XSLT - select nodes that came after another node

Question

I'm trying to select all nodes that 1) come after a node with a particular property and 2) have a particular property themselves. So if I had the following XML:

<node id="1"><child attr="valueOfInterest"/></node>
<node id="2"><child attr="boringValue"/></node>
...
<node id="3"><child attr="valueOfInterest"/></node>
<node id="4"><child attr="boringValue"/></node>
<node id="5"><child attr="boringValue"/></node>
<node id="6"><child attr="boringValue"/></node>
...

My XSLT traverses through each node tag. At each node, I want it to select all previous nodes that occurred since the most recent node that had a child whose attr was valueOfInterest. So if I were at node #2, I would want an empty node set. If I were at node #6, I would want to select node #'s 4 and 5. I currently have the following XSLT:

<xsl:variable name="prev_vals"
    select="preceding-sibling::node/child[@attr = $someValueICareAbout]/@attr"/>

So this XSLT gets all preceding attr values that are a particular value. How do I only get those preceding attr values that are in nodes that come after the most recent node whose child has a particular attr value (i.e., "valueOfInterest")? The id attribute on node tags is not guaranteed to be increasing, so we can't compare against that.

Edit: I thought these might be of use:

<xsl:variable name="prev_children_of_interest"
    select="preceding-sibling::node/child[@attr != $someValueICareAbout]"/>
<xsl:variable name="mru_child_of_interest"
    select="$prev_children_of_interest[count($prev_children_of_interest)]"/>

So that's all previous child tags with attr=valueOfInterest and then the most recently used (closest to current node) child tag that has the attribute I'm looking for. From mru_child_of_interest we can find the most recently used parent tag, but how do we then look for nodes that come after that tag?

This can be done using grouping. Can you use XSLT 2.0 (grouping features), or are you limited to 1.0? — LarsH, Jun 13 '11 at 21:34
`"$prev_children_of_interest[count($prev_children_of_interest)]"`: it looks like you are expecting the inner `$prev_children_of_interest` to be reevaluated in the context of each element in the outer `$prev_children_of_interest` node set. But in fact, the value of that variable is only evaluated at the point where the variable is defined. So this XPath expression is equivalent to `"$prev_children_of_interest"`. — LarsH, Jun 13 '11 at 21:39
Good question, +1. See my answer for a complete, short and easy solution that uses the Kayessian formula for the intersection of two node-sets. :) — Dimitre Novatchev, Jun 14 '11 at 03:31
@Sarah Vessels: Was my answer useful or did you have any problems in using my solution? — Dimitre Novatchev, Jun 15 '11 at 02:50
@Dimitre Novatchev: your answer is useful and I appreciate the long reply. :) — Sarah Vessels, Jun 16 '11 at 14:56

Grzegorz Szpetkowski · Accepted Answer · 2011-06-13T22:34:29.447

I am not sure if I understand your question correctly, but here is some XSL 1.0 (additional each-nodes attributes are informational only):

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
    <xsl:output method="xml" indent="yes"/>

    <xsl:template match="nodes">
        <xsl:copy>
            <xsl:apply-templates select="node"/>
        </xsl:copy>
    </xsl:template>

    <xsl:template match="node">
        <xsl:variable name="someValueICareAbout">valueOfInterest</xsl:variable>

        <xsl:variable name="recentParticularNode"
            select="preceding-sibling::node[child/@attr = $someValueICareAbout][1]"/>

        <xsl:variable name="recentParticularNodePosition"
            select="count($recentParticularNode/preceding-sibling::node) + 1"/>

        <xsl:variable name="currentPosition" select="position()"/>

        <xsl:if test="child/@attr != $someValueICareAbout">
            <each-nodes id="{@id}" cp="{$currentPosition}" 
                    rpnp="{$recentParticularNodePosition}">
                <xsl:copy-of
                    select="../node[position() &gt; $recentParticularNodePosition
                    and position() &lt; $currentPosition]"/>
            </each-nodes>
        </xsl:if>
    </xsl:template>
</xsl:stylesheet>

Input XML:

<?xml version="1.0" encoding="UTF-8"?>
<nodes>
    <node id="1"><child attr="valueOfInterest"/></node>
    <node id="2"><child attr="boringValue2"/></node>
    <node id="3"><child attr="valueOfInterest"/></node>
    <node id="4"><child attr="boringValue4"/></node>
    <node id="5"><child attr="boringValue5"/></node>
    <node id="6"><child attr="boringValue6"/></node>
</nodes>

Result XML:

<?xml version="1.0" encoding="UTF-8"?>
<nodes>
   <each-nodes id="2" cp="2" rpnp="1"/>
   <each-nodes id="4" cp="4" rpnp="3"/>
   <each-nodes id="5" cp="5" rpnp="3">
      <node id="4">
         <child attr="boringValue4"/>
      </node>
   </each-nodes>
   <each-nodes id="6" cp="6" rpnp="3">
      <node id="4">
         <child attr="boringValue4"/>
      </node>
      <node id="5">
         <child attr="boringValue5"/>
      </node>
   </each-nodes>
</nodes>

score 3 · Answer 2 · edited May 23 '17 at 11:53

It looks like you want the intersection of two sets. Set 1 is all the nodes after the last valueOfInterest. Set 2 is all the nodes before the current node that do not contain valueOfInterest. For XPath 1.0, the following will give you the intersection (reference found here).

