PythonPackageNameCheck: fix TypeError on empty PyPi remote

[falbrechtskirchinger]

Skip check when the PyPI remote-id in metadata.xml is empty and add a test package.

Addresses the following crash observed with a template metadata.xml (<remote-id type="pypi"></remote-id>):

$ pkgcheck scan .
pkgcheck scan: error: Traceback (most recent call last):
  File "/usr/lib/python3.14/site-packages/pkgcheck/pipeline.py", line 191, in _run_checks
    if results := sorted(
                  ~~~~~~^
        chain.from_iterable(pipes[i][-1][scope][j].run(restrict) for j in runners)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ):
    ^
  File "/usr/lib/python3.14/site-packages/pkgcheck/runners.py", line 71, in run
    yield from check.feed(item)
  File "/usr/lib/python3.14/site-packages/pkgcheck/checks/python.py", line 1114, in feed
    if pkg.package != normalize(pypi_name):
                      ~~~~~~~~~^^^^^^^^^^^
  File "/usr/lib/python3.14/site-packages/pkgcheck/checks/python.py", line 1111, in normalize
    return PROJECT_SYMBOL_NORMALIZE_RE.sub("-", project).lower()
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^
TypeError: expected string or bytes-like object, got 'NoneType'

Expected result:

PkgMetadataXmlEmptyElement: metadata.xml: empty element 'remote-id' on line 9

The test is primarily a regression test, for which I could find no precedent. python/dev-python/PythonMismatchedPackageName4 seems a bit out of place. Should I drop it, put it elsewhere, ...?

[ferringb]

Unrelated- this should be a warning IMO so we can isolate how to handle this better. That situation shouldn't exist- if it does, it points at a bad metadata.xml definition, or something (again) we should address.

[ferringb]

I think you're doing the inverse of what is needed- isn't this a malformed metadata.xml per glep68? the 'value'- name- is required. If they've specified a remote-id but left that out, it's malformed.

[falbrechtskirchinger]

I think you're doing the inverse of what is needed- isn't this a malformed metadata.xml per glep68? the 'value'- name- is required. If they've specified a remote-id but left that out, it's malformed.

While I see your point regarding GLEP 68, my goal with this PR is just to stop pkgcheck from crashing outright.

The crash prevents PackageMetadataXmlCheck from catching the empty element. While PkgMetadataXmlEmptyElement may not fail CI, that is a separate problem that another PR can address.

[falbrechtskirchinger]

Also:

pkgcheck/src/pkgcheck/checks/metadata_xml.py

Lines 572 to 576 in 9ecf675

	def _check_remote_id(self, pkg, loc, doc):
	for u in pkg.upstreams:
	# empty values are already reported as PkgMetadataXmlEmptyElement
	if not u.name:
	continue

[ferringb]

While I see your point regarding GLEP 68, my goal with this PR is just to stop pkgcheck from crashing outright.

The crash prevents PackageMetadataXmlCheck from catching the empty element. While PkgMetadataXmlEmptyElement may not fail CI, that is a separate problem that another PR can address.

No argument that this shouldn't fail out, so your PR is merged.

Offhand- this shouldn't even be necessary, but pkgcore is letting through schema violating objects- see pkgcore/pkgcore#466 . I added a commit tweaking your comment to clarify this is suppressing invalid xml, and linking to the upstream.

Thanks!