[xml.dom.minidom] nodeValue attr is not defined on Node class but directly on all derived classes
According to the documentation, the Node class is supposed to have an attribute nodeValue that "has a different meaning for each node type; see the DOM specification for details.. [...] The value is a string or None."". However, this attribute is undefined on the Node class, but is then implemented on all derived classes (DocumentFragment, Attr, Element, ProcessingInstruction, CharacterData, DocumentType, Entity, Notation, Document).
As a result, IDEs and static analysers raise false positives when analysing "correct" code.
from xml.dom import minidom
dummy_xml: str = """<?xml version="1.0"?>
<Envelope>
<Body>
<StockName>T</StockName>
<StockName>N</StockName>
</Body>
</Envelope>
"""
dom: minidom.Document = minidom.parseString(dummy_xml)
body: minidom.Element = dom.getElementsByTagName("Body")[0]
child_node: minidom.Node
for child_node in body.childNodes:
inner_node: minidom.Node | None = child_node.firstChild
if inner_node is not None:
print(inner_node.nodeValue)The example above prints T and N as expected, but Pylance reports the following issue:
Cannot access member "nodeValue" for type "Node"
Member "nodeValue" is unknown - Pylance (reportGeneralTypeIssues)
Indeed the nodes we process happen to be Elements - therefore the attribute is defined at runtime. However during static analysis this is unknown despite this being legal according to the documentation.
Proposed solution
As far as i can tell this would be very easy to fix: just by initializing this attribute to None in the Node class (as is done for other similar attributes), this problem could be fully resolved:
|
class Node(xml.dom.Node): |
|
namespaceURI = None # this is non-null only for elements and attributes |
|
parentNode = None |
|
ownerDocument = None |
|
nextSibling = None |
|
previousSibling = None |
|
|
While we're at it it'd be worth double checking if other attributes are affected by the same issue. I can submit a PR in a few hours if you'd like.
My environment
- CPython versions tested on: 3.8.2, 3.10.8, 3.11.0
- Operating system and architecture: RHEL7.9 x86_64, Windows 10 x86_64
Linked PRs
[xml.dom.minidom] nodeValue attr is not defined on Node class but directly on all derived classes
According to the documentation, the Node class is supposed to have an attribute
nodeValuethat "has a different meaning for each node type; see the DOM specification for details.. [...] The value is a string or None."". However, this attribute is undefined on the Node class, but is then implemented on all derived classes (DocumentFragment, Attr, Element, ProcessingInstruction, CharacterData, DocumentType, Entity, Notation, Document).As a result, IDEs and static analysers raise false positives when analysing "correct" code.
The example above prints T and N as expected, but Pylance reports the following issue:
Indeed the nodes we process happen to be Elements - therefore the attribute is defined at runtime. However during static analysis this is unknown despite this being legal according to the documentation.
Proposed solution
As far as i can tell this would be very easy to fix: just by initializing this attribute to
Nonein the Node class (as is done for other similar attributes), this problem could be fully resolved:cpython/Lib/xml/dom/minidom.py
Lines 34 to 40 in b99ac1d
While we're at it it'd be worth double checking if other attributes are affected by the same issue. I can submit a PR in a few hours if you'd like.
My environment
Linked PRs