How to concatenate the contents of all children of a node in XPath?

Question

Is it possible with XPath to get a concatenated view of all of the children of a node? I am looking for something like the JQuery .html() method.

For example, if I have the following XML:

<h3 class="title">
    <span class="content">this</span>
    <span class="content"> is</span>
    <span class="content"> some</span>
    <span class="content"> text</span>
</h3>

I would like an XPath query on "h3[@class='title']" that would give me "this is some text".

That is the real question, but if more context/background is helpful, here it is: I am using XPath and I used this post to help me write some complex XSL. My source XML looks like this.

<h3 class="title">Title</h3>
<p>
    <span class="content">Some</span>
    <span class="content"> text</span>
    <span class="content"> for</span>
    <span class="content"> this</span>
    <span class="content"> section</span>
</p>
<p>
    <span class="content">Another</span>
    <span class="content"> paragraph</span>
</p>
<h3 class="title">
    <span class="content">Title</span>
    <span class="content"> 2</span>
    <span class="content"> is</span>
    <span class="content"> complex</span>
</h3>
<p>
    <span class="content">Here</span>
    <span class="content"> is</span>
    <span class="content"> some</span>
    <span class="content"> text</span>
</p>

My output XML considers each <h3> as well as all <p> tags until the next <h3>. I wrote the XSL as follows:

<xsl:template match="h3[@class='title']">
...
    <xsl:apply-templates select="following-sibling::p[
        generate-id(preceding-sibling::h3[1][@class='title'][text()=current()/text()])
        =
        generate-id(current())
    ]"/>
...
</xsl:template>

The problem is that I use the text() method to identify h3s that are the same. In the example above, the "Title 2 is complex" title's text() method returns whitespace. My thought was to use a method like JQuery's .html that would return me "Title 2 is complex".

Update: This might help clarify. After the transform, the desired output for the above would look something like this:

<section>
    <title>Title</title>
    <p>
        <content>Some</content>
        <content> text</content>
        <content> for</content>
        <content> this</content>
        <content> section</content>
    </p>
    <p>
        <content>Another</content>
        <content> paragraph</content>
    </p>
</section>
<section>
    <title>
        <content>Title</content>
        <content> 2</content>
        <content> is</content>
        <content> complex</content>
    </title>
    <p>
        <content>Here</content>
        <content> is</content>
        <content> some</content>
        <content> text</content>
    </p>
</section>

Answer 1

h3[@class='title']/span[@class='content']/text()

Like this?

h3[@class='title']/descendant::*/text()

Or this?

Answer 2

From http://www.w3.org/TR/xpath/#dt-string-value

The string-value of an element node is the concatenation of the string-values of all text node descendants of the element node in document order.

So, you are looking for the function string()

From http://www.w3.org/TR/xpath/#function-string

A node-set is converted to a string by returning the string-value of the node in the node-set that is first in document order. If the node-set is empty, an empty string is returned

Anyway, keep in mind that the string() is often applied implicitly, for example: value-of, node-set comparisons, etc.

But if you really need is to apply templates to the elements "p" that follow a certain element "h3", I recommend using the expression for node-set subtraction. In this case:

<xsl:apply-templates select="following::p[count(.|following::p) > count(current()/following::h3/following::p)]"/>

Note 1: It is seen in this example as the semantics of XHTML is in some cases rather vague.

Note 2: Why is it necessary to apply templates explicitly to the elements "p"? If this is so because it takes some data that it provides the element "h3" which precedes it, the same can be achieved with the expression "preceding::h3" within a template for the elements "p".

Edit: Now, with your output sample...

XML input:

<root>
<h3 class="title">Title</h3> 
<p> 
    <span class="content">Some</span> 
    <span class="content"> text</span> 
    <span class="content"> for</span> 
    <span class="content"> this</span> 
    <span class="content"> section</span> 
</p> 
<p> 
    <span class="content">Another</span> 
    <span class="content"> paragraph</span> 
</p> 
<h3 class="title"> 
    <span class="content">Title</span> 
    <span class="content"> 2</span> 
    <span class="content"> is</span> 
    <span class="content"> complex</span> 
</h3> 
<p> 
    <span class="content">Here</span> 
    <span class="content"> is</span> 
    <span class="content"> some</span> 
    <span class="content"> text</span> 
</p> 
</root>

XSLT stylesheet:

<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">

<xsl:template match="/">
<root>
<xsl:apply-templates select="/root/h3" />
</root>
</xsl:template>

<xsl:template match="h3">
<section>
<title>
<xsl:apply-templates select="node()" />
</title>
<xsl:apply-templates select="following-sibling::p[count(preceding-sibling::h3[1]|current())=1]"/>
</section>
</xsl:template>

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

<xsl:template match="span">
<content>
<xsl:value-of select="." />
</content>
</xsl:template>

</xsl:stylesheet>

XML output:

<root>
<section>
<title>Title</title>
<p>
<content>Some</content>
<content> text</content>
<content> for</content>
<content> this</content>
<content> section</content>
</p>
<p>
<content>Another</content>
<content> paragraph</content>
</p>
</section>
<section>
<title>
<content>Title</content>
<content> 2</content>
<content> is</content>
<content> complex</content>
</title>
<p>
<content>Here</content>
<content> is</content>
<content> some</content>
<content> text</content>
</p>
</section>
</root>