[ENG-9044] Add manage command to resync preprint dois v1

api/crossref/views.py

@@ -7,6 +7,7 @@
 
 from api.crossref.permissions import RequestComesFromMailgun
 from osf.models import Preprint, NotificationTypeEnum
+from osf.models.base import Guid
 from website import settings
 from website.preprints.tasks import mint_doi_on_crossref_fail
 
@@ -48,6 +49,23 @@ def post(self, request):
                 if record.get('status').lower() == 'success' and doi:
                     msg = record.find('msg').text
                     created = bool(msg == 'Successfully added')
+                    # Unversioned DOIs (no _vN suffix, e.g. 10.31233/osf.io/tnaqp) are routing
+                    # aliases that always resolve to the latest version via OSF's GUID routing.
+                    # Store them as 'doi_unversioned' on the v1 preprint so we can track which
+                    # preprint series have had their unversioned DOI registered.
+                    _, version = Guid.split_guid(guid) if guid else (None, None)
+                    if not version:
+                        logger.info(f'Unversioned DOI confirmed by CrossRef: {doi}')
+                        if created and guid:
+                            v1_preprint = Preprint.objects.filter(
+                                versioned_guids__guid___id=guid,
+                                versioned_guids__version=1,
+                            ).first()
+                            if v1_preprint:
+                                v1_preprint.set_identifier_value(category='doi_unversioned', value=doi)
+                        dois_processed += 1
+                        continue
+
                     legacy_doi = preprint.get_identifier(category='legacy_doi')
                     if created or legacy_doi:
                         # Sets preprint_doi_created and saves the preprint
@@ -67,9 +85,14 @@ def post(self, request):
                     if 'Relation target DOI does not exist' in record.find('msg').text:
                         logger.warning('Related publication DOI does not exist, sending metadata again without it...')
                         mint_doi_on_crossref_fail.apply_async(kwargs={'preprint_id': preprint._id})
-                    # This error occurs when a single preprint is being updated several times in a row with the same metadata [#PLAT-944]
-                    elif 'less or equal to previously submitted version' in record.find('msg').text and record_count == 2:
-                        break
+                    # This error occurs when a single preprint is being updated several times in a row
+                    # with the same metadata [#PLAT-944]. Previously this broke out of the loop when
+                    # record_count == 2 (single DOI submitted twice). Now batches legitimately contain
+                    # 2 records (versioned + unversioned DOI), so we continue instead of break to allow
+                    # the remaining record to be processed.
+                    elif 'less or equal to previously submitted version' in record.find('msg').text:
+                        dois_processed += 1
+                        continue
                     else:
                         unexpected_errors = True
             logger.info(f'Creation success email received from CrossRef for preprints: {guids}')

api/users/views.py

@@ -1477,7 +1477,8 @@ def post(self, request, *args, **kwargs):
 
         user.date_last_logged_in = timezone.now()
         user.external_identity[provider][provider_id] = 'VERIFIED'
-        user.social[provider.lower()] = provider_id
+        if provider.lower() in OSFUser.SOCIAL_FIELDS:
+            user.social[provider.lower()] = provider_id
         del user.email_verifications[token]
         user.verification_key = generate_verification_key()
         user.save()

api_tests/crossref/views/test_crossref_email_response.py

@@ -220,3 +220,91 @@ def test_confirmation_marks_legacy_doi_as_deleted(self, app, url, preprint):
             app.post(url, context_data)
 
         assert preprint.identifiers.get(category='legacy_doi').deleted
+
+    def test_unversioned_doi_confirmation_skips_identifier_update(self, app, url, preprint):
+        versioned_doi = settings.DOI_FORMAT.format(
+            prefix=preprint.provider.doi_prefix, guid=preprint._id
+        )
+        preprint.set_identifier_value(category='doi', value=versioned_doi)
+
+        base_guid = preprint.get_guid()._id  # no _vN suffix
+        unversioned_doi = settings.DOI_FORMAT.format(
+            prefix=preprint.provider.doi_prefix, guid=base_guid
+        )
+        dual_confirmation_xml = """
+            <?xml version="1.0" encoding="UTF-8"?>
+            <doi_batch_diagnostic status="completed" sp="cs3.crossref.org">
+               <submission_id>1390675999</submission_id>
+               <batch_id>{batch_id}</batch_id>
+               <record_diagnostic status="Success">
+                  <doi>{versioned_doi}</doi>
+                  <msg>Successfully updated</msg>
+               </record_diagnostic>
+               <record_diagnostic status="Success">
+                  <doi>{unversioned_doi}</doi>
+                  <msg>Successfully added</msg>
+               </record_diagnostic>
+               <batch_data>
+                  <record_count>2</record_count>
+                  <success_count>2</success_count>
+                  <warning_count>0</warning_count>
+                  <failure_count>0</failure_count>
+               </batch_data>
+            </doi_batch_diagnostic>
+        """.format(
+            batch_id=preprint._id,
+            versioned_doi=versioned_doi,
+            unversioned_doi=unversioned_doi,
+        )
+
+        context_data = self.make_mailgun_payload(crossref_response=dual_confirmation_xml)
+        with capture_notifications(expect_none=True):
+            app.post(url, context_data)
+
+        preprint.reload()
+        assert preprint.get_identifier_value('doi') == versioned_doi
+        assert preprint.get_identifier_value('doi_unversioned') == unversioned_doi
+
+    def test_unversioned_doi_confirmation_update_does_not_store_doi_unversioned(self, app, url, preprint):
+        versioned_doi = settings.DOI_FORMAT.format(
+            prefix=preprint.provider.doi_prefix, guid=preprint._id
+        )
+        preprint.set_identifier_value(category='doi', value=versioned_doi)
+
+        base_guid = preprint.get_guid()._id
+        unversioned_doi = settings.DOI_FORMAT.format(
+            prefix=preprint.provider.doi_prefix, guid=base_guid
+        )
+        update_confirmation_xml = """
+            <?xml version="1.0" encoding="UTF-8"?>
+            <doi_batch_diagnostic status="completed" sp="cs3.crossref.org">
+               <submission_id>1390676000</submission_id>
+               <batch_id>{batch_id}</batch_id>
+               <record_diagnostic status="Success">
+                  <doi>{versioned_doi}</doi>
+                  <msg>Successfully updated</msg>
+               </record_diagnostic>
+               <record_diagnostic status="Success">
+                  <doi>{unversioned_doi}</doi>
+                  <msg>Successfully updated</msg>
+               </record_diagnostic>
+               <batch_data>
+                  <record_count>2</record_count>
+                  <success_count>2</success_count>
+                  <warning_count>0</warning_count>
+                  <failure_count>0</failure_count>
+               </batch_data>
+            </doi_batch_diagnostic>
+        """.format(
+            batch_id=preprint._id,
+            versioned_doi=versioned_doi,
+            unversioned_doi=unversioned_doi,
+        )
+
+        context_data = self.make_mailgun_payload(crossref_response=update_confirmation_xml)
+        with capture_notifications(expect_none=True):
+            app.post(url, context_data)
+
+        preprint.reload()
+        assert preprint.get_identifier_value('doi') == versioned_doi
+        assert preprint.get_identifier_value('doi_unversioned') is None