[Zenodo] Prevent mis-importing software as report

[katrinleinweber]

As discussed in the forum this ensures that Software items on Zenodo are imported as computerProgram, even though along the way, Zenodo is offering a non-CSL item type

Potential/Remaining isues to discuss

• Scaffold imports the full author names as lastName. This could occur in https://github.com/zotero/translators/blame/master/Zenodo.js#L172-L180, but I'm not sure whether or how to adjust that. Or, should I adjust the new test case to expect lastName and firstName normally?
• Would it be helpful, if Zenodo stopped putting a clearly mis-matching CSL type in the JSON that Zotero parses, and instead send none? If yes, could we then omit both the "fix" and this "undo"?
• Scaffold added - and + in the doWeb test output (near the bottom in the diff). Not sure what those mean, but I guess both symbols need to be removed from the JSON block.

[adam3smith]

How are you generating the tests in Scaffold? This is fine now, but to save you the work going forward, if you actually generate the test using the Scaffold testing panel, you won't get the + and - (which are from a diff display Scaffold shows).

[adam3smith]

what would happen if we replace this whole thing, including l. 197 by item.itemType = detectWeb(doc, url)
That seems like a cleaner solution, provided our detect function is good.

[katrinleinweber]

Could the var zoteroType block be removed as well, then?

[adam3smith]

exactly, yes

[adam3smith]

The Zenodo item types in the CSL are fine -- they are as good as it gets for e.g. automated citation generation directly from the CSL and Zotero can customize at will in the import.

For the names, we can't really parse these correctly without breaking institutional names since they're entered incorrectly in Zenodo, but we can switch fieldMode to true along the lines for

for (let i =0; i< item.creators.length; i++) {
  if (!item.creators[i].firstName) {
    item.creators[i].fieldMode = true;
  }
}

This makes them easier to fix manually.

[zuphilip]

Does this work? No changes in the test cases??

I think what we want is rather want is:

if (!item.creators[i].firstName) {
	if (item.creators[i].lastName.indexOf(",")!=-1) {
		item.creators[i].firstName = item.creators[i].lastName.replace(/.+?,\s*/, "");
		item.creators[i].lastName = item.creators[i].lastName.replace(/,.+/, "");
	} else {
		item.creators[i].fieldMode = true;
	}
}

[zuphilip]

Can we delete the first introduction of the variable type as well?

[katrinleinweber]

Do you mean L42 or L133? Removing the latter indeed doesn't break the tests :-)

[zuphilip]

Yes, the latter one seems not necessarily anymore. We certainly still need the first one.

[zuphilip]

Add ; at the end.

[zuphilip]

Rerun the tests should yield some fieldMode fields.

[zuphilip]

🎉 Thank you @katrinleinweber !

[katrinleinweber]

You're welcome, and thanks for merging :-)