Xpath selecting ancestors

Question

I am trying to find a formula that creates a URL for an element based on its position in the XML hierarchy.

This is my sample xml:

<Xml>
    <Site Url="http://mysite.abc">
        <Content></Content>
        <SubSites>
            <Site Url="/sub1">
                <Content></Content>
                <SubSites>
                    <Site Url="/sub2">
                        <Content></Content>
                        <SubSites>
                            <Site Url="/sub3">
                                <Content></Content>
                            </Site>
                        </SubSites>
                    </Site>
                </SubSites>
            </Site>
        </SubSites>
    </Site>
</Xml>

I have a function in Powershell that recursively iterates down from the top and on each 'Content' element I want to generate a concatenation of the ancestors Url values. So it should generate consecutively for each 'Content' Node:

http://mysite.abc
http://mysite.abc/sub1
http://mysite.abc/sub1/sub2
http://mysite.abc/sub1/sub2/sub3

I use at the moment as a start: ( $Node = the 'Content' element )

$Sites = $Node | Select-XML -XPath  "//ancestor::Site"

But for every $Node it selects all the 'Site' elements. It would expect it to find more ancestors while going down in the xml structure.

If someone would know how to concatenate the values directly with Xpath that would be especially great, but for starters, I would be happy to know what is going wrong with my current approach.

Answer 1

To offer an alternative to Mathias R. Jessen's helpful answer (which explains the problem with your approach well and offers an effective solution):

Since the Site nodes seem to always be the parent node of any given Content node, you can simply refer to the respective Site node with an .. path component.

This approach allows you to process the entire document at once:

Select-Xml -LiteralPath sample.xml -XPath  "//Content/.." | ForEach-Object -Begin {
    $ancestralUrl = ''
  } -Process {
    $thisUrl = $_.Node.Url
    if ($thisUrl -match '^https?://') {
      $ancestralUrl = $thisUrl
    } else {
      $thisUrl = $ancestralUrl += $thisUrl
    }
    $thisUrl
  }

The above yields:

http://mysite.abc
http://mysite.abc/sub1
http://mysite.abc/sub1/sub2
http://mysite.abc/sub1/sub2/sub3

---

In fact, you can even combine the above approach with the ancestor function (though it would be overkill here):

Select-Xml -LiteralPath sample.xml '//Content/ancestor::Site' | ForEach-Object -Begin {
  $ancestralUrl = ''
} -Process {
  $thisUrl = $_.Node.Url
  if ($thisUrl -match '^https?://') {
    $ancestralUrl = $thisUrl
  } else {
    $thisUrl = $ancestralUrl += $thisUrl
  }
  $thisUrl
}

Answer 2

//ancestor::Site will give you the ancestral Site node relative to any node (//) in the tree.

Use ./ancestor::Site to grab only the ancestor relative to the current node (.):

$Sites = $Node | Select-XML -XPath  "./ancestor::Site"