Xpath - Group matches as combined results - xpath

Question

I'm having the following problem, I want to create a Xpath expression that results into 3 matches, where match1=text1text1text, match2=text2text2text2 etc.
I'm able to find each p seperate, but i just can't 'group' them.
What the expression should say is: "after a[#class="random-a-tag"] give me all the text of all the p-tags that exist before each element where div[#id]"
<a class="random-a-tag"></a>
<p>text1</p>
<p>text1</p>
<p>text1</p>
<div id="1"></div>
<p>text2</p>
<p>text2</p>
<p>text2</p>
<div id="2"></div>
<p>text3</p>
<p>text3</p>
<p>text3</p>
<div id="3"></div>

Answer 1

You can do this in two steps.
find all divs
//a[#class="random-a-tag"]/following-sibling::div[#id]
and for each div #id do:
//a[#class="random-a-tag"]/following-sibling::p[following-sibling::div[1][#id="3"]]/text()

Answer 2

This cannot be done with a single XPath 1.0 expression.
The most you can select is the set of p for a given value of #id:
/*/a/following-sibling::div
[#id=$pId]
/preceding-sibling::p
[count(.
|
/*/a/following-sibling::div
[#id=$pId]
/preceding-sibling::div[1]
/preceding-sibling::p
)
=
count(/*/a/following-sibling::div
[#id=$pId]
/preceding-sibling::div[1]
/preceding-sibling::p
)
+1
]
If $pId is (substituted by) 2, and the above XPath expression is applied on this XML document (your XML fragment, wrapped in a top element to make it well-formed XML document):
<t>
<a class="random-a-tag"></a>
<p>text1</p>
<p>text1</p>
<p>text1</p>
<div id="1"></div>
<p>text2</p>
<p>text2</p>
<p>text2</p>
<div id="2"></div>
<p>text3</p>
<p>text3</p>
<p>text3</p>
<div id="3"></div>
</t>
then this selects the following nodes:
<p>text2</p>
<p>text2</p>
<p>text2</p>
In the above XPath expression we use the wellknown Kayessian (created by #Michael Kay) formula for node-set intersection:
$ns1[count(.|$ns2) = count($ns2)]
is the intersection of the nodesets $ns1 and $ns2.
II. XPath 2.0 solution:
(a/following-sibling::div
[#id=$pId]
/preceding-sibling::p
except
a/following-sibling::div
[#id=$pId]
/preceding-sibling::div[1]
/preceding-sibling::p
)/string()
When this XPath 2.0 expression is evaluated against the same XML document (above) and $pId is 2, the result is exactly the wanted text:
text2 text2 text2