error in XML document. Unexpected XML declaration. XML declaration must be the first node in the document

Question

There is an error in XML document (8, 20). Inner 1: Unexpected XML declaration. The XML declaration must be the first node in the document, and no white space characters are allowed to appear before it.

OK, I understand this error.

How I get it, however, is what perplexes me.

I create the document with Microsoft's Serialize tool. Then, I turn around and attempt to read it back, again, using Microsoft's Deserialize tool.

I am not in control of writing the XML file in the correct format - that I can see.

Here is the single routine I use to read and write.

private string xmlPath = System.Web.Hosting.HostingEnvironment.MapPath(WebConfigurationManager.AppSettings["DATA_XML"]);
private object objLock = new Object();
public string ErrorMessage { get; set; }

public StoredMsgs Operation(string from, string message, FileAccess access) {
    StoredMsgs list = null;
    lock (objLock) {
        ErrorMessage = null;
        try {
            if (!File.Exists(xmlPath)) {
                var root = new XmlRootAttribute(rootName);
                var serializer = new XmlSerializer(typeof(StoredMsgs), root);
                if (String.IsNullOrEmpty(message)) {
                    from = "Code Window";
                    message = "Created File";
                }
                var item = new StoredMsg() {
                    From = from,
                    Date = DateTime.Now.ToString("s"),
                    Message = message
                };
                using (var stream = File.Create(xmlPath)) {
                    list = new StoredMsgs();
                    list.Add(item);
                    serializer.Serialize(stream, list);
                }
            } else {
                var root = new XmlRootAttribute("MessageHistory");
                var serializer = new XmlSerializer(typeof(StoredMsgs), root);
                var item = new StoredMsg() {
                    From = from,
                    Date = DateTime.Now.ToString("s"),
                    Message = message
                };
                using (var stream = File.Open(xmlPath, FileMode.Open, FileAccess.ReadWrite)) {
                    list = (StoredMsgs)serializer.Deserialize(stream);
                    if ((access == FileAccess.ReadWrite) || (access == FileAccess.Write)) {
                        list.Add(item);
                        serializer.Serialize(stream, list);
                    }
                }
            }
        } catch (Exception error) {
            var sb = new StringBuilder();
            int index = 0;
            sb.AppendLine(String.Format("Top Level Error: <b>{0}</b>", error.Message));
            var err = error.InnerException;
            while (err != null) {
                index++;
                sb.AppendLine(String.Format("\tInner {0}: {1}", index, err.Message));
                err = err.InnerException;
            }
            ErrorMessage = sb.ToString();
        }
    }
    return list;
}

Is something wrong with my routine? If Microsoft write the file, it seems to me that it should be able to read it back.

It should be generic enough for anyone to use.

Here is my StoredMsg class:

[Serializable()]
[XmlType("StoredMessage")]
public class StoredMessage {
    public StoredMessage() {
    }
    [XmlElement("From")]
    public string From { get; set; }
    [XmlElement("Date")]
    public string Date { get; set; }
    [XmlElement("Message")]
    public string Message { get; set; }
}

[Serializable()]
[XmlRoot("MessageHistory")]
public class MessageHistory : List<StoredMessage> {
}

The file it generates doesn't look to me like it has any issues.

I saw the solution here:

Error: The XML declaration must be the first node in the document

But, in that case, it seems someone already had an XML document they wanted to read. They just had to fix it.

I have an XML document created my Microsoft, so it should be read back in by Microsoft.

Answer 1

The problem is that you are adding to the file. You deserialize, then re-serialize to the same stream without rewinding and resizing to zero. This gives you multiple root elements:

<?xml version="1.0"?>
<StoredMessage>
</StoredMessage
<?xml version="1.0"?>
<StoredMessage>
</StoredMessage

Multiple root elements, and multiple XML declarations, are invalid according to the XML standard, thus the .NET XML parser throws an exception in this situation by default.

For possible solutions, see XML Error: There are multiple root elements, which suggests you either:

1. Enclose your list of StoredMessage elements in some synthetic outer element, e.g. StoredMessageList.

   This would require you to load the list of messages from the file, add the new message, and then truncate the file and re-serialize the entire list when adding a single item. Thus the performance may be worse than in your current approach, but the XML will be valid.

