Question

I’m trying to filter an XML file using XPath. The XPath that I’m using is definitely filtering to the data that I want, but I’m just not sure how to filter the file overall.

Here’s the sample XML file:

<fields>
    <field name='F'>
        <field name='0'><value>F.0 stuff</value></field>
        <field name='1'><value>F.1 stuff</value></field>
        <field name='2'><value>F.2 stuff</value></field>
    </field>
    <field name='B'>
        <field name='0'><value>B.0 stuff</value></field>
        <field name='1'><value>B.1 stuff</value></field>
        <field name='2'><value>B.2 stuff</value></field>
        <field name='3'><value>B.3 stuff</value></field>
    </field>
</fields>

Here’s the desired output:

<fields>
    <field name='F'>
        <field name='1'><value>F.1 stuff</value></field>
        <field name='2'><value>F.2 stuff</value></field>
    </field>
    <field name='B'>
        <field name='3'><value>B.3 stuff</value></field>
    </field>
</fields>

The solution does not necessarily have to be solved by XPath, but since this is a .NET application, .NET APIs would be appreciated! The following code can be cut and pasted into LINQPad with no edits to see what I’m trying to do.

var doc = XDocument.Parse(@"
<fields>
    <field name='F'>
        <field name='0'><value>F.0 stuff</value></field>
        <field name='1'><value>F.1 stuff</value></field>
        <field name='2'><value>F.2 stuff</value></field>
    </field>
    <field name='B'>
        <field name='0'><value>B.0 stuff</value></field>
        <field name='1'><value>B.1 stuff</value></field>
        <field name='2'><value>B.2 stuff</value></field>
        <field name='3'><value>B.3 stuff</value></field>
    </field>
</fields>");
doc.Dump("Original XML");

var xpath = "//fields/field[@name='F']/field[@name='1' or @name='2'] | //fields/field[@name='B']/field[@name='3']";
doc.XPathSelectElements(xpath).Dump("XPath Combined");

var desired = XDocument.Parse(@"
<fields>
    <field name='F'>
        <field name='1'><value>F.1 stuff</value></field>
        <field name='2'><value>F.2 stuff</value></field>
    </field>
    <field name='B'>
        <field name='3'><value>B.3 stuff</value></field>
    </field>
</fields>");
desired.Dump("Desired Filtered XML");

EDIT: I completely missed XML Transforms – thank you for the solution! Here’s the solution you can paste into LINQPad to see it working:

var filterString = @"@name=""F""]/field[@name=""0""]  | field[@name=""B""]/field[not(@name=""3"")";
var xslFmt = @"
<xsl:stylesheet version='1.0'
 xmlns:xsl='http://www.w3.org/1999/XSL/Transform'>
 <xsl:output omit-xml-declaration='yes' indent='yes'/>

 <xsl:template match='node()|@*'>
  <xsl:copy>
   <xsl:apply-templates select='node()|@*'/>
  </xsl:copy>
 </xsl:template>

 <xsl:template match=
  'field[{0}]
  '/>
</xsl:stylesheet>";
var xslMarkup = string.Format(xslFmt, filterString);

var xmlTree = XDocument.Parse(@"
<fields>
    <field name='F'>
        <field name='0'><value>F.0 stuff</value></field>
        <field name='1'><value>F.1 stuff</value></field>
        <field name='2'><value>F.2 stuff</value></field>
    </field>
    <field name='B'>
        <field name='0'><value>B.0 stuff</value></field>
        <field name='1'><value>B.1 stuff</value></field>
        <field name='2'><value>B.2 stuff</value></field>
        <field name='3'><value>B.3 stuff</value></field>
    </field>
</fields>");
xmlTree.Dump("Original XML");

// Code from MSDN: http://msdn.microsoft.com/en-us/library/bb675186.aspx
var newTree = new XDocument();
using (var writer = newTree.CreateWriter()) {
    // Load the style sheet.
    var xslt = new XslCompiledTransform();
    xslt.Load(XmlReader.Create(new StringReader(xslMarkup)));

    // Execute the transform and output the results to a writer.
    xslt.Transform(xmlTree.CreateReader(), writer);
}

newTree.Dump("Transformed XML");

Answer 1

XPath is a query language and it cannot be used to produce a modified XML document.

The technology which was especially designed for such transformations is called XSLT.

You can use the XDocument.CreateNavigator() method and then use one of the overloads of the XslCompiledTransform.Transform() method to carry out the transformation.

The XSLT transformation itself is very simple:

<xsl:stylesheet version="1.0"
 xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
 <xsl:output omit-xml-declaration="yes" indent="yes"/>

 <xsl:template match="node()|@*">
  <xsl:copy>
   <xsl:apply-templates select="node()|@*"/>
  </xsl:copy>
 </xsl:template>

 <xsl:template match=
  "field[@name='F']/field[@name='0']
  |
   field[@name='B']/field[not(@name='3')]
  "/>
</xsl:stylesheet>

when applied on the provided XML document, it produces the wanted, correct result:

<fields>
    <field name="F">
        <field name="1">
            <value>F.1 stuff</value>
        </field>
        <field name="2">
            <value>F.2 stuff</value>
        </field>
    </field>
    <field name="B">
        <field name="3">
            <value>B.3 stuff</value>
        </field>
    </field>
</fields>