fix etree parse on literal string

[tlambert03]

This should fix the issue that @JacksonMaxfield had in #50 (comment) ... wherein ElementTree.parse() is looking for a path name or file-like object, but receiving a literal xml string (when parsing structured_annotations).

cc @jmuhlich

[tlambert03]

actually... @jmuhlich, I'm getting an error in roundtrips on python 3.7 now (not related to this PR ... it's even on master) with the C14NWriterTarget backport:

testing/test_model.py:117:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
testing/test_model.py:113: in canonicalize
    xml_out = util.canonicalize(xml_out, strip_text=True)
testing/util.py:36: in canonicalize
    parser.feed(xml_data)
testing/util.py:205: in start
    self._start(tag, attrs, new_namespaces)
testing/util.py:234: in _start
    n: parse_qname(n) for n in sorted(qnames, key=lambda n: n.split("}", 1))
testing/util.py:234: in <dictcomp>
    n: parse_qname(n) for n in sorted(qnames, key=lambda n: n.split("}", 1))
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

self = <util.C14NWriterTarget object at 0x1110322d0>, qname = '{http://www.openmicroscopy.org/Schemas/OME/2016-06}OME'
uri = 'http://www.openmicroscopy.org/Schemas/OME/2016-06'

    def _qname(self, qname, uri=None):
        if uri is None:
            uri, tag = qname[1:].rsplit("}", 1) if qname[:1] == "{" else ("", qname)
        else:
            tag = qname

        prefixes_seen = set()
        for u, prefix in self._iter_namespaces(self._declared_ns_stack):
            if u == uri and prefix not in prefixes_seen:
                return f"{prefix}:{tag}" if prefix else tag, tag, uri
            prefixes_seen.add(prefix)

        # Not declared yet => add new declaration.
        if self._rewrite_prefixes:
            if uri in self._prefix_map:
                prefix = self._prefix_map[uri]
            else:
                prefix = self._prefix_map[uri] = f"n{len(self._prefix_map)}"
            self._declared_ns_stack[-1].append((uri, prefix))
            return f"{prefix}:{tag}", tag, uri

        if not uri and "" not in prefixes_seen:
            # No default namespace declared => no prefix needed.
            return tag, tag, uri

        for u, prefix in self._iter_namespaces(self._ns_stack):
            if u == uri:
                self._declared_ns_stack[-1].append((uri, prefix))
                return f"{prefix}:{tag}" if prefix else tag, tag, uri

>       raise ValueError(f'Namespace "{uri}" is not declared in scope')
E       ValueError: Namespace "http://www.openmicroscopy.org/Schemas/OME/2016-06" is not declared in scope

any thoughts?

[evamaxfield]

Just wanted to comment here and say I just tried out this branch and it works for my file. I figured I would give a shot at understanding the broken tests. Interestingly, the tests work just fine with pytest testing/ but fail on tox. Will keep investigating.

[tlambert03]

Oh darn, sorry I didn't push this through faster. Thanks for trying it out, I'll make sure to finish it up this week one way or another

[evamaxfield]

No worries on timing. I was busy as well and just started working again. Thanks for all existing work on this haha 🙂

[tlambert03]

fixed the tests error. had to do with the "imported" xml.etree.ElementTree._namespace_map diverging from the actual one. merging now

[evamaxfield]

Thanks for the patch!

[tlambert03]

@JacksonMaxfield, if a new bug-fix release would make your life easier when testing this out, let me know and I'll push one

[evamaxfield]

@JacksonMaxfield, if a new bug-fix release would make your life easier when testing this out, let me know and I'll push one

I will keep testing with the latest master branch now. I may have a PR of my own to this lib, still debugging though. If I do have a PR would be best to patch out together probably.

[tlambert03]

lol ... oops, just pushed 0.2.1, figuring we wouldn't have anything soon. No worries though! If you have something, that's great! no problem releasing again soon

[evamaxfield]

Hey @tlambert03 I am also having issues where my pytest stuff passes but tox doesn't. Specifcally, it breaks on ElementTree.tostring, since you had the same issue I looked through commits and see you single forced name spaced ElementTree. on all functions and such. I tried the same but can't get it to work - any ideas?

[tlambert03]

can I see the full trace? Does it fail in tox in all python versions? or just in one of them? We vendored some of ElementTree for testing purposes on py3.7, and there may still be some hiccups in the approach.

[evamaxfield]

Wow. That was it. I kept interrupting the tests after they failed on py37 and didn't let them run through py38 or py39. From xml docs I now see that the xml_declaration parameter is only available on 3.8+. Remove that kwarg fixed it. My bad!

Thanks for the help.