$set1[count($set2|.)=count($set2)]

Given your input, the following XSL demonstrates the nodeset you are looking for

<?xml version="1.0" encoding="utf-8"?>
<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
  <xsl:output method="xml" indent="yes"/>
  <xsl:strip-space elements="*"/>
  <xsl:variable name="someValueICareAbout">valueOfInterest</xsl:variable>
  <xsl:template match="@* | node()">
    <xsl:copy>
      <xsl:apply-templates select="@* | node()"/>
    </xsl:copy>
  </xsl:template>

  <xsl:template match="node[child/@attr='boringValue']">
    <xsl:copy>
      <xsl:apply-templates select="@* | node()"/>

      <xsl:variable name="set1" select="preceding-sibling::node[child/@attr='valueOfInterest'][1]/following-sibling::node "/>

      <xsl:variable name="set2" select="preceding-sibling::node[child/@attr='boringValue']"/>

      <predecessors>
          <xsl:copy-of select="$set1[count($set2|.)=count($set2)]"/>
      </predecessors>

    </xsl:copy>
  </xsl:template>
</xsl:stylesheet>

Here is the output

<xml>
  <node id="1">
    <child attr="valueOfInterest"/>
  </node>
  <node id="2">
    <child attr="boringValue"/>
    <predecessors/>
  </node>
  <node id="3">
    <child attr="valueOfInterest"/>
  </node>
  <node id="4">
    <child attr="boringValue"/>
    <predecessors/>
  </node>
  <node id="5">
    <child attr="boringValue"/>
    <predecessors>
      <node id="4">
        <child attr="boringValue"/>
      </node>
    </predecessors>
  </node>
  <node id="6">
    <child attr="boringValue"/>
    <predecessors>
      <node id="4">
        <child attr="boringValue"/>
      </node>
      <node id="5">
        <child attr="boringValue"/>
      </node>
    </predecessors>
  </node>
</xml>

Note, the reason I use [1] in preceding-sibling::node[child/@attr='valueOfInterest'][1] is because the order of the nodeset is reversed by preceding-sibling see here.

If you have XPath 2.0, you can use the intersect operator

  <predecessors>
    <xsl:copy-of select="$set1 intersect $set2"/>
  </predecessors>

This produces the same result.

score 3 · Answer 3 · answered Jun 14 '11 at 03:29

This transformation copies exactly the wanted nodes:

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

 <xsl:template match=
 "node[not(child/attr='valueOfInterest')]">

 <xsl:variable name="vFollowing" select=
 "preceding-sibling::node
          [child/@attr='valueOfInterest'][1]
            /following-sibling::node"/>

 <xsl:variable name="vPreceding" select=
  "preceding-sibling::node"/>

  <xsl:copy-of select=
  "$vFollowing[count(. | $vPreceding)
              =
               count($vPreceding)
              ]
  "/>
======================
 </xsl:template>
 <xsl:template match="text()"/>
</xsl:stylesheet>

when applied on this XML document (based on the provided XML fragment and wrapping it in a top element to make it a well-formed XML document):

<t>
    <node id="1">
        <child attr="valueOfInterest"/>
    </node>
    <node id="2">
        <child attr="boringValue"/>
    </node>...
    <node id="3">
        <child attr="valueOfInterest"/>
    </node>
    <node id="4">
        <child attr="boringValue"/>
    </node>
    <node id="5">
        <child attr="boringValue"/>
    </node>
    <node id="6">
        <child attr="boringValue"/>
    </node>...
</t>

the wanted, correct result is produced:

======================

======================
 <node id="2">
   <child attr="boringValue"/>
</node>
======================

======================
 <node id="4">
   <child attr="boringValue"/>
</node>
======================
 <node id="4">
   <child attr="boringValue"/>
</node>
<node id="5">
   <child attr="boringValue"/>
</node>
======================

Explanation:

Here we use the well-known Kayessian formula (discovered by the SO user @Michael Kay) for the intersection of two node-sets $ns1 and $ns2:

ns1[count(.|$ns2) = count($ns2)]
We simply substitute $vFollowing and $vPreceding for $ns1 and $ns2 in the above formula.

$vFollowing is defined to contain exactly all the following sibling elements namednodeof the nearestnode` that satisfies the condition (to be interesting).

$vPreceding is the set of all node elements that are preceding siblings of the current (matched) node.

.3. Their intersection is exactly the wanted node-set.

LarsH · Answer 4 · 2011-06-13T22:22:39.420

2

Here's one way to do it in XSLT 2.0:

<xsl:variable name="prevVOI"
      select="(preceding-sibling::node[child/@attr = 'valueOfInterest'])[last()]" />
<xsl:variable name="prevNodesAfterVOI"
      select="preceding-sibling::node[. >> $prevVOI]" />

edited Jun 13 '11 at 22:22

answered Jun 13 '11 at 21:49

LarsH

27,481
8
94
152

2

Do note you need to select the first matched preceding sibling to compare against a single node, and make the second instruction works correctly. – Emiliano Poggi Jun 13 '11 at 22:07
@empo: thanks, you're entirely right. I've edited my answer accordingly. – LarsH Jun 13 '11 at 22:22
1

May be more intuitive `preceding-sibling::node[child/@attr = 'valueOfInterest'][1]` :)) – Emiliano Poggi Jun 13 '11 at 22:32
@empo: in a way, I agree; but for me it's always hard to remember when the `[1]` is bound closely enough to the reverse axis so that `[1]` actually means last in document order; and when it isn't. But if I use `(...)[last()]` then I know for sure that I'm selecting the last in document order. – LarsH Jun 14 '11 at 02:10

XSLT - select nodes that came after another node

4 Answers4