33

Consider the following xml fragment:

<Obj>
   <Name><![CDATA[SomeText]]></Name>
</Obj>

How do I retrieve the "SomeText" value via XPath? I'm using Nauman Leghari's (excellent) Visual XPath tool.
/Obj/Name returns the element
/Obj/Name/text() returns blank

I don't think its a problem with the tool (I may be wrong) - I also read XPath can't extract CDATA (See last response in this thread) - which sounds kinda weird to me.

Gishu
  • 134,492
  • 47
  • 225
  • 308

6 Answers6

22

/Obj/Name/text() is the XPath to return the content of the CDATA markup.

What threw me off was the behavior of the Value property. For an XMLNode (DOM world), the XmlNode.Value property of an Element (with CDATA or otherwise) returns Null. The InnerText property would give you the CDATA/Text content. If you use Xml.Linq, XElement.Value returns the CDATA content.

string sXml = @"
<object>
    <name><![CDATA[SomeText]]></name>
    <name>OtherName</name>
</object>";

XmlDocument xmlDoc = new XmlDocument();
xmlDoc.LoadXml( sXml );
XmlNamespaceManager nsMgr = new XmlNamespaceManager(xmlDoc.NameTable);

Console.WriteLine(@"XPath = /object/name" );
WriteNodesToConsole(xmlDoc.SelectNodes("/object/name", nsMgr));

Console.WriteLine(@"XPath = /object/name/text()" );
WriteNodesToConsole( xmlDoc.SelectNodes("/object/name/text()", nsMgr) );

Console.WriteLine(@"Xml.Linq = obRoot.Elements(""name"")");
XElement obRoot = XElement.Parse( sXml );
WriteNodesToConsole( obRoot.Elements("name") );

Output:

XPath = /object/name
        NodeType = Element
        Value = <null>
        OuterXml = <name><![CDATA[SomeText]]></name>
        InnerXml = <![CDATA[SomeText]]>
        InnerText = SomeText

        NodeType = Element
        Value = <null>
        OuterXml = <name>OtherName</name>
        InnerXml = OtherName
        InnerText = OtherName

XPath = /object/name/text()
        NodeType = CDATA
        Value = SomeText
        OuterXml = <![CDATA[SomeText]]>
        InnerXml =
        InnerText = SomeText

        NodeType = Text
        Value = OtherName
        OuterXml = OtherName
        InnerXml =
        InnerText = OtherName

Xml.Linq = obRoot.Elements("name")
        Value = SomeText
        Value = OtherName

Turned out the author of Visual XPath had a TODO for the CDATA type of XmlNodes. A little code snippet and I have CDATA support now. alt text

MainForm.cs 

private void Xml2Tree( TreeNode tNode, XmlNode xNode)
{
   ...
   case XmlNodeType.CDATA:
      //MessageBox.Show("TODO: XmlNodeType.CDATA");
      // Gishu                    
      TreeNode cdataNode = new TreeNode("![CDATA[" + xNode.Value + "]]");
      cdataNode.ForeColor = Color.Blue;
      cdataNode.NodeFont = new Font("Tahoma", 12);
      tNode.Nodes.Add(cdataNode);
      //Gishu
      break;
Community
  • 1
  • 1
Gishu
  • 134,492
  • 47
  • 225
  • 308
  • 1
    > the behavior of the Value property You are seeing the effect of different data models reflected by different APIs. Many APIs follow the XML Infoset which distinguishes text and CData content. But XPath does not. .NET's XPathDocument API follows the XPath model so you cannot explicitly see CDATA, – Richard Feb 21 '09 at 18:58
13

CDATA sections are just part of what in XPath is known as a text node or in the XML Infoset as "chunks of character information items".

Obviously, your tool is wrong. Other tools, as the XPath Visualizer correctly highlight the text of the Name element when evaluating this XPath expression:

/*/Name/text()

One can also write a simple XSLT transformation:

<xsl:stylesheet version="1.0"
 xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
 <xsl:output method="text"/>
<xsl:template match="/">
  "<xsl:value-of select="/*/Name"/>"
</xsl:template>
</xsl:stylesheet>

When this transformation is applied on the provided XML document:

<Obj>
    <Name><![CDATA[SomeText]]></Name>
</Obj>

the correct result is produced:

  "SomeText"
Dimitre Novatchev
  • 240,661
  • 26
  • 293
  • 431
10

i think the thread you referenced says that the CDATA markup itself is ignored by XPATH, not the text contained in the CDATA markup.

my guess is that its an issue with the tool, the source code is available for download, maybe you can debug it...

Jason
  • 15,915
  • 3
  • 48
  • 72
  • 1
    It basically had a giant DisplayNode() switch case on the type of XMLNode scanned... with the specific case block with a TODO Inside it :) Issue with the tool it is. – Gishu Apr 24 '09 at 14:38
3

See if this helps - http://www.zrinity.com/xml/xpath/
XPATH = /Obj/Name/text()

shahkalpesh
  • 33,172
  • 3
  • 63
  • 88
0

Just in case you run into a similar issue with jdom2, text() will be an array.

To recover CDATA, use /Obj/Name/text()

S.Walsh
  • 3,105
  • 1
  • 12
  • 23
Jesus Marval
  • 1
-4

A suggestion would be to have another field of the md5 hash of the cdata. You can then use xpath to query based off the md5 with no issue 

<sites>
  <site>
    <name>Google</name>
    <url><![CDATA[http://www.google.com]]></url>
    <urlMD5>ed646a3334ca891fd3467db131372140</urlMD5>
  </site>
</sites>

Then you can search:

/sites/site[urlMD5=ed646a3334ca891fd3467db131372140]
MANCHUCK
  • 2,424
  • 1
  • 16
  • 22