2. When deserializing a file containing concatenated root elements, create an XML writer using XmlReaderSettings.ConformanceLevel = ConformanceLevel.Fragment and iteratively walk through the concatenated root node(s) and deserialize each one individually as shown, e.g., here. Using ConformanceLevel.Fragment allows the reader to parse streams with multiple root elements (although multiple XML declarations will still cause an error to be thrown).

   Later, when adding a new element to the end of the file using XmlSerializer, seek to the end of the file and serialize using an XML writer returned from XmlWriter.Create(TextWriter, XmlWriterSettings) with XmlWriterSettings.OmitXmlDeclaration = true. This prevents output of multiple XML declarations as explained here.

For option #2, your Operation would look something like the following:

private string xmlPath = System.Web.Hosting.HostingEnvironment.MapPath(WebConfigurationManager.AppSettings["DATA_XML"]);
private object objLock = new Object();
public string ErrorMessage { get; set; }

const string rootName = "MessageHistory";
static readonly XmlSerializer serializer = new XmlSerializer(typeof(StoredMessage), new XmlRootAttribute(rootName));

public MessageHistory Operation(string from, string message, FileAccess access)
{
    var list = new MessageHistory();
    lock (objLock)
    {
        ErrorMessage = null;
        try
        {
            using (var file = File.Open(xmlPath, FileMode.OpenOrCreate))
            {
                list.AddRange(XmlSerializerHelper.ReadObjects<StoredMessage>(file, false, serializer));
                if (list.Count == 0 && String.IsNullOrEmpty(message))
                {
                    from = "Code Window";
                    message = "Created File";
                }
                var item = new StoredMessage()
                {
                    From = from,
                    Date = DateTime.Now.ToString("s"),
                    Message = message
                };
                if ((access == FileAccess.ReadWrite) || (access == FileAccess.Write))
                {
                    file.Seek(0, SeekOrigin.End);
                    var writerSettings = new XmlWriterSettings
                    {
                        OmitXmlDeclaration = true,
                        Indent = true, // Optional; remove if compact XML is desired.
                    };
                    using (var textWriter = new StreamWriter(file))
                    {
                        if (list.Count > 0)
                            textWriter.WriteLine();
                        using (var xmlWriter = XmlWriter.Create(textWriter, writerSettings))
                        {
                            serializer.Serialize(xmlWriter, item);
                        }
                    }
                }
                list.Add(item);
            }
        }
        catch (Exception error)
        {
            var sb = new StringBuilder();
            int index = 0;
            sb.AppendLine(String.Format("Top Level Error: <b>{0}</b>", error.Message));
            var err = error.InnerException;
            while (err != null)
            {
                index++;
                sb.AppendLine(String.Format("\tInner {0}: {1}", index, err.Message));
                err = err.InnerException;
            }
            ErrorMessage = sb.ToString();
        }
    }
    return list;
}

Using the following extension method adapted from Read nodes of a xml file in C#:

public partial class XmlSerializerHelper
{
    public static List<T> ReadObjects<T>(Stream stream, bool closeInput = true, XmlSerializer serializer = null)
    {
        var list = new List<T>();

        serializer = serializer ?? new XmlSerializer(typeof(T));
        var settings = new XmlReaderSettings
        {
            ConformanceLevel = ConformanceLevel.Fragment,
            CloseInput = closeInput,
        };
        using (var xmlTextReader = XmlReader.Create(stream, settings))
        {
            while (xmlTextReader.Read())
            {   // Skip whitespace
                if (xmlTextReader.NodeType == XmlNodeType.Element)
                {
                    using (var subReader = xmlTextReader.ReadSubtree())
                    {
                        var logEvent = (T)serializer.Deserialize(subReader);
                        list.Add(logEvent);
                    }
                }
            }
        }

        return list;
    }    
}

Note that if you are going to create an XmlSerializer using a custom XmlRootAttribute, you must cache the serializer to avoid a memory leak.

Sample fiddle.