51
votes

I'm trying to create a child XML element for this xml:

<?xml version="1.0" encoding="utf-8"?>
<configuration>
</configuration>

I use this PowerShell script:

[xml] $doc = Get-Content($filePath)
$child = $doc.CreateElement("newElement")
$doc.configuration.AppendChild($child)

I have an error: Method invocation failed because [System.String] doesn't contain a method named 'AppendChild'.

1
This has got to be the single dumbest "quirk" in powershell.StingyJack
This works fine in powershell 6. (Tested on 6.2)Tamir Daniely

1 Answers

83
votes

If you use dot notation to navigate an XML file (e.g. $doc.configuration), Powershell tries to be clever about what it returns.

  • If the target element is empty or only contains a single text node, PS will return a String.
  • If the target element contains child nodes other than text nodes, it will return an XmlElement.
  • If multiple target elements exist, it will return an Object[], where each individual array element is again subject to these rules, e.g. it will either be a String or an XmlElement depending on its contents.
  • If the target element does not exist, PS returns $null.

In your case it's easy since you want to append nodes to the document element:

$doc = New-Object System.Xml.XmlDocument
$doc.Load($filePath)
$child = $doc.CreateElement("newElement")
$doc.DocumentElement.AppendChild($child)

but you could use $doc.SelectNodes() or $doc.SelectSingleNode() to navigate around the XML document and always have a node/node list returned.


One could argue about the sensibility of this behavior, but as a matter of fact it makes consuming (sanely structured) XML quite straight-forward - for example tasks such as reading values from a config file, or from an API response. That's the purpose of this simple syntax.

It's not a good tool for creating XML, which is a more complex task. Using DOM API methods from the start is the better approach here.