From e6773cb68162e62d90eae5331103afdf72421cf2 Mon Sep 17 00:00:00 2001
From: Abram Booth <boothaa@gmail.com>
Date: Wed, 30 Aug 2023 16:27:04 -0400
Subject: [PATCH] fix: infer funderIdentifierType from identifier

---
 osf/metadata/osf_gathering.py                     |  1 -
 osf/metadata/rdfutils.py                          |  1 +
 .../serializers/datacite/datacite_tree_walker.py  | 15 +++++++++++++--
 3 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/osf/metadata/osf_gathering.py b/osf/metadata/osf_gathering.py
index 60d8a6210602..11e60b5c0156 100644
--- a/osf/metadata/osf_gathering.py
+++ b/osf/metadata/osf_gathering.py
@@ -839,7 +839,6 @@ def gather_funding(focus):
                 yield (_funder_ref, RDF.type, DCTERMS.Agent)
                 yield (_funder_ref, DCTERMS.identifier, _funder_uri)
                 yield (_funder_ref, FOAF.name, _funder_name)
-                yield (_funder_ref, OSF.funderIdentifierType, _funding.get('funder_identifier_type'))
             _award_uri = _funding.get('award_uri')
             _award_title = _funding.get('award_title')
             _award_number = _funding.get('award_number')
diff --git a/osf/metadata/rdfutils.py b/osf/metadata/rdfutils.py
index 259efdf08249..6dbe7039c5d4 100644
--- a/osf/metadata/rdfutils.py
+++ b/osf/metadata/rdfutils.py
@@ -12,6 +12,7 @@
 OSFIO = rdflib.Namespace(website_settings.DOMAIN)
 # external pid namespaces:
 DOI = rdflib.Namespace('https://doi.org/')
+DxDOI = rdflib.Namespace('http://dx.doi.org/')  # "earlier but no longer preferred" DOI namespace
 ORCID = rdflib.Namespace('https://orcid.org/')
 ROR = rdflib.Namespace('https://ror.org/')
 # external terminology namespaces:
diff --git a/osf/metadata/serializers/datacite/datacite_tree_walker.py b/osf/metadata/serializers/datacite/datacite_tree_walker.py
index a00911fc1d0d..08990d466a9e 100644
--- a/osf/metadata/serializers/datacite/datacite_tree_walker.py
+++ b/osf/metadata/serializers/datacite/datacite_tree_walker.py
@@ -12,6 +12,7 @@
     RDF,
     DCTERMS,
     DOI,
+    DxDOI,
     FOAF,
     ORCID,
     OSF,
@@ -154,6 +155,8 @@ def _visit_creators(self, parent_el, focus_iri):
     def _identifier_type_and_value(self, identifier: str):
         if identifier.startswith(DOI):
             return ('DOI', without_namespace(identifier, DOI))
+        elif identifier.startswith(DxDOI):
+            return ('DOI', without_namespace(identifier, DxDOI))
         elif identifier.startswith(ROR):
             return ('ROR', identifier)  # ROR keeps the full IRI
         elif identifier.startswith(ORCID):
@@ -162,6 +165,13 @@ def _identifier_type_and_value(self, identifier: str):
             return ('URL', identifier)
         logger.warning('skipping non-IRI-shaped identifier "%s"', identifier)
 
+    def _funder_identifier_type(self, identifier: str):
+        if identifier.startswith(DxDOI) or identifier.startswith(DOI):
+            return 'Crossref Funder ID'
+        if identifier.startswith(ROR):
+            return 'ROR'
+        return 'Other'
+
     def _get_name_type(self, agent_iri):
         if (agent_iri, RDF.type, FOAF.Person) in self.basket:
             return 'Personal'
@@ -261,12 +271,13 @@ def _visit_funding_references(self, parent_el):
         for _funder in self.basket[OSF.funder]:
             fundref_el = self.visit(fundrefs_el, 'fundingReference')
             self.visit(fundref_el, 'funderName', text=next(self.basket[_funder:FOAF.name], ''))
+            funder_identifier = next(self.basket[_funder:DCTERMS.identifier], '')
             self.visit(
                 fundref_el,
                 'funderIdentifier',
-                text=next(self.basket[_funder:DCTERMS.identifier], ''),
+                text=funder_identifier,
                 attrib={
-                    'funderIdentifierType': next(self.basket[_funder:OSF.funderIdentifierType], ''),
+                    'funderIdentifierType': self._funder_identifier_type(funder_identifier),
                 },
             )
             for _funding_award in self.basket[OSF.hasFunding]: