Paul Selles

Computers and cats

Parsing Xml with Invalid Characters in C#

The Problem

I’ve stumbled upon an interesting predicament. I need to parse some SQL relationships from an automatically generated XML file that contains invalid characters. Here is an example XML file that I will use to highlight the problem that I saw:

<?xml version="1.0" encoding="utf-8"?>
<Cats>
	<Cat Id="1" Type="Tabby">
		<Property Name="Fur" Value="Coarse"/>
		<Property Name="Color" Value="Orange" />
		<Part Name="Paws">
			<Property Name="Claws" Value="Very sharp" />
		</Part>
		<Part Name="Nose">
			<Property Name="Cute" Value="true" />
		</Part>
		<Info>
			I have an invalid character.&#x13;
		</Info>
	</Cat>
	<Cat Id="2" Type="Short hair">
		<Property Name="Fur" Value="Soft"/>
		<Property Name="Color" Value="Black" />
		<Part Name="Paws">
			<Property Name="Polydactyl" Value="true" />
			<Property Name="Claws" Value="Sharp" />
		</Part>
		<Part Name="Nose">
			<Property Name="Cute" Value="true" />
		</Part>
		<Info>
			I don't have an invalid character.
		</Info>
	</Cat>
</Cats>

So above we have a small XML file cataloging my two cats. Within the Info tags you may notice that the first Cat entry has a superfluous character, 0x13; this falls outside of the valid XML character set [1]. The W3C recommendation, however, is no guarantee that every XML file that you encounter will follow the recommendations to a tee.

In C# we can try using the two most common XML parsing libraries System.Xml and System.Xml.Linq to import the XML file to the XmlDocument and XDocument objects using their respective Load functions [2][3]. If we try to do this we can expect to see the following exception:

‘ ‘, hexadecimal value 0x13, is an invalid character. Line 13, position 35.

The Solution

There is a workaround that is made possible with the lightweight disposable XmlReader class and the XmlReaderSettings support class that allows us to customize the behavior of XmlReader [4][5]. The XmlReaderSettings property that interests us the most is the Boolean CheckCharacters. Setting CheckCharacters property to false will let us read the XML document without verifying if the processed text data is within the valid XML character set [6]. The XmlDocument and XDocument objects can now be loaded from the XmlReader incident free:

static XmlDocument ReadXmlDocumentWithInvalidCharacters(string filename)
{
    XmlDocument xmlDocument = new XmlDocument();

    XmlReaderSettings xmlReaderSettings = new XmlReaderSettings { CheckCharacters = false };

    using (XmlReader xmlReader = XmlReader.Create(filename, xmlReaderSettings))
    {
        // Load our XmlDocument
        xmlReader.MoveToContent();
        xmlDocument.Load(xmlReader);
    }

    return xmlDocument;
}
static XDocument ReadXDocumentWithInvalidCharacters(string filename)
{
    XDocument xDocument = null;

    XmlReaderSettings xmlReaderSettings = new XmlReaderSettings { CheckCharacters = false };

    using (XmlReader xmlReader = XmlReader.Create(filename, xmlReaderSettings))
    {
        // Load our XDocument
        xmlReader.MoveToContent();
        xDocument = XDocument.Load(xmlReader);
    }

    return xDocument;
}

Once we load our XML code then we are free to parse it, and since I prefer working with the System.Xml.Linq library, that’s all I will do:

static void PrintXDocument(XDocument xDocument)
{
    foreach (XElement xElement in xDocument.Elements(xDocument.Root.Name).DescendantsAndSelf())
    {
           Console.Write(("".PadRight(xElement.Ancestors().Count() * 4) +
            (xElement.HasElements == true || string.IsNullOrEmpty(xElement.Value) ?
                xElement.Name.LocalName :
                (xElement.Name.LocalName + " \"" + xElement.Value.Trim() + "\""))));

        foreach (XAttribute xAttribute in xElement.Attributes())
            Console.Write(" " + xAttribute.Name.LocalName + "=\"" + xAttribute.Value + "\"");

        Console.WriteLine();
    }

    Console.ReadLine();
}

And the results:

Cats
Cat Id=”1″ Type=”Tabby”
Property Name=”Fur” Value=”Coarse”
Property Name=”Color” Value=”Orange”
Part Name=”Paws”
Property Name=”Claws” Value=”Very sharp”
Part Name=”Nose”
Property Name=”Cute” Value=”true”
Info “I have an invalid character.‼”
Cat Id=”2″ Type=”Short hair”
Property Name=”Fur” Value=”Soft”
Property Name=”Color” Value=”Black”
Part Name=”Paws”
Property Name=”Polydactyl” Value=”true”
Property Name=”Claws” Value=”Sharp”
Part Name=”Nose”
Property Name=”Cute” Value=”true”
Info “I don’t have an invalid character.”

We are not out of the woods yet

We are dealing with damaged goods here: that invalid character is still present, so we have to be careful. Notice the in the output above, that is 0x13.

An example of what can go wrong is evident if we try to print out the contents of our XDocument object:

Console.WriteLine(XDocument.Load(filename).ToString());

Normally we will get a printout of the containing XML. In this case we will see the exception we saw above.

Paul

 

References

[1] Extensible Markup Language (XML) 1.0 (Fifth Edition). 26 Nov 2008. W3C Recommendation

[2] XmlDocument Class. MSDN Library

[3] XDocument Class. MSDN Library

[4] XmlReader Class. MSDN Library

[5] XmlReaderSettings Class. MSDN Library

[6] XmlReaderSettings.CheckCharacters Property. MSDN Library

Advertisements

4 responses to “Parsing Xml with Invalid Characters in C#

  1. Pingback: XAML Formatting: Programmatically Tabify XAML and XML Files in C# | Paul Selles

  2. Nico April 9, 2015 at 7:20 am

    Does not work for my case.

    You & Me

    I thought that your solution allows it to parse faulty XML but I can’t load my faulty XML. Any hints?

    • Nico April 9, 2015 at 7:21 am

      Oh XML tags are not displayed in the comment.
      With other words: I have a ‘&’ in a textfield and I want to parse/repair it.

  3. Braks July 30, 2015 at 4:40 am

    Doesn’t work for my case either, I have an EXT character (0x03) in my XML files. It’s not abnormal for copy-pasted content to contain this (this data is exported from a CMS), yet this makes it invalid XML, and it seems CheckCharacters = false isn’t going to get around that. Will have to preprocess the file as plain text I guess.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: