Question

Just to clarify something first. I am not trying to convert a byte array to a single string. I am trying to convert a byte-array to a string-array.

I am fetching some data from the clipboard using the GetClipboardData API, and then I’m copying the data from the memory as a byte array. When you’re copying multiple files (hence a CF_HDROP clipboard format), I want to convert this byte array into a string array of the files copied.

Here’s my code so far.

//Get pointer to clipboard data in the selected format
var clipboardDataPointer = GetClipboardData(format);

//Do a bunch of crap necessary to copy the data from the memory
//the above pointer points at to a place we can access it.
var length = GlobalSize(clipboardDataPointer);
var @lock = GlobalLock(clipboardDataPointer);

//Init a buffer which will contain the clipboard data
var buffer = new byte[(int)length];

//Copy clipboard data to buffer
Marshal.Copy(@lock, buffer, 0, (int)length);

GlobalUnlock(clipboardDataPointer);

snapshot.InsertData(format, buffer);

Now, here’s my code for reading the buffer data afterwards.

var formatter = new BinaryFormatter();
using (var serializedData = new MemoryStream(buffer))
{
    paths = (string[]) formatter.Deserialize(serializedData);
}

This won’t work, and it’ll crash with an exception saying that the stream doesn’t contain a binary header. I suppose this is because it doesn’t know which type to deserialize into.

I’ve tried looking the Marshal class through. Nothing seems of any relevance.

Answer 1

If the data came through the Win32 API then a string array will just be a sequence of null-terminated strings with a double-null-terminator at the end. (Note that the strings will be UTF-16, so two bytes per character). You’ll basically need to pull the strings out one at a time into an array.

The method you’re looking for here is Marshal.PtrToStringUni, which you should use instead of Marshal.Copy since it works on an IntPtr. It will extract a string, up to the first null character, from your IntPtr and copy it to a string.

The idea would be to continually extract a single string, then advance the IntPtr past the null byte to the start of the next string, until you run out of buffer. I have not tested this, and it could probably be improved (in particular I think there’s a smarter way to detect the end of the buffer) but the basic idea would be:

var myptr = GetClipboardData(format);
var length = GlobalSize(myptr);

var result = new List<string>();

var pos = 0;
while ( pos < length )
{
    var str = Marshal.PtrToStringUni(myptr);
    var count = Encoding.Unicode.GetByteCount(str);

    myptr = IntPtr.Add(myptr, count + 1);
    pos += count + 1;

    result.Add(str);
}

return result.ToArray();

(By the way: the reason your deserialization doesn’t work is because serializing a string[] doesn’t just write out the characters as bytes; it writes out the structure of a string array, including additional internal bits that .NET uses like the lengths, and a binary header with type information. What you’re getting back from the clipboard has none of that present, so it cannot be deserialized.)