Unexpected behaviour with ElementTree.ParseError

[pmezard]

Hello,

No sure if it is a bug or feature, but this is certainly a regression in 0.7.0. To reproduce:

import defusedxml.ElementTree
import xml.etree.ElementTree

try:
    defusedxml.ElementTree.fromstring("")
    print("OK")
except xml.etree.ElementTree.ParseError as e:
    print("PARSEERROR", e.__class__)
except Exception as e:
    print("NOT PARSEERROR", e.__class__)

which outputs:

NOT PARSEERROR <class 'xml.etree.ElementTree.ParseError'>

Unsurprisingly, it can be bisected to:

commit 3a48453780793c3e98251b0139e9e89e310fb617 (HEAD, refs/bisect/bad)
Author: Christian Heimes <[email protected]>
Date:   Tue Jan 12 16:37:17 2021 +0100

    Restore xml.etree.ElementTree after patch
    
    Restore ``ElementTree`` attribute of ``xml.etree`` module after patching

Is this a bug or should we do this in a different manner? Regardless, this is very confusing.

Thanks.

[tiran]

This is an unforeseen side-effect of module reloading code. The exception classes look to be the same, but they are actually different classes.

>>> import defusedxml.ElementTree
>>> import xml.etree.ElementTree
>>> defusedxml.ElementTree.ParseError is xml.etree.ElementTree.ParseError
False
>>> defusedxml.ElementTree.ParseError
<class 'xml.etree.ElementTree.ParseError'>
>>> id(defusedxml.ElementTree.ParseError)
94084011057936
>>> xml.etree.ElementTree.ParseError
<class 'xml.etree.ElementTree.ParseError'>
>>> id(xml.etree.ElementTree.ParseError)
94084011051104

Looking into a fix now...

[tiran]

Fixed in release 0.7.1.

Thanks for the report!

[pmezard]

Merci !