字节数组到struct

Question

I'm having trouble converting the string parts of the byte array.

My struct looks like this:

[StructLayout(LayoutKind.Sequential, Pack = 1)]
struct Message
{
    public int id;

    [MarshalAs(UnmanagedType.ByValTStr, SizeConst = 10)]
    public string text;
}

Creation of test byte array:

private static byte[] CreateMessageByteArray()
{
    int id = 69;
    byte[] intBytes = BitConverter.GetBytes(id);

    string text = "test";
    byte[] stringBytes = GetBytes(text);

    IEnumerable<byte> rv = intBytes.Concat(stringBytes);

    return rv.ToArray();
}

Method to convert my bytearray to a struct:

static T ByteArrayToStructure<T>(byte[] bytes) where T : struct
{
    var handle = GCHandle.Alloc(bytes, GCHandleType.Pinned);
    var result = (T)Marshal.PtrToStructure(handle.AddrOfPinnedObject(), typeof(T));
    handle.Free();
    return result;
}

When I call ByteArrayToStructure with the result from CreateMessageByteArray() I get a struct with an id=60 and text="t".

Why don't I get the whole string eg "test" ?

Edit: This is the code I forgot to pase:

    static byte[] GetBytes(string str)
    {
        byte[] bytes = new byte[str.Length * sizeof(char)];
        System.Buffer.BlockCopy(str.ToCharArray(), 0, bytes, 0, bytes.Length);
        return bytes;
    }

Answer 1

The problem is at this line:

byte[] stringBytes = GetBytes(text);

How are you converting the string to a byte array? You are probably using a Unicode encoding, which will store each character as two bytes, and because your string is in the ASCII set, every other byte will be zero:

byte[] stringBytes = new UnicodeEncoding().GetBytes(text);
// will give you { 't', '\0', 'e', '\0', 's', '\0', 't', '\0' }

These zeroes mislead the marshalling mechanism into assuming they are terminal characters and so the string ends just after the 't' .

Instead, you can use an ASCII encoding (which stores one byte per character):

byte[] stringBytes = new ASCIIEncoding().GetBytes(text);
// will give you { 't', 'e', 's', 't' }
// but will lose non-ASCII character information

Or you can use a UTF8 encoding (which is variable length):

byte[] stringBytes = new UTF8Encoding().GetBytes(text);
// will give you { 't', 'e', 's', 't' }    
// and retain non-ASCII character information, but it's somewhat
// trickier to rebuild the string correctly in case of non-ASCII
// information present

Answer 2

除了其他两个答案之外，如果您希望text字段中的字符串始终为Unicode，则可以在[StructLayout]属性中包含CharSet = CharSet.Unicode

Answer 3

Maybe GetBytes method doesn't work as you expect. This linqpad works fine for me:

void Main()
{
    var result = ByteArrayToStructure<Message>(CreateMessageByteArray());
    result.Dump();
}

[StructLayout(LayoutKind.Sequential, Pack = 1)]
struct Message
{
    public int id;

    [MarshalAs(UnmanagedType.ByValTStr, SizeConst = 10)]
    public string text;
}

private static byte[] CreateMessageByteArray()
{
    int id = 69;
    byte[] intBytes = BitConverter.GetBytes(id);

    string text = "test";
    byte[] stringBytes = Encoding.UTF8.GetBytes(text);

    IEnumerable<byte> rv = intBytes.Concat(stringBytes);

    return rv.ToArray();
}

static T ByteArrayToStructure<T>(byte[] bytes) where T : struct
{
    var handle = GCHandle.Alloc(bytes, GCHandleType.Pinned);
    var result = (T)Marshal.PtrToStructure(handle.AddrOfPinnedObject(), typeof(T));
    handle.Free();
    return result;
}

Output:

id    69 
text  test