,
[param: MarshalAs(UnmanagedType.LPStr)]
StringBuilder strB,
[param: MarshalAs(UnmanagedType.LPStr)]
StringBuilder strC);

changeg need were to remove the ref keyword, changes string to stringbuilder and the unmanaged type to LPStr

[Ian Shlasko]

I'm no expert on C++ interfacing, so I'd defer to Richard on this, but I think marshaling as UnmanagedType.LPStr might take care of this. According to the MSDN docs, that'd send them through as single-byte, null-terminated strings. Would eliminate the need to convert from Unicode. EDIT: Never mind, looks like you already figured this out :)

Proud to have finally moved to the A-Ark. Which one are you in?
Author of the Guardians Saga (Sci-Fi/Fantasy novels)

[Luc Pattyn]

Some of the replies you got contain too much attributes IMO. Passing strings between C# and native code is pretty easy, you don't even need all those Marshal thingies, a simple CharSet = CharSet.Unicode [ADDED] or CharSet = CharSet.ASCII [ADDED] in your DllImport should do it. You must get rid of the ref keywords obviously, as a string is always passed as a pointer anyway. And if one or more of those strings are meant to be output parameters, the easiest way is by using a StringBuilder. I think I covered all these essentials in this article[^]. :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.

modified on Tuesday, March 22, 2011 12:32 PM

[KeithF]

Hi Luc, I tried what you said as well but I only get see the first character in the C++ method, here is the code:

[DllImport("MyDll.dll", CharSet = CharSet.Unicode)]
public static extern int MyMethod(
string strA,
string strB,
string strC);

I took out the

[param: MarshalAs(UnmanagedType.LPStr)]

but it seems that this is needed

[Luc Pattyn]

sorry, my mistake, if your native code wants ASCII, just say CharSet.ASCII. There is automatic marshaling that will allocate and later free buffers, translate the strings and everything for you. Warning: you are not allowed to remember those string pointers (actually any pointer unless the object remains pinned) and use them after the function return! :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.

[KeithF]

Hi Luc, Thanks that works perfectly, you learn something new every day ;)

[Lost User]

Great stuff. I see that you also got some help from Ian, and Luc who knows far more about this than I do.

I must get a clever new signature for 2011.