Fix word cloud modules not exporting correctly.

[peter-fogg]

@chrisndodge @cahrens Fixes STUD-146. I also looked through other modules to see if this might be an issue -- it looks like word cloud and video are the only ones which inherit from RawDescriptor and could end up with no data.

[auraz]

Please explain why there is not assertions on values of metadata fields in exported word_cloud xml (in test)

[cahrens]

If the tag doesn't matter, I suggest that we put a default definition into definition_to_xml and remove the wordcloud and videoalpha-specific workarounds.

@chrisndodge, what do you think of this?

[auraz]

@cahrens good point.

[peter-fogg]

@cahrens @chrisndodge The latest commit has a more general approach. We still need to raise an exception if the XML isn't well-formed, so we only provide a default when self.data is empty (None or the empty string).

[chrisndodge]

Can I rope @cpennington in on this? I say that because this is how I originally implemented the video-module fix and Cale thought it would be better not to do this in the base class. Maybe now that we're seeing multiple instances of the same problem this changes how we approach it.

I guess another approach could be to make a RawDescriptorWithOnlyMetadata class which derives from RawDescriptor and overrides the definition_to_xml() method. Then we have video_module.py and word_cloud derive from that intermediate class?

[peter-fogg]

@chrisndodge That would be doable, but since RawDescriptor only has two methods it seems odd to subclass it just to stub out one of them.

[cahrens]

Personally, I like this approach. If we go with it, write a unit test for this new functionality (independent of word cloud or video alpha).

[cpennington]

If self.data isn't empty, then this will delete it, which isn't what we want, I don't think, since during import we import content to self.data if there is content to import.

[peter-fogg]

@cpennington So we should check if data is empty, return the empty tag if so, and read in the XML otherwise?

[cpennington]

Yeah.

On Tue, Jul 30, 2013 at 1:16 PM, Peter Fogg notifications@github.comwrote:

In common/lib/xmodule/xmodule/raw_module.py:

+class EmptyDataRawDescriptor(XmlDescriptor, XMLEditingDescriptor):

• """
• Version of RawDescriptor for modules which may have no XML data,
• but use XMLEditingDescriptor for import/export handling.
• """
• data = String(default='', scope=Scope.content)

• @classmethod
• def definition_from_xml(cls, xml_object, system):

•    if len(xml_object) == 0 and len(xml_object.items()) == 0:
  

•        return {'data': ''}, []
  

•    return {'data': etree.tostring(xml_object, pretty_print=True, encoding='unicode')}, []
  
• def definition_to_xml(self, resource_fs):

•    return etree.Element(self.category)
  

So we should check if data is empty, return the empty tag if so, and read
in the XML otherwise?

—
Reply to this email directly or view it on GitHubhttps://github.com/edx/edx-platform/pull/492/files#r5482617
.

[peter-fogg]

Done.

[peter-fogg]

@cpennington @cahrens @chrisndodge Is this good to merge?

[chrisndodge]

I see you ended up making a new subclass of RawDescriptor.

LGTM

[cahrens]

I'll defer to @cpennington and @cdodge.

[cpennington]

👍