`parseXml` unexpectedly creates `xnText` nodes after encountering an xml comment

[doublespiral]

Nim Version

Nim Compiler Version 2.2.4 [Linux: amd64]
Compiled at 2025-04-22
Copyright (c) 2006-2025 by Andreas Rumpf

git hash: f7145dd26efeeeb6eeae6fff649db244d81b212d
active boot switches: -d:release

Description

Example that reproduces the issue:

import std/[xmltree, xmlparser, parsexml]

const xml_text = """
<?xml version="1.0" encoding="UTF-8"?>
<foo>
    <bar>text bar</bar>
    <!-- xml comment -->
    <baz>text baz</baz>
</foo>
"""

when isMainModule:
    let root_node = xml_text.parseXml()

    for child in root_node:
        echo "child.kind() = ", child.kind()

Current Output

child.kind() = xnElement
child.kind() = xnComment
child.kind() = xnText
child.kind() = xnElement

Expected Output

child.kind() = xnElement
child.kind() = xnComment
child.kind() = xnElement

Known Workarounds

Removing the characters between the end of the comment and the start of the new element

Additional Information

The proc causing this issue is rawGetTok in lib/pure/parsexml.nim. It assumes that when we're not creating a new element we're creating a text node, which in the case of the ending of a comment, is not expected

[Araq]

But after the comment whitespace follows which is text, I'm not sure and I don't remember what the XML spec says. Avoid XML if you can.