restrict xpath within nested element - xpath

Question

I want to find all index terms by section, but sections are nested. Here is a simple example.
<chapter>
<section><title>First Top Section</title>
<indexterm text="dog"/>
<para>
<indexterm text="tree"/>
</para>
<section><title>SubSection</title>
<indexterm text="cat"/>
</section>
</section>
<section><title>Second Top Section</title>
<indexterm text="elephant" />
</section>
</chapter>
Is there any xpath expression to get a result like this:
First Top Section = ["dog", "tree"]
Subsection = ["cat"]
Second Top Section = ["elephant"]
Of course I get all the descendant indexterms under a section with an expression like this:
/chapter/section//indexterm
But indexterms can be inside other elements in a section--they're not necessarily children.
Is it possible to get indexterms specific to their parent section using xpath?

Answer 1

You can put a predicate at the section level:
/chapter/section[title = 'First Top Section']//indexterm
but this will include all indexterm elements under the given section, including those in subsections. To exclude them you could do something like
/chapter/section[title = 'First Top Section']//indexterm[count(ancestor::section) = 1]
to pick out those indexterm elements that have exactly one section ancestor (i.e. the "First Top Section" you started with).
More generally, if you have a reference to a specific section element then you can get all the indexterm elements inside it but not inside a subsection by first evaluating
count(ancestor-or-self::section)
as a number, and with the current section element as the context node, and then build up another expression
.//indexterm[count(ancestor::section) = thenumberyoujustcounted]
and evaluate that as a node set, again with the original section element as the context node.

Answer 2

If you can use XPath 2.0, you could do:
XML Input
<chapter>
<section><title>First Top Section</title>
<indexterm text="dog"/>
<para>
<indexterm text="tree"/>
</para>
<section><title>SubSection</title>
<indexterm text="cat"/>
</section>
</section>
<section><title>Second Top Section</title>
<indexterm text="elephant" />
</section>
</chapter>
XPath 2.0
for $section in //section
return concat($section/title,' - ["',
string-join($section//indexterm[ancestor::section[1] is $section]/#text,
'", "'),'"]
')
Output
First Top Section - ["dog", "tree"]
SubSection - ["cat"]
Second Top Section - ["elephant"]