Delete / Remove node from XPathNodeIterator, given an XPath

First of all, if someone has a different, perhaps shorter (or better), solution to the problem, it's welcome as well.

I'm trying to "simply" remove (almost) duplicate elements in XSLT. There's some (metadata) nodes i don't want to include when comparing, and i couldn't figure out how do do that in XSLT so thought i'd extend it with a function removing these nodes. Like so:

<xsl:for-each select="abx:removeNodes(d/df600|d/df610|d/df611|d/df630|d/df650|d/df651|d/df655, '*[@key=&quot;i1&quot; or @key=&quot;i2&quot; or key=&quot;db&quot;]')">
   <xsl:if test="not(node()=preceding-sibling::*)">
      blah
   </xsl:if>
</xsl:for-each>

And the extension, which doesn't work so well... (C#)

public XPathNodeIterator removeNodes(XPathNodeIterator p_NodeIterator, String removeXPath)
{
   Logger Logger = new Logger("xslt");
   Logger.Log("removeNodes(removeXPath={0}):", removeXPath);

   foreach (XPathNavigator CurrentNode in p_NodeIterator)
   {
      Logger.Log("removeNodes(): CurrentNode.OuterXml={0}.", CurrentNode.OuterXml);

      foreach (XPathNavigator CurrentSubNode in CurrentNode.Select(removeXPath))
      {
         Logger.Log("removeNodes(): CurrentSubNode.OuterXml={0}.", CurrentSubNode.OuterXml);
         // How do i delete this node!?
         //CurrentSubNode.DeleteSelf();
      }
   }

   return p_NodeIterator;
}

My initial approach using 'CurrentSubNode.DeleteSelf();' doesn't work because it gets confused and loses its position in the XPathNavigator, causing it to only delete the first item it finds using "removeXPath". Something like a DeleteAndMoveNext() would be nice but there seems to be no such method...

Example data:

<df650>
  <df6开发者_运维知识库50 key="i1"> </df650>
  <df650 key="i2">0</df650>
  <df650 key="a">foo</df650>
  <df650 key="x">bar</df650>
  <df650 key="db">someDB</df650>
  <df650 key="id">b2</df650>
  <df650 key="dsname">someDS</df650>
</df650>

..and then another identical node (if you ignore the meta fields; db,id,dsname).

<df650>
  <df650 key="i1"> </df650>
  <df650 key="i2">0</df650>
  <df650 key="a">foo</df650>
  <df650 key="x">bar</df650>
  <df650 key="db">someOtherDB</df650>
  <df650 key="id">b2</df650>
  <df650 key="dsname">someOtherDS</df650>
</df650>

The result should be...

<df650>
  <df650 key="i1"> </df650>
  <df650 key="i2">0</df650>
  <df650 key="a">foo</df650>
  <df650 key="x">bar</df650>
</df650>

You can do that in XSLT alone easily, an extension function really not necessary. Consider this:

<!-- make a template that matches all nodes that cold be removed -->
<xsl:template match="d/df600|d/df610|d/df611|d/df630|d/df650|d/df651|d/df655">
  <!-- check the your condition for node removal, whatever it may be -->
  <xsl:if test="not(@key='i1' or @key='i2' or @key='db')">
    <!-- ...if it is *not* met, copy the node -->
    <xsl:copy-of select="." />
  </xsl:if>
  <!-- ...in all other cases, nothing happens, i.e. the node is removed -->
</xsl:template>

Thanks for the hint RymdPung, I was able to remove blank rows in a repeating section using your List suggestion.

I added a reference to the System.Collections.Generic namespace in my code.

Here is the method I created to scan through a set of nodes, identify the ones I want to remove, and then delete these in a separate loop.

         private void deleteEmptyRows(string path)
    {
        XPathNodeIterator nodesToCheck = MainDataSource.CreateNavigator().Select(path, NamespaceManager);
        List<XPathNavigator> nodesToDelete = new List<XPathNavigator>();
        foreach (XPathNavigator currentItem in nodesToCheck)
            if (currentItem.Value.Trim().Length == 0)
                nodesToDelete.Add(currentItem);

        foreach(XPathNavigator deleteMe in nodesToDelete)
            deleteMe.DeleteSelf();
    }

The problem can be solved like this (however, it doesn't solve MY actual problem...).

• Create a List of type XPathNavigator that will contain nodes you want to delete.
• Add the nodes to this list instead of using DeleteSelf().
• When done finding all nodes you want to delete, iterate through your List and delete the nodes. Since these nodes are Navigators, there is no issue with lost position.

I gave up on trying to paste the code in after 10 minutes...