This should not be needed because it should be already covered by the changes in EM translator.

it's needed because this is recognized by EM as a journal article so it doesn't move the DOI. We change that above.

We could also change the itemType in the translatorObject cf. https://github.com/zotero/translators/blob/master/Embedded%20Metadata.js#L815 and then (I assume) it is not necessary anymore. I.e.

	translator.getTranslatorObject(function(trans) {
		trans.itemType = type;
		trans.doWeb(doc, url);
	});

Maybe we could here even do

if (item.extra.search(/(^|\n)DOI:/) == -1) {

I haven't tested, but shouldn't ^DOI find the string at the beginning of any new line?

Hm... this could also be...

No, you were right; changing

Wouldn't it be better to do this in the called translator rather then the special one here?

turns out the whole passage isn't necessary -- the DOI is in the RIS and the RIS translator does handle this now.

Why did you do this change in the RDF translator? Is this the best place for this?

because the RDF translator handles all other dublin core code.

I haven't thought this through, but RDF is usually called by EM. Is there also another (common) use case for RDF translator?

Common no. General -- other RDF imports, including Zotero RDF, DC RDF, and Bibliontology RDF. Arguably those should all recognize the (valid) DC type dataset.

Shouldn't this be newItem.extra = "itemType: dataset" as documented here https://www.zotero.org/support/dev/translators/datasets ? Otherwise we should update the documentation.

again, I need to test this, but I'm pretty sure type is correct here, as this is used by citeproc and hence relies on CSL json (which uses type, not itemType).

here I was right: testing shows that "type: dataset" works, itemType: dataset doesn't. Changed in the documentation

The authors in CSL JSON look fine for me and also importing the CSL JSON works for me correctly. Can you give an example where this fix is needed?

https://www.zenodo.org/record/569304/export/csl#.WUgtNsaQy70
but it's also in one of the tests. I don't understand the problem -- have double-checked data entry for this one and we did that exactly as recommended.

Okay, I haven't seen this example. But I cannot find it in any test case...

(BTW I read the comment above also, that there is always something wrong with zenodo's author metadata, which is not the case.)

I would suggest to move this to the place below where it is actually used. Otherwise I will have forgotten what this is when reading the code below...

done

What about data set?

see above -- I'm leaving that as document on purpose (though feel free to disagree

Maybe add a comment here, what these lines are doing and why.

done

Maybe also check that we only do this once, i.e.

if (item.extra.search(/(^|\n)DOI:/) == -1) {

as below.

Since there is no way for the Extra field to contain anything other than note (given the CSL import from Zotero) I just empty the field above and can do without all this code.