In dev system, new study metadata is prepopulated with irrelevant information

[jar398]

If I go to devtree.opentreeoflife.org and try to add a new study, after I say 'continue' from the license dialog, I'm shown metadata from a previously entered study (looks like one in the ot_34...ot_37 range, according to the duplicate-DOI feature). The study id of the new study is ot_43, which indicates a new study is being created.

The URL I entered was 'http://mumble.net/' so it's not that the metadata is populated from Crossref.

I can delete the metadata and proceed, as a workaround, but blank fields would be nicer.

[jimallman]

The URL I entered was 'http://mumble.net/' so it's not that the metadata is populated from Crossref.

Actually, I thought that's exactly what's happening here. CrossRef.org does very fuzzy matching and replaces the original URL with its best guess at a DOI.

But I see that entering "http://mumble.net" into search.crossref.org gives different results, so I'll dig a bit further to see what's going on here.

[jimallman]

I see what's happening here. In phylesystem-api, we try to use the submitted DOI/URL to guess at the intended publication. This logic expects a DOI with "10." in it and munges it into a searchable form. Other URLs (any string not including "10.") results in an empty query to CrossRef, which always returns the lucky study (Humble, 2011) as the top result. It appears the mumble/humble similarity is a coincidence!

In any case, I'll check for the empty query and bail on DOI lookups in this case. This should give the desired result (empty fields throughout, but showing the original URL entered).

[jar398]

Tested, ok