I am generating an XML document from a StringBuilder, basically something like:
string.Format(\"{0}
{1}
Without your actual data or source, it will be hard for us to diagnose what is going wrong. However, I can make a few suggestions:
Given what you wrote, I suspect whatever converts the database data to XML is broken; it's propagating non-XML characters.
Create some database entries with non-XML characters (NULs, DELs, control characters, et al.) and run your XML converter on it. Output the XML to a file and look at it in a hex editor. If this contains non-XML characters, your converter is broken. Fix it or, if you cannot, create a preprocessor that rejects output with such characters.
If the converter output looks good, the problem is in your XML consumer; it's inserting non-XML characters somewhere. You will have to break your consumption process into separate steps, examine the output at each step, and narrow down what is introducing the bad characters.
Update: I just ran into an example of this myself! What was happening is that the producer was encoding the XML as UTF16 and the consumer was expecting UTF8. Since UTF16 uses 0x00 as the high byte for all ASCII characters and UTF8 doesn't, the consumer was seeing every second byte as a NUL. In my case I could change encoding, but suggested all XML payloads start with a BOM.
I also get the same error in an ASP.NET application when I saved some unicode data (Hindi) in the Web.config file and saved it with "Unicode" encoding.
It fixed the error for me when I saved the Web.config file with "UTF-8" encoding.
I'm using IronPython here (same as .NET API) and reading the file as UTF-8 in order to properly handle the BOM fixed the problem for me:
xmlFile = Path.Combine(directory_str, 'file.xml')
doc = XPathDocument(XmlTextReader(StreamReader(xmlFile.ToString(), Encoding.UTF8)))
It would work as well with the XmlDocument
:
doc = XmlDocument()
doc.Load(XmlTextReader(StreamReader(xmlFile.ToString(), Encoding.UTF8)))
In my case, it took some digging, but found it.
My Context
I'm looking at exception/error logs from the website using Elmah. Elmah returns the state of the server at the of time the exception, in the form of a large XML document. For our reporting engine I pretty-print the XML with XmlWriter.
During a website attack, I noticed that some xmls weren't parsing and was receiving this '.', hexadecimal value 0x00, is an invalid character.
exception.
NON-RESOLUTION: I converted the document to a byte[]
and sanitized it of 0x00, but it found none.
When I scanned the xml document, I found the following:
...
<form>
...
<item name="SomeField">
<value
string="C:\boot.ini�.htm" />
</item>
...
There was the nul byte encoded as an html entity �
!!!
RESOLUTION: To fix the encoding, I replaced the �
value before loading it into my XmlDocument
, because loading it will create the nul byte and it will be difficult to sanitize it from the object. Here's my entire process:
XmlDocument xml = new XmlDocument();
details.Xml = details.Xml.Replace("�", "[0x00]"); // in my case I want to see it, otherwise just replace with ""
xml.LoadXml(details.Xml);
string formattedXml = null;
// I have this in a helper function, but for this example I have put it in-line
StringBuilder sb = new StringBuilder();
XmlWriterSettings settings = new XmlWriterSettings {
OmitXmlDeclaration = true,
Indent = true,
IndentChars = "\t",
NewLineHandling = NewLineHandling.None,
};
using (XmlWriter writer = XmlWriter.Create(sb, settings)) {
xml.Save(writer);
formattedXml = sb.ToString();
}
LESSON LEARNED: sanitize for illegal bytes using the associated html entity, if your incoming data is html encoded on entry.
As kind of a late answer:
I've had this problem with SSRS ReportService2005.asmx when uploading a report.
Public Shared Sub CreateReport(ByVal strFileNameAndPath As String, ByVal strReportName As String, ByVal strReportingPath As String, Optional ByVal bOverwrite As Boolean = True)
Dim rs As SSRS_2005_Administration_WithFOA = New SSRS_2005_Administration_WithFOA
rs.Credentials = ReportingServiceInterface.GetMyCredentials(strCredentialsURL)
rs.Timeout = ReportingServiceInterface.iTimeout
rs.Url = ReportingServiceInterface.strReportingServiceURL
rs.UnsafeAuthenticatedConnectionSharing = True
Dim btBuffer As Byte() = Nothing
Dim rsWarnings As Warning() = Nothing
Try
Dim fstrStream As System.IO.FileStream = System.IO.File.OpenRead(strFileNameAndPath)
btBuffer = New Byte(fstrStream.Length - 1) {}
fstrStream.Read(btBuffer, 0, CInt(fstrStream.Length))
fstrStream.Close()
Catch ex As System.IO.IOException
Throw New Exception(ex.Message)
End Try
Try
rsWarnings = rs.CreateReport(strReportName, strReportingPath, bOverwrite, btBuffer, Nothing)
If Not (rsWarnings Is Nothing) Then
Dim warning As Warning
For Each warning In rsWarnings
Log(warning.Message)
Next warning
Else
Log("Report: {0} created successfully with no warnings", strReportName)
End If
Catch ex As System.Web.Services.Protocols.SoapException
Log(ex.Detail.InnerXml.ToString())
Catch ex As Exception
Log("Error at creating report. Invalid server name/timeout?" + vbCrLf + vbCrLf + "Error Description: " + vbCrLf + ex.Message)
Console.ReadKey()
System.Environment.Exit(1)
End Try
End Sub ' End Function CreateThisReport
The problem occurs when you allocate a byte array that is at least 1 byte larger than the RDL (XML) file.
Specifically, I used a C# to vb.net converter, that converted
btBuffer = new byte[fstrStream.Length];
into
btBuffer = New Byte(fstrStream.Length) {}
But because in C# the number denotes the NUMBER OF ELEMENTS in the array, and in VB.NET, that number denotes the UPPER BOUND of the array, I had an excess byte, causing this error.
So the problem's solution is simply:
btBuffer = New Byte(fstrStream.Length - 1) {}
To add to Sonz's answer above, following worked for us.
//Instead of
XmlString.Replace("�", "[0x00]");
// use this
XmlString.Replace("\x00", "[0x00]");