CDATA nodes' content can only be retrieved using XMLParser.get_node_name().

[elenakrittik]

Tested versions

Reproducible on 4.2.1 (and it seems ever since 3.5 at least, #65416 )

System information

Godot v4.2.1.stable - Android - GLES3 (Compatibility) - Mali-G52 - (8 Threads)

Issue description

From #65416:

The CDATA content can be read via get_node_name() (wich is odd .. imho).

get_node_name() currently says

This method will raise an error if the currently parsed node is not of NODE_ELEMENT or NODE_ELEMENT_END type.

But in practice does not for NODE_CDATA nodes and instead returns CDATA content. As it seems like fair amount of people already rely on this, (aside from correcting documentation, of course) i propose to soft-deprecate the current behaviour and instead make get_node_data() responsible for returning CDATA nodes' content.

Steps to reproduce

Download the test project and run it. It will print the following (excluding the arrow note):

Node name: data
   At: res://main.gd:12:_ready()
Is node a CDATA node: false
   At: res://main.gd:13:_ready()
Node name: 
   At: res://main.gd:12:_ready()
Is node a CDATA node: false
   At: res://main.gd:13:_ready()
Node name:  Hello World          <-- here
   At: res://main.gd:12:_ready()
Is node a CDATA node: true
   At: res://main.gd:13:_ready()
Node name: data
   At: res://main.gd:12:_ready()
Is node a CDATA node: false
   At: res://main.gd:13:_ready()

Minimal reproduction project (MRP)

xmlcdata.zip

[timothyqiu]

Thanks for the report!

The documentation for XMLParser is written way after the implementation. So I believe it's get_node_name()'s description that's wrong. This method errors when the node is a text node, exactly opposite of get_node_data().

The name (as well as the API) is indeed confusing. But on the other side, changing the behavior breaks compatibility with existing projects. Personally, I don't think bad naming is a good reason for compat breaks :P