Reading from memory stream to string

tigerswithguitars picture tigerswithguitars · May 4, 2012 · Viewed 133.6k times · Source

I am trying to write an object to an Xml string and take that string and save it to a DB. But first I need to get the string...

    private static readonly Encoding LocalEncoding = Encoding.UTF8;

    public static string SaveToString<T> (T settings)
    {
        Stream stream = null;
        TextWriter writer = null;
        string settingsString = null;

        try
        {
            stream = new MemoryStream();

            var serializer = new XmlSerializer(typeof(T));

            writer = new StreamWriter(stream, LocalEncoding);

            serializer.Serialize(writer, settings);

            var buffer = new byte[stream.Length];

            stream.Read(buffer, 0, (int)stream.Length);

            settingsString = LocalEncoding.GetString(buffer);
        }
        catch(Exception ex)
        {
            // If the action cancels we don't want to throw, just return null.
        }
        finally
        {
            if (stream != null)
                stream.Close();

            if(writer != null)
                writer.Close();
        }

        return settingsString;
    }

This seems to work, the stream gets filled with bytes. But when I come to read it back into the buffer and then into the string... the buffer is filled with '0'! Not sure what I doing wrong here guys.

Answer

Jon Skeet picture Jon Skeet · May 4, 2012

If you'd checked the results of stream.Read, you'd have seen that it hadn't read anything - because you haven't rewound the stream. (You could do this with stream.Position = 0;.) However, it's easier to just call ToArray:

settingsString = LocalEncoding.GetString(stream.ToArray());

(You'll need to change the type of stream from Stream to MemoryStream, but that's okay as it's in the same method where you create it.)

Alternatively - and even more simply - just use StringWriter instead of StreamWriter. You'll need to create a subclass if you want to use UTF-8 instead of UTF-16, but that's pretty easy. See this answer for an example.

I'm concerned by the way you're just catching Exception and assuming that it means something harmless, by the way - without even logging anything. Note that using statements are generally cleaner than writing explicit finally blocks.