XPath - how to use position() in XMLInput

[Anonymous]

Hello folks,
i have come across this file that has no identifier for the <b>Item</b>:

<Items>
  <Item>
    <Detail1>A</Detail1>
    <Detail3>B</Detail2>
    <Detail4>B</Detail4>
    ...
  </Item>
  <Item>
    <Detail2>B</Detail2>
    <Detail4>B</Detail4>
    ...
  </Item>
  ...
</Items>

i would like to get the output of this file normalised to:

Item   |Detail1   |Detail2     |Detail3     |Detail4
1      |A         |            |B           |B

where the value of item would be returned by the XPath expression position()
But i am failing to understand how the expression should be entered in the tFileInputXML
this is a similar question to:
https://community.talend.com/t5/Design-and-Development/XML-record-count/td-p/74328

and i would very much appreciate assistance with this.
regards

[Anonymous]

This shows an XPath that returns the sequence of the parent node.
http://blogs.msdn.com/b/asanto/archive/2004/09/08/226663.aspx
count(parent::*/preceding-sibling::*) + 1

However when run in 'Refresh Preview' or in a job the column is empty.
When the job is configured with 'Get Nodes' enabled on the tFileInputXML then the results show that the sequence is being calculated.
e.g. from a different dataset:
.-------------------+----------------------------------+--------.
| tLogRow_1 |
|=------------------+----------------------------------+-------=|
|key |value |entityid|
|=------------------+----------------------------------+-------=|
|key="jurisdiction" |value="GB" |1.0 |
|key="domicile" |value="UK" |1.0 |
How can the column I call 'entityid', computed from the earlier 'count...' XPath be accessed?
The first two columns of output show the full node value of their attribute, so could potentially be parsed.