CAFxX wrote:
> I found just a minor error in your code, it should be
> 
> //*[contains(child::text(),'Y')]/ancestor::a[1]
> 
> and not
> 
> //*[contains(.,'Y')]/ancestor::a[1]
> 
> right?

In some circumstances...it depends on what other markup is in your 
real-life document. If you want to find all a elements which have a Y in 
any element within them, regardless of how deep, then my example is 
correct, because <a><b>Y</b></a> contains Y somewhere in it.But the same 
is then true of your outer a as well, because 
<a><b>X</b><a><c><b>Y</b></c></a></a> contains XY, which contains a Y.

If what you really want is the a ancestor of all b elements that 
directly contain Y, then you need to specify that:

//b[contains(text(),'Y')]/ancestor::a[1]

This will return your nested a element rather than both a elements.

///Peter
-- 
XML FAQ: http://xml.silmaril.ie/


> Peter Flynn ha scritto:
>> CAFxX wrote:
>>> Hi everybody, I have a problem with a xpath query.
>>> Suppose you have something like this:
>>>
>>> <a>
>>>   <b>X</b>
>>>   <a>
>>>     <c>
>>>       <b>Y</b>
>>>     </c>
>>>   </a>
>>> </a>
>>>
>>> Right now to obtain a list of <a> containg a certain token I do:
>>>
>>> //a[ contains( descendant-or-self::*, 'token' ) ]
>>>
>>> the problem is that, in the case above, searching for the token Y 
>>> would give both <a>s, whereas I need only the innermost one (i.e. the 
>>> nearest <a> ancestor of the fragment containing the searched token). 
>>> How can I translate this in xpath?
>>
>> //*[contains(.,'Y')]/ancestor::a[1]
>>
>> ///Peter