markdown: 3.2.1 breaks when defusedxml is loaded

I was upgrading one of my projects from markdown==3.1 to markdown==3.2.1, and had a very weird behaviour that was quite hard to track down.

In my test case, I had this in my imports:

from markdown import Markdown
from nbconvert import HTMLExporter

def render_markdown(raw):
    return markdown_renderer.reset().convert(raw)

def render_ipynb(raw):
    exporter = HTMLExporter()
    exporter.template_file="basic"
    return smartypants(exporter.from_notebook_node(reads(renumber_notebook(raw), 4))[0]

When I called only render_markdown, I got this pretty obscure exception:

Traceback (most recent call last):
  File "markdown_test.py", line 81, in <module>
    print(render_markdown(test_block))
  File "markdown_test.py", line 79, in render_markdown
    return markdown_renderer.reset().convert(raw)
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/core.py", line 269, in convert
    newRoot = treeprocessor.run(root)
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/extensions/footnotes.py", line 388, in run
    footnotesDiv = self.footnotes.makeFootnotesDiv(root)
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/extensions/footnotes.py", line 181, in makeFootnotesDiv
    self.parser.parseChunk(surrogate_parent, self.footnotes[id])
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/blockparser.py", line 105, in parseChunk
    self.parseBlocks(parent, text.split('\n\n'))
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/blockparser.py", line 123, in parseBlocks
    if processor.run(parent, blocks) is not False:
  File "/Users/rcarmo/.pyenv/versions/3.6.9/lib/python3.6/site-packages/markdown/blockprocessors.py", line 591, in run
    p = etree.SubElement(parent, 'p')
TypeError: SubElement() argument 1 must be xml.etree.ElementTree.Element, not Element
make: *** [smoke-test] Error 1

Reverting back to markdown==3.1 “fixed” it, and it took me a long time to figure out what caused it in the first place.

As it turns out, it went away when I removed the nbconvert import or moved it inside render_ipynb (which is bad form, but turned out to be the quickest workaround). I assume that nbconvert does some kind of conflicting initialization to markdown that does not go away with reset()…

But since it wasn’t happening in markdown==3.1 and happens in both still versions of nbconvert I’ve tried (nbconvert==5.5.0 and nbconvert==5.6.1), I’m assuming something changed in markdown to cause this issue.