5 replies to this topic

[stafil]

Hi, all, I have a problem with encode.utf8 with a chars ľščťžýáíé. Please help me. Thx

[Stefan]

Hi stafil, Can you post some code, so we could try to see what's going on? Which firmware version do you have on your Netduino?

[stafil]

I used 4.1 framework.

[stafil]

using (FileStream fStr = new FileStream(@"\SD\aa.txt", FileMode.Open, FileAccess.Read))
                {
                    const int lenBuff = 1;
                    byte[] buff = new byte[lenBuff];
                    int bRead = 0;
                    char[] buffc = new char[lenBuff];
                    string line = "";


                    Debug.Print(DateTime.Now.TimeOfDay.ToString());

                    while ((bRead = fStr.Read(buff, 0, lenBuff)) > 0)
                    {
                             line += new string(Encoding.UTF8.GetChars(buff));
                            try
                            {
                                curSocket.Send(Encoding.UTF8.GetBytes(line), line.Length, SocketFlags.None);
                            }
                            catch (SocketException e)
                            {
                                Debug.Print("THROW");
                                //                   break;
                            }

}
}

[CW2]

Are you sure the file content is UTF8 encoded? If yes, you may need to skip the preamble, if present (0xEF, 0xBB, 0xBF). Also, one byte buffer is not enough for reading UTF8 encoded characters (they can take up to 4 bytes). Try reading and decoding the whole line.

[Stefan W.]

To elaborate further on that: There is no way to read a fixed number of bytes and then decode this as UTF-8, you might always end up in the middle of a codepoint. If the file is too big to fit in memory whole, you could, if the lines have a maximum size that fits in memory, do as CW2 says and read until end-of-line (assuming they use \n for end-of-line and not the more elaborate unicode line/page breaks ) and decode the line, which will work since the "usual" end-of-line character is a single byte, or you really have to buffer and specifically look for codepoint boundaries.