Some funderIdentifiers are not indexed in ElasticSearch

[KellyStathis]

Describe the bug

Expected Behaviour

If a funderIdentifier is in the XML, it should be indexed in ElasticSearch and therefore show in the REST API response.

Current Behaviour

Some funderIdentifiers are not indexed.

This was identified because there are more records with funderIdentifierType than funderIdentifier in the index, even though funderIdentifierType cannot exist without funderIdentifier.

Steps to Reproduce

1. Go to https://api.datacite.org/dois/10.5285/8e59f849-5b93-438e-a5e0-3c65636f9053 - see fundingReference is missing funderIdentifier:

{
"awardUri": "http://gotw.nerc.ac.uk/list_full.asp?pcode=NE%2FL002434%2F1",
"awardTitle": "NERC GW4+ Doctoral Training Partnership studentship",
"funderName": "Natural Environment Research Council, UK Research & Innovation",
"awardNumber": "NE/L002434/1",
"funderIdentifierType": "ISNI"
}

2. Decode the XML from base64 - it shows this:

<fundingReference>
      <funderName>Natural Environment Research Council, UK Research &amp; Innovation</funderName>
      <funderIdentifier funderIdentifierType="ISNI">0000 0001 2181 0377</funderIdentifier>
      <awardNumber awardURI="http://gotw.nerc.ac.uk/list_full.asp?pcode=NE%2FL002434%2F1">NE/L002434/1</awardNumber>
      <awardTitle>NERC GW4+ Doctoral Training Partnership studentship</awardTitle>
    </fundingReference>

Context (Environment)

Screenshots

n/a

Further details

n/a

Proposal

Hypothesis

As @richardhallett identified, it looks like funderIdentifier is validating for a URL in bolognese:
https://github.com/datacite/bolognese/blob/master/lib/bolognese/readers/datacite_reader.rb#L161. Because this funderIdentifier is not a URL, it is excluded.

Possible Implementation

Ideally ISNIs would be entered as URLs, e.g.: https://isni.org/isni/0000000121810377

However, this is not enforced by the XSD. When funderIdentifiers are not entered as URLs, our parsing could be more forgiving so as not to exclude them from indexing entirely.

[KellyStathis]

Related: #137 where there was discussion of normalization logic for ISNI.

[codycooperross]

This bug is fixed for new metadata imports. Existing indexed metadata without funderIdentifiers will need to be reimported for the fix to be reflected in the